Exhibit 4 To The Comments Of Relpromax Antitrust Inc.

TO THE COMMENTS

OF RELPROMAX ANTITRUST INC.

GARY L. REBACK (Bar No. 218594)
SUSAN A. CREIGHTON
DAVID A. KILLAM
NEIL M. NATHANSON
IRWIN R. GROSS
WILSON, SONSINI, GOODRICH & ROSATI
Professional Corporation
650 Page Mill Road
Palo Alto, California 94304-1050
Telephone: (415) 493-9300

FOR THE DISTRICT OF COLUMBIA

TABLE OF CONTENTS

1. INTRODUCTION AND SUMMARY
2. THE PERMISSIBLE SCOPE OF THIS COURT'S REVIEW
3. THE ECONOMIC CHARACTERISTICS OF THE SOFTWARE INDUSTRY

    

    

   1. Market And Technology Background

       

       

       

       

       

       

       

       

       

       

      1. The Business Desktop
      2. The Intrabusiness Server
      3. The Home-to-Business Server
      4. Home Computer Market
      5. OLE
      6. Windows

          

          

         1. Desktop
         2. Server
   2. Free Market Forces in Increasing Return Industries
4. MICROSOFT'S TACTICS AND PROSPECTS FOR SUCCESS
   1. Microsoft's Strategy

       

       

       

       

      1. The Business Desktop

          

          

          

          

          

          

          

          

          

          

         1. Effect of the Monopoly on Operating Systems
         2. Effect of the Monopoly on Applications
         3. Unfair Early Access
         4. Predatory Bundling
         5. Predatory Unbundling
         6. Other Uses of Leverage
      2. The Intrabusiness Server
      3. The Home-to-Business Market (Server and Client)

          

          

          

          

         1. Predatory Bundling
         2. Unfair Use of Information
         3. Unfair Head Start
5. MICROSOFT'S NETWORK-WIDE MONOPOLY

    

    

    

    

    

    

   1. Leverage of the Installed Base by Competitors
   2. Alliances
   3. "Tiered" Monopoly
   4. Efficiencies of Integration
6. ANTITRUST ENFORCEMENT

    

    

   1. Deficiencies of the Proposed Judgment

       

   2. Comparable Consent Decrees

       

       

       

       

      1. Parke, Davis Decree (Pharmaceuticals)
      2. International Business Machines Corp. (Computers)
      3. American Telephone and Telegraph (Telecommunications)

       

   3. Case Law
7. PROPOSED PROCEDURES UNDER SECTION 16(f)

TABLE OF AUTHORITIES

CASES

Alaska Airlines, Inc. v. United Airlines, Inc., 948 F.2d 536 (9th Cir. 1991), cert. denied, 112 S. Ct. 1603 (1992)

Digidyne Corp. v. Data General Corp., 734 F.2cl 1336 (9th Cir. 1984), cert. denied, 473 U.S. 908 (1985)

Eastman Kodak Co. v. Image Technical Services, Inc., 112 S. Ct. 2072 (1992)

Grappone, Inc. v. Subaru of New England, Inc., 858 F.2d 792 (1st Cir. 1988)

Lee v. Life Ins. Co., 829 F. Supp. 529 (D.R.I. 1993), aff'd, 23 F.3d 14 (1st Cir.), cert. denied, 1994 U.S. LEXIS 7596 (1994)

Ortho Diagnostic Systems, Inc, v.. Abbott Laboratories, Inc., 822 F. Supp. 145 (S.D.N.Y. 1993)

United States v. American Tel. & Tel., 552 F.Supp. 131 (D.D.C. 1982), aff'd sub nom. Maryland v. United States, 460 U.S. 1001 (1983)

United States v. BNS Inc., 858 F.2d 456 (9th Cir. 1988)

United States v. Gillette Co., 406 F.Supp. 713 (D. Mass. 1975)

United States v. International Business Machines Corp., 1956 Trade Cas. (CCH) . 68,245 (S.D.N.Y. 1956)

United States v. Parke, Davis and Co. and Eli Lilly and Co., 1951 Trade Cas. (CCH) . 62,914 (E.D. Mich. 1951)

United States v. Parke, Davis and Co. and Eli Lilly and Co., 1987-2 Trade Cas. (CCH) . 67,834 (E.D. Mich. 1987)

United States v. Western Electric Co., 900 F.2d 283 (D.C. Cir. 1990), cert denied, 498 U.S. 911 (1990)

United States v. Western Electric Co., Inc., 993 F.2d 1572 (D.C. Cir. 1993)

United States v. Yoder, 1989-2 Trade Cas. (CCH) . 68,723 (N.D. Ohio 1986)

Viacom International, Inc. v. Time Inc., 785 F. Supp. 371 (S.D.N.Y 1992)

STATUTES

47 Fed. Reg. 23,320, 23,335 (May 27, 1982)

59 Fed. Reg. 59,426 (Nov. 17, 1994)

59 Fed. Reg. 59,427 (Nov. 17, 1994)

59 Fed. Reg. 59,429 (Nov. 17, 1994)

59 Fed. Reg. 42,847 (Aug. 19, 1994)

59 Fed. Reg. 42,850 (Aug. 19, 1994)

15 U.S.C. .. 16(b)-(h) (1994)

15 U.S.C. . 16(e)

15 U.S.C. . 16(f)

Additional Materials

All Things Considered (NPR broadcast, Nov. 17, 1994)

The Antitrust Procedures and Penalties Act: Hearings on S. 782 and S. 1088 before the Subcommittee on Antitrust and Monopoly of the Committee on the Judiciary, 93rd Cong., 1st Sess. (1973)

Byte, May 1992, at 159

The Computer Industry Survey: Reboot System and Start Again, The Economist, Feb. 27-Mar. 5, 1993

Computerworld, Dec. 6, 1993 (International Data Corp. Table)

A Fierce Battle Brews Over the Simplest Software Yet, Business Week, November 21, 1983, at 114

Microsoft Corp. Employee Performance Review, dated Nov. 2, 1987

Microsoft Corp. Employee Performance Review, dated May 4, 1987

Microsoft's Domination, San Jose Mercury News, December 21, 1994

Personal Computing Software Worldwide, Dataquest, June 27, 1994

OS Overview, Computer Reseller News, Aug. 22, 1994

PC Week, Feb. 21, 1994 (Paine Webber, Inc. Table)

Quote of the Week, InformationWeek, Aug. 1, 1994

Tr. of Status Call, Sept. 29, 1994

Will Sun Melt the Software Barrier, Business Week, April 18, 1988, at 72

Paul Andrews, Windows Is No JFK, But Its Visual Appeal Is Outstanding, Seattle Times, May 22, 1990

Paul Andrews, Can Microsoft Just Do It?, Seattle Times, March 18, 1991

W. Brian Arthur, Positive Feedback in the Economy, Scientific American, Feb. 1990

W. Brian Arthur, Increasing Returns and Path Dependence in the Economy (1994)

Doug Barney, Microsoft, Stac Resolve Dispute; Microsoft Finally Pays Up, InfoWorld, June 27, 1994, at 14

Doug Barney, Microsoft to Announce New On-Line Service at Comdex, InfoWorld, Oct. 24, 1994, at 1, 140

Doug Barney and Ilan Greenberg, ISVs Dampen Microsoft Furor for OLE, InfoWorld, July 18, 1994, at 1

Amy Bernstein, Microsoft Goes Online, U.S. News & World Report, Nov. 21, 1994, at 84

Jesse Berst, Behind The Smoke: Microsoft Wins Again, PC Week, July 25, 1994, at 106

Jesse Berst, Microsoft's On-Line Rivals Could End Up In 'Cyberia', PC Week, Dec. 12, 1994, at 120

Jeff Bertolucci, Microsoft Settles: Business As Usual, PC World, October 1994, at 72

Stanley M. Besen and Joseph Farrell, Choosing How to Compete, J. of Econ.t Perspectives, Spring 1994

Stanley Besen and Leland Johnson, Compatibility Standards, Competition, and Innovation in the Broadcasting Industry (1986)

Stanley Besen and Garth Saloner, The Economics of Telecommunications Standards, in Changing the Rules: Technological Change, International Competition, and Regulation in Communications 177 (1989)

William Brandel, Developing for Next Generation of Windows May Mean Running on NT, Computerworld, November 18, 1994, at 4

Richard Brandt, Microsoft Is Like an Elephant Rolling Around, Squashing Ants, Business Week, Oct. 30, 1989, at 148

Larry Campbell, Novell to Introduce SuperNOS Strategy, South China Morning Post, Sept. 20, 1994, at 1

Jim Carlton & G. Pascal Zachary, Microsoft Sells A 20% Interest In Planned Unit, Wall St. J., Dec. 22, 1994

William Casey, Let's Stop Beating On Microsoft, Washington Post, July 25, 1994, at F15

Don Clark, Microsoft to Buy Intuit In Stock Pact, Wall St. J., Oct. 14, 1994, at A3

Don Clark and Laurie Hays, Microsoft's New Marketing Tactics Draw Complaints, Wall St. J., Dec. 12, 1994, at B6

Elizabeth Corcoran, Microsoft Deal Came Down to a Phone Call, Washington Post, July 18, 1994, at A1

Elizabeth Corcoran, Microsoft Heads Home: Software Giant Targets Huge Consumer Market With a Host of High-Tech Innovations, Washington Post, Nov. 13, 1994, at H1, H6

O. Casey Corr, IBM vs. Microsoft -- Software Superbowl -- IBM to Kick Off New Version of OS/2, but will Microsoft Make Winning Goat, Seattle Times, March 29, 1992, at C1

O. Casey Corr, A Look Behind Stac Deal, Seattle Times, June 26, 1994, at F1

Amy Cortese, Next Stop, Chicago, Business Week, Aug. 1, 1994, at 24

Amy Cortese, No Slack for Microsoft Rivals, Business Week, Dec. 19, 1994, at 35

Michael Csenger & Adam Griffin, Microsoft Free At Last?, Ruling Still Lets Firm Incorporate Apps Into Its OS'es, Network World, July 25, 1994, at 4

Cara A. Cunningham, IBM and Microsoft Wage Open Doc vs. OLE Find, InfoWorld, Aug. 15, 1994, at 25

Barbara Darrow, Developers Brace for Shakeout, Computer Reseller News, Feb. 1, 1993, at 28

Paul A. David, Clio and the Economics of QWERTY, Amer. Econ. Rev., May 1985, at 332

Department of Justice Press Conference (July 16, 1994)

Laura DiDio, NetWare, NT Server to Divide Lion's Share, Dec. 26, 1994, at 77

John Eckhouse, Giant Microsoft Buys Intuit for $1.5 Billion, San Francisco Chronicle, Oct. 14, 1994, at A1

David Einstein, Microsoft Unscathed by Settlement, S.F. Chronicle, July 18, 1994, at A1

Karen Epper, Software Deal Shakes Up Home Banking, Amer. Banker, Oct. 17, 1994, at 1,25

Joseph Farrell, Hunter K. Monroe and Garth Saloner, The Vertical Organization of Industry and Systems Competition Versus Component Competition, October 1994 (working paper)

Joseph Farrell and Garth Saloner, Standardization, Compatibility, and Innovation, Rand J. of Econ., Spring 1985

Joseph Farrell & Garth Saloner, Installed Base and Compatibility: Innovation, Product Pre-Announcements, and Predation, Amer. Econ. Rev., Dec. 1986

Joseph Farrell and Carl Shapiro, Dynamic Competition with Switching Costs, Rand J. of Econ., Spring 1988, at 123-137

Michelle Flores, Probe of Microsoft is Extended -- Justice Dept. Asks For More Information, Seattle Times, Nov. 22, 1994, at B11

Laurie Flynn & Rachel Parker, Extending its Reach, InfoWorld, August 7, 1989, at 43

Mary Jo Foley, Microsoft Lays Foundation For On-Line Network, PC Week, Nov. 14, 1994, at 1

Adam Gaffin & Peggy Watt, Microsoft, Lotus Battle Shifting to On-Line Services, Network World, Nov. 21, 1994, at 1

Lee Gomes, Microsoft to Acquire Intuit, San Jose Mercury News, Oct. 14, 1994, at 1D

John M. Goodman, The DOS Heavyweights Go Another Round, InfoWorld, Aug. 29, 1994, at 87

David A. Harvey, Ergonomic Issues Have Taken a Backseat to Performance, Resulting in a Growing Tide of Computer-Related Injuries. Change is Needed - Now!, Byte, Oct. 1, 1991, at 119

Stuart J. Johnston, Dangerous Liasons, InfoWorld, April 8, 1991, at 44

Stuart J. Johnston,

'No Chinese Wall' at Microsoft, Infoworld, Dec. 30, 1991, at 107

Stuart J. Johnston, Microsoft Settles for Piece of Stac, Computerworld, June 27, 1994, at 30

Stuart J. Johnston, Decree: Deal or Dodge?, Computerworld, July 25, 1994

Stuart J. Johnston and Ed Scannell, Server Suite Could Squeeze Market, Computerworld, Oct. 10, 1994, at 4
Joseph Katten, Market Power in the Presence of an Installed Base, 62 Antitrust L.J. 1 (1993)
Michael Katz and Carl Shapiro, Systems Competition and Network Effects, J. of Econ. Perspectives, Spring 1994

Michael Katz and Carl Shapiro, Network Externalities, Competition, and Compatibility, Amer. Econ. Rev., June 1985

Miles B. Keyhoe, The Winds of Change, HP Professional, Aug. 1994, at 40

Bob Lewis, Ten Troublesome Trends in Computing That Are Sure to Spook You, InfoWorld, Oct. 31, 1994, at 82

Yael Li-Ron, PC DOS 6.3: DOS and DOS: Separated At Birth, PC-Computing, July 1994, at 94

Brian Livingston, Undocumented Windows Calls, InfoWorld, Nov. 16, 1992, at 98

Brian Livingston, Will 'Windows' Compatible Really Mean What It Says?, InfoWorld, November 14, 1994, at 40

Claudia Maclachlan, Software Makers Mull Over Microsoft Legal Challenge, National Law Journal, Aug. 1, 1994, at B1

Lawrence J. Magid, Microsoft: Not So Marvelous, Bay Area Computer Currents, Dec. 1, 1994, at 98, 101

John Markoff, Microsoft's Future Barely Limited, N.Y. Times, July 18, 1994, at D1

Charles McCoy, Microsoft to Pay Stac Judgment of $120 Million, Wall St. J., Feb. 24, 1994, at A4

Start Miastkowski, Digital Research Creates a Better DOS, Byte, Nov. 1991, at 68

Michael J. Miller, The World According to Microsoft, PC Magazine, Jan. 24, 1995, at 80

Charles R. Morris and Charles H. Ferguson, How Architecture Wins Technology Wars, Harv. Bus. Rev., Mar. 1993

Michael Morris, Microsoft Deal: Too Little, Too Late, S.F. Examiner, July 24, 1994, at C-5

Jane Morrissey, DOJ Accord Fosters 'Too Little, Too Late' Perception, PC Week, July 25, 1994, at 1

Jane Morrissey, Microsoft's Application Unit Seeks Market Dominance, PC Week, Nov. 18, 1991, at 1

Viveca Novak, Antitrust's Bingaman Talks Tough in Microsoft Case, Wall St. J., July 19, 1994, at B5

Rory J. O'Connor, Microsoft, Intel Set to Define Technology, San Jose Mercury News, Nov. 13, 1994, at 1-A

Christopher O'Malley, The New Operating Systems, Personal Computing, October 1986, at 181

Janusz A. Ordover and Garth Saloner, Predation, Monopolization, and Antitrust, in Handbook of Industrial Organization 537 (R. C. Schmalensee and R. Willis eds., 1989)

Carroll Patton, Bundles Are Bad News, Computerworld, Nov. 14, 1994, at 57

Alan Padding, IBM PC Orphans Hang On To A Good Thing, Computerworld, March 7, 1988, at 81

Kathy Rebello, et al., Is Microsoft Too Powerful, Business Week, March 1, 1993, at 88

Earle Robinson, DOS-version Madness? Integration Coping with DOS, Windows Sources, Oct. 1994, at 163

Jeffrey Rohlfs, A Theory of Interdependent Demand for a Communications Service, Bell J. of Econ., Spring 1974

Garth Saloner, Economic Issues in Computer Interface Standardization, Econ. Innov. New Techn., 1990

Brent Schlender, Bill Gates: What Doesn't He Want, Fortune, Jan. 16, 1995

Jerry Schneider, Dump DOS? No Way, Not Yet, Computer Decisions, March 1988, at 50

Jack Schofield, Computing 94: Processor Wars and Rumors of Delays, Guardian, Dec. 29, 1994, at T14

Michael Schrage, Microsoft Can Make Lots of Money; Can It Shape the Management 0f It?, Washington Post, Oct. 21, 1994, at B3

Andrew Schulman, Microsoft's Grip On Software Tightened By Antitrust Deal, Dr. Dobb's Journal of Software Tools, Oct. 1994, at 143

J. William Semich, The Long View From Microsoft: Component DBMSs, Datamation, Aug. 1, 1994, at 40

Sean Silverthorne, AMD Files $2 Billion Antitrust Suit Against Intel, Investor's Daily, August 30, 1991, at 1

Gina Smith, Merger Misgivings: Will Intuit Go 'Soft?, S.F. Chronicle, Dec. 4, 1994, at B5

Michael Stroud, FTC Widens Probe of Microsoft Dominence, Investor's Daily, Apr. 15, 1991, at I

Eamonn Sullivan & Matt Kramer, Microsoft Marvel Beta Leverages WIN 95 Desktop, PC Week, Nov. 7, 1994, at 169

Jean Tirole, The Theory of Industrial Organization (1988)

L. D. Tyson, Who's Bashing Whom? Trade Conflict in High-Technology Industries (1992)

Doug VanKirk, Integrated Office Suites, InfoWorld, Feb. 7, 1994, at 51

James Wallace & Jim Erickson, Hard Drive: Bill Gates and the Making of the Microsoft Empire (1992)

Ray Weiss, Windows Stars at SD 91, Electronic Engineering Times, Feb. 18, 1991

Ron White, Microsoft Gives the New Word, PC Week, Oct. 20, 1987, at 95

Ethan Winer, BASIC, Yes; Feeble, No., PC Magazine, Oct. 30, 1989, at 187

G. Pasquel Zachary, Showstopper: Breakneck Race To Create Windows NT and the Next Generation at Microsoft, 27 (1994)

Victor F. Zonana, $14-Million Deal Microsoft Buys Software Competitor, L.A. Times, July 31, 1987, at 4

The Department of Justice has determined that from 1988 through July 1994, a period during which the number of personal computers in the United States virtually exploded. Microsoft Corporation successfully used a variety of unlawful and "anticompetitive" practices to maintain its monopoly position in the market for "operating systems" for use with personal computers. As a result of these unlawful practices, Microsoft has been able to preclude any meaningful competition in the market while increasing the installed base of Microsoft operating systems from well under 20 million in 1988 to approximately 120 million in 1994.

This memorandum¹ will show that under established economic theory, this now- massive installed base will enable Microsoft, if unchecked, both to maintain its monopoly in the operating systems market, and to leverage its installed base to dominate and monopolize the markets for applications and other software products. This brief also will show that the Department's proposed decree completely fails to address the consequences of the huge increase in installed base that Microsoft has procured through illegal practices. Instead, the Department simply proposes to shut the barn door now that the horse has already gone.

Under established economic theory, it is clear that the proposed decree will neither result in an increase in competition in the operating systems market, nor prevent Microsoft from monopolizing the remainder of the software industry. These amici accordingly urge the Court to require further submissions from the Department, both by way of expert affidavits and the production of documents, to explain how permitting Microsoft to profit from its illegal conduct not just by continuing, but by expanding, its monopolization of the software industry can be argued to be in the "public interest."

INTRODUCTION AND SUMMARY

This Court has been asked to endorse the proposed Consent Decree between the Department of Justice and Microsoft without being provided with any of the information upon which a meaningful determination under the Tunney Act could be based. Thus, for example, the Department's investigation ostensibly inquired regarding "alleged false product preannouncements" by Microsoft. 59 Fed. Reg. 59,426, 59,427 (Nov. 17, 1994). At the September 29, 1994 hearing on this matter, the Court referred to this issue, noting that in the book Hard Drive,² Microsoft was said "time after time" to predatorially preannounce products "with the intent [to] freeze other people from coming out with their product." Tr. of Status Call, Sept. 29, 1994, at 16:21-22. The following colloquy then took place between Microsoft's counsel and the Court:

Id. at 15:7-12, 16:18-17:1.

Microsoft's representations, however, are belied by Microsoft's own documents, produced to the Government during the course of its investigation. (Examples of such documents are attached hereto at Appendix Exs. 21 and 22.)³ Thus, for example, a Microsoft manager was involved in spearheading two product preannouncements during one six-month period. In one instance, the manager wrote that in response to "Borland's announce[ment of] TurboBASIC at the November Comdex," he simultaneously worked "to develop a [Microsoft] spec[ification] that could beat TurboB," while also formulating a promotional campaign "that could hold our position until [QB3, the Microsoft product] hit the market."⁴ He stated that he "reviewed [this] promotion plan with Bill G. before implementation." Id. The Microsoft documents state that Steve Ballmer, one of Microsoft's top executives, favorably commented on this strategy, saying that the "best way to stick it" to Borland was such a "QB3 preannounce to hold off Turbo buyers."⁵

In the same document, the Microsoft manager wrote that Microsoft was "not as far along on the response to [Borland's] Turbo C," a second product, because Microsoft was "further from product announcement." According to the Microsoft document, the Microsoft manager:

• developed a rollout plan for [Microsoft's products] QuickC and CS that focused on minimizing Borland's first mover advantage by preannouncing with an aggressive communication campaign.⁶

The manager was given the highest possible rating on his performance review (a "5-") for his "public relations" handling of this "C preannouncement."⁷

Perhaps even more striking than the incongruence between Microsoft's representations and its own documents is the silence by the Department, both in its written submissions and in its oral presentation to the Court, regarding its findings on this and other matters. The Department has not taken the position (nor, presumably, could it, without some explanation of the documents that have been submitted to it) that Microsoft has not engaged in practices such as predatory, preannouncements, or the seeding of what are referred to as "undocumented calls" (secret elements in an operating system that make a competitor's applications program operate less well than a rival Microsoft program).⁸ Instead, the Department simply has asserted that it had determined that "no further action was warranted" on these matters -- presumably a conclusion that it asks this Court to take completely on faith, since it has provided the Court with literally no explanation for its decision.

Most remarkable of all, however, is the absence of any information in any of the Department's submissions regarding the adequacy of its proposed remedy for Microsoft's illegal monopolistic conduct. Based on the Department's own allegations, from 1988 to 1994 Microsoft used a variety of illegal tactics to maintain its monopolistic share in the rapidly growing operating systems market -- and thus increased the size of its installed base through the use of illegal tactics from no more than 18 million⁹ to approximately 120 million users.¹⁰ Having acknowledged that Microsoft thus illegally acquired its massive installed base, the Department nonetheless has failed to proffer any basis for concluding that simply prohibiting these practices in the future will remedy the unassailable position that Microsoft has gained as a result of its unfair and illegal practices.

Certainly no one in the industry believes that the Department's proposed remedies will have the slightest effect in unseating Microsoft from the position that it now illegally occupies. As one competitor observed after the consent decree was announced, "[t]he consent decree seems to have set [Microsoft] free .... Now, they are running rampant over everything."¹¹ Microsoft entirely agrees. As Bill Gates observed in his response to the proposed decree:

• None of the people who run [Microsoft's seven] divisions are going to change what they do or think or forecast. Nothing. There's one guy in charge of [hardware company] licenses. He'll read the agreement.

Elizabeth Corcoran, Microsoft Deal Came Down to a Phone Call, Washington Post, July 18. 1994, at Al (Ex. 42).

Nor have events since the decree was proposed provided the slightest basis for believing that the Department's proposed remedy will have any effect. In a nationally televised press conference on July 16, 1994, Attorney General Janet Reno predicted that the Department's settlement with Microsoft would have two results: it "will save consumers money [and] enable them to have a choice when selecting operating systems."¹² In fact, however, in the six months since the proposed settlement was announced, press reports indicate that Microsoft has literally doubled the price of its operating system to computer manufacturers.¹³

Moreover, far from the decree leading to an increase in competition in the operating systems market, a key competitor in that market, the maker of DR DOS, has subsequently withdrawn from the market. The competitor observed in withdrawing from the market that "the battle for the desktop is over and MS DOS and Windows have won."¹⁴ The withdrawal of DR DOS from the market is of particular note since it was DR DOS that the authors of Hard Drive pointed to as providing the most likely source of meaningful competition to Microsoft in the operating systems market. See Hard Drive, supra, at 398.¹⁵

Having failed to explain how its proposal will remedy Microsoft's illegal acquisition of its massive installed base in the operating systems market, the Department's submission does not even touch on Microsoft's use of that illegally acquired installed base to leverage into -- and acquire market power in -- other software markets. In analyzing the strength of the Department's case against Microsoft, Hard Drive identified Microsoft's weakness in application programs as the principal reason (apart from the competition provided by products such as DR DOS) why Microsoft's dominant position arguably would not hurt consumers. With respect to application programs, the authors in 1992 argued that

• Microsoft does not come close to dominating the Big Three of applications--word processing, databases and spreadsheets. WordPerfect is far ahead of Microsoft Word, Lotus 1-2-3 is still ahead of Excel, and Microsoft has nothing to compete against Ashton-Tate's dBASE.

Hard Drive, supra, at 398.

What a difference three years can make -- at least when, like Microsoft, a company can leverage its installed base in operating systems, and finance early losses in applications with monopoly profits from operating systems. Under the headline "MICROSOFT'S DOMINATION," Dataquest Inc. has reported the 1994 market revenue and share figures for the applications market:

• "Lotus 1-2-3, WordPerfect, dBASE, Paradox and Harvard Graphics once dominated their respective categories," said Dataquest analyst Karl Wong. "Today, Microsoft products have replaced each of these one-time product category leaders."

Microsoft's Domination, San Jose Mercury News, December 21, 1994, at 1F (Ex. 35).¹⁶

Microsoft did not achieve its dominant position in operating systems and applications through free and open competition on a level playing field. Rather, it used the illegal tactics challenged in the Government's complaint to create a huge installed base in operating systems. Then, it took unfair advantage of its installed base to give its own applications group a head start and its programs a performance advantage over applications competitors -- precisely the concern voiced in Hard Drive¹⁷ and echoed by this Court.¹⁸

"Microsoft has never had a hit among its MS-DOS applications programs."¹⁹ Yet, in the past few years, Microsoft has come from nowhere to provide the lion's share of business application programs.²⁰

As explained in this brief, Microsoft achieved that result by the illegal tactics charged by the Government, and by illegal tying techniques, monopoly leveraging, and otherwise predatorially exploiting its monopoly position in one market to achieve market power in other markets. Because of the type of economic forces that prevail in these markets, rigorous economic analysis predicts that, unless restrained by Government action, Microsoft will succeed in using its dominance in operating systems to monopolize all other aspects of transaction software, from desktop applications to online systems. Microsoft's goal is to identify and control every "strategic component," "choke point" or "leverage point" in the information economy.²¹ And Microsoft is already close to achieving a complete lock-in in desktop applications.

This Memorandum of Amici argues that the Proposed Final Judgment is not in the public interest and should not be entered by this Court. Indeed it is economically impossible to achieve the stated goals of greater choices and lower prices for operating systems without (1) addressing the increase in installed base that Microsoft has procured through illegal practices and (2) restraining Microsoft's use of that installed base to dominate the markets for applications and other software products.

This Memorandum of Amici is divided into seven sections. This first section provides a summary and overview of the brief. The second section addresses the scope of investigation and power of this Court under the Tunney Act. In particular, the second section argues that, under 15 U.S.C. ｧ 16(e), the Court not only can but should consider the effect of the proposed decree beyond the operating systems market. The section further argues that the Department's submission falls far short of providing the Court with an adequate record upon which to act, and provides no factual predicate for concluding that the decree's remedy is even arguably within the "public interest" under Section 16(e).

The remainder of the brief explains that the Government cannot effectively restore and maintain competition -- even in the operating systems market -- without addressing both the consequences of the "installed base" that Microsoft increased through illegal means, and the use of Microsoft's resulting market power more broadly. Section III describes the markets and technologies in which Microsoft operates and lays a foundation for an understanding of Microsoft's conduct and strategic direction. The section begins by describing the interrelationships among complicated software technologies and demonstrates that the various markets in which Microsoft competes are parts of a large network that can be entered by a competitor's product through a few key gateways, the principal gateway being the desktop operating system. Using economic analysis, the section then argues that the economic characteristics of the technologies and markets at issue differ markedly from other, more conventional industries, in that these products (software products) and markets (networks) exhibit "increasing returns," also sometimes called "network effects." The section discusses the underlying characteristics of the technology that gave rise to these conditions and the likely consequences that these circumstances will produce.

Section IV of the brief explains Microsoft's strategy and evaluates Microsoft's prospects for complete domination of all of the interconnected software markets. The section begins by explaining that Microsoft increased its "installed base" in operating systems through the illegal practices charged in the Government's complaint. The section then explains and documents the fact that Microsoft pursues a strategy of leverage from "gateway" markets, like the desktop operating system in which it is dominant, to strategic markets in which its competitive position is weak (as was the case in applications). Microsoft targets such strategic markets, establishes marketing and technological links to those markets from established monopolies in gateway markets, and leverages its power to monopolize the target markets. In other words, it transfers the installed base of a gateway market it dominates to create an installed base in the strategic target market. The section focuses primarily on the desktop market, describing in some detail the method by which Microsoft (according to the Government's Tunney Act filing) used illegal activities to increase its installed base in operating systems and then leveraged its monopoly over the operating system to dominate applications. In particular, the section describes Microsoft's tactics of bundling and unbundling functions into and out of its operating system to disadvantage its competitors in the applications market.

Section V of the Memorandum of Amici applies "increasing returns" economics to suggest that Microsoft likely will achieve a monopoly position for its products throughout the entire personal computer network unless restrained by Government action. The section rejects various arguments that could be put forward to justify such monopolization, including the arguments (1) that alternative networks created by alliances of competitors will provide competition, and (2) that the benefits derived from integration of a single product line are worth the cost in loss of free competition throughout the network. The section concludes by suggesting that absent meaningful governmental intervention, the American software industry will be monopolized by Microsoft, with the only competition coming from protected markets and competition abroad.

Section VI evaluates the possibilities and prospects for governmental intervention from the legal 'perspective. The section begins with an evaluation of the proposed Final Judgment, observing that the Government's Tunney Act filing concedes that Microsoft. through the use of illegal practices, has acquired an enormous installed base that constitutes an overwhelming barrier to entry. The only sanction proposed by the Government, requiring Microsoft to cease the behavior that permitted it to acquire this entrenched installed base, will have no effect in diminishing the installed base, easing barriers to entry, or otherwise precluding Microsoft from using the illegally acquired installed base to monopolize the operating system market or other markets. The section considers specific strategies for relief adopted by previous Administrations in comparable situations and analyzes legal precedents supporting such strategies.

Finally, Section VII of the brief proposes procedures this Court may wish to adopt in order to exercise its appropriate role in Tunney Act proceedings. The section urges the Court to order the production of key Microsoft documents and to require the Government to produce detailed and predictive economic models of the type previously employed to support consent decrees adopted through Tunney Act procedures.

THE PERMISSIBLE SCOPE OF THIS COURT'S REVIEW

In 1974 Congress enacted the Antitrust Procedures and Penalties Act ("APPA"), also known as the "Tunney Act." 15 U.S.C. .. 16(b)-(h) (1994), out of concern with "prior practice, which gave the [Justice] Department almost total control of the consent decree process, with only minimal judicial oversight." United States v. American Tel. & Tel., 552 F.Supp. 131. 148 (D.D.C. 1982) ("AT&T"), aff'd sub nom. Maryland v. United States, 460 U.S. I001 (1983). To remedy this practice, Congress sought to eliminate "judicial rubber stamping" of such consent decrees,²² providing that "[b]efore entering any consent judgment ... the court shall determine that the entry of such judgment is in the public interest." 15 U.S.C. ｧ 16(e). Circuit Judge Aldrich, sitting by designation in United States v. Gillette Co., 406 F.Supp. 713 (D. Mass. 1975) (cited by both the Department and Microsoft), observed upon reviewing the legislative history of the Act:

• The legislative history shows clearly that Congress did not intend the court's action to be merely pro forma, or to be limited to what appears on the surface. Nor can one overlook the circumstances under which the act was passed, indicating Congress' desire to impose a check not only on the government's expertise -- or at the least, its exercise of it -- but even on its good faith.

Id. at 715.²³

Despite this clear statutory intent, the oral and written submissions in the present case have suggested that the Court's review should be circumscribed in ways not supported either by the statute or by existing case law. First, the submissions may be taken as suggesting that the Court should look only to the impact of the proposed decree on the operating system market in determining whether the decree is in the public interest. See, e.g., 59 Fed. Reg., at 59,429. The law, however, plainly is otherwise. For example, in United States v. BNS Inc., 858 F.2d 456 (9th Cir. 1988), -- a case relied upon by the Department -- the Court observed that "the statute suggests that a court may, and perhaps should, look beyond the strict relationship between complaint and remedy in evaluating the public interest." 858 F.2d at 462 (quoting United States v. Bechtel Corp., 648 F.2d 660, 666 (9th Cir.), cert. denied, 454 U.S. 1083 (1981)). While the court's public interest determination may not be based on a different market from the one identified in the complaint, the Ninth Circuit emphasized that this did not mean that only effects on that market can or should be considered:

• [T]he statute clearly indicates that the court may consider the impact of the consent judgment on the public interest, even though that effect may be on an unrelated sphere of economic activity. For example, the government's complaint might allege a substantial lessening of competition in the marketing of grain in a specified area. It would be permissible for the court to consider the resulting increase in the price of bread in related areas.

Id. at 463 (emphasis added).

Under the Department's own authority, therefore, the Court's inquiry is not limited to the effect of the proposed judgment on the operating system market. To the contrary, the Court can (and, it is submitted, should) determine the effect of the proposed judgment on other areas impacted by Microsoft's monopolistic conduct. As will be discussed in more detail in Section IV, infra, for example, Microsoft has used its illegally acquired market position to leverage into and acquire a monopoly in other related markets. The failure of the decree to "break up or render impotent [this] monopoly power found to be in violation of the Act." AT&T, 552 F. Supp. at 150 -- indeed, its tacit decision to leave Microsoft free to profit from its unlawful market power by leveraging into other software markets -- is something that the Court should consider in evaluating the public interest served (or disserved) by the proposed decree.

A second limitation implied in the submissions to the Court also is without authority in the case law. namely, that the Court is limited to considering those matters that the Department has identified in its complaint. That is not the law. See, e.g., BNS, 858 F.2d at 462 ("a court may consider matters not discussed in the complaint"); Gillette, 406 F.Supp. at 715 ("Congress did not intend the court's action to be... limited to what appears on the surface"). Indeed, simply accepting at face value the Department's analysis -- and even its good faith -- amounts to precisely the kind of "rubber stamping" that the APPA expressly rejects. The Court is required, in evaluating the Department's proposed decree, to determine whether it "meets the requirements for an antitrust remedy -- that is, if it effectively opens the relevant markets to competition and prevents the recurrence of anticompetitive activity." AT&T, 552 F. Supp. at 153. If the Department has determined not to address a practice -- for example, Microsoft's "bundling" of operating and applications programs, discussed in more detail in Section IV, infra -- which forecloses any meaningful chance of competition in the operating systems market, that fact must be considered by the Court in assessing the adequacy of the decree as a remedy for the charged violations. That is so regardless of whether the Department has chosen to turn a blind eye to the consequences of such bundling on the effectiveness of its proposed decree.

Finally, prior submissions to the Court have emphasized that in assessing whether the decree is in the "public interest" under Section 16(e), the Court should not "determine whether the resulting array of rights and liabilities is the one that will best serve society, but only to confirm that the resulting settlement is within the reaches of the public interest." United States v. Western Electric Co., 900 F.2d 283, 309 (D.C. Cir. 1990), cert. denied, 498 U.S. 911 (1990) (citations and quotations omitted; emphasis in original). This standard clearly is correct. but the parties' further assertion -- that the submissions already made by the Department are sufficient to satisfy this standard -- equally clearly is not.

A comparison of the information provided in those cases relied upon by the Department, with that provided here, highlights just how far short the Department has fallen in providing this Court with an adequate record upon which to act. For example, the Department relies heavily upon the Court of Appeals' decision affirming a modification of the consent decree in United States v. Western Electric Co., Inc., 993 F.2d at 1572. See 59 Fed. Reg. at 59,429.²⁴ However, in finding that there was a sufficient "factual foundation for the judgment call made by the Department of Justice and to make its conclusion reasonable," 993 F. 2d at 1582, the Court of Appeals in that case expressly pointed to the "array of prominent economists (including two Nobel laureates, Stigler and Arrow)," who had submitted affidavits in the record that supported the Department's position. These affidavits provided detailed support for the factual predicates underlying the Department's proposal, including the view that the Bell operating companies would not be able to discriminate or engage in cross-subsidization: that government oversight would be effective in regulating their behavior; and that the proposal would enhance competition in the relevant markets. See id. at 1578-82.

This Court, by contrast, has not been provided with the affidavit of any economist, or for that matter of anyone else, that would provide a factual predicate for any of the matters that it must decide in reviewing the adequacy of the proposed decree. The Department has provided no factual basis (other than its say-so) for believing that the remedies proposed in the decree would be sufficient to "pry open to competition" the operating systems market, AT&T, 552 F. Supp. at 150; that Microsoft's other anticompetitive practices (undocumented calls, predatory preannouncements, anticompetitive bundling and unbundling, early disclosure to Microsoft applications programmers) will not undermine the effectiveness of the decree; and so forth. Although this case involves an industry of unquestioned significance to the future of the American economy -- one of comparable importance to AT&T itself -- the Department has in fact given this Court nothing to go on other than the purest ipse dixit. Indeed, it is hard to imagine how the Department could claim that its request for approval of the decree amounts to anything but a request for a "rubber stamp" when it has so notably failed to say anything other than "trust us."

Nor does the Department's submission compare favorably with the information available to other courts in cases cited by the Department. In Gillette, for example, which first formulated the "reaches of the public interest" standard, see 406 F. Supp. at 716, Judge Aldrich concluded that he was able to make an independent determination regarding the adequacy of the proposed decree because "the record [in the case] is both open and extensive." Id. at 715. Here, the record is neither. Indeed, the transcripts of the hearings on September 29, 1994 and November 2, 1994 are replete with inquiries by the Court regarding matters inextricably tied to the adequacy of the proposed remedy -- inquiries that repeatedly failed to yield any information at all, or (even worse) information that is at odds with the record.

The example of preannouncements already has been discussed above: despite Microsoft's unequivocal denial, and the Department's silence, the documentary record shows that such predatory preannouncements in fact are used by Microsoft. Nor is this the only example highlighted by the transcript. Equally striking is the Court's effort to ascertain whether the Department had concluded that a "Chinese Wall" exists between Microsoft's operating system and applications divisions. Noting the discussion of this point in Hard Drive, the Court may have been left with the impression during the hearing that such a "Chinese Wall" in fact exists. See Tr. of Status Call, Sept. 29, 1994, at 27:11-28:1. Certainly that is the impression that Microsoft previously has sought to convey, dating all the way back to 1983.²⁵ Indeed, throughout the spring and summer of 1991, after the FTC announced its investigation of Microsoft in March 1991, Microsoft persisted in its claim that the company's applications and systems development groups were separated.²⁶

Now, however, at the end of a long footnote in its written submission, Microsoft disavows that any such "Chinese Wall" exists -- and, indeed, derides the idea as "irrational." See Microsoft Mem. at 7 n. 12. The Department, again, has been silent. Was its determination that "no further action [is] warranted" on this issue, 59 Fed. Reg. at 59,427, based on Microsoft's earlier representation that a "Chinese Wall" in fact exists? Was it based on the conclusion that there is no "Chinese.Wall," but it does not matter? If not, why not'.)

The answers to these and other questions may remain unanswered because no satisfactory answer is available. As shown in Sections III through VI, infra, the Government cannot effectively restore competition in the operating systems market without addressing the consequences of Microsoft's illegally-acquired "installed base," and its broader use of its acquired market power. The Government's proposed consent decree, however, fails to do either.

THE ECONOMIC CHARACTERISTICS OF THE SOFTWARE INDUSTRY

Section III is divided into two parts. Subsection A provides background by describing the structure of the software industry and how it has changed over time in response to Microsoft's prior conduct in the market. Subsection B describes the economic characteristics of the technologies and markets at issue here.

1. • Market And Technology Background

The relevance of much of the material in this section, particularly the schematic diagrams, is fleshed out and explained to a great extent in the subsequent sections. If the Court is unfamiliar with these markets, the Court may find it useful at this point to read The Economist²⁷ article, and the Harvard Business Review²⁸ article, both found in the Appendix.

At the outset, two characteristics of these markets and technologies should be emphasized. First, the products at issue are software products, composed almost entirely of intellectual property content. Because of the nature of software, there can be greater flexibility in the formation of vertical relationships than often is present with respect to more conventional products. Unlike a pipeline, for example, many competitors can vertically link their software, through software compatibility, to products in the markets above and below them. So, for example, a number of different companies can make word processing application programs that work equally well with Microsoft's operating system so long as they all have the same technical information on a timely basis. It is not necessary for Microsoft to bundle -- or literally tie together -- its operating system and word processing program in order to ensure that the two programs work well together. With software, the efficiency benefits of vertical integration can be achieved without foreclosing access to competitors.

Second, the Stipulated Complaint and Final Judgment in this case focus on the Personal computer operating system and the applications that run on top of it. Together, the personal computer operating system and the applications that run on it are sometimes known as the "business desktop." But the desktop is really only an interrelated component of a network that contains desktops (or "clients") and "servers." These software networks bear many of the characteristics that economists have associated with networks in other industries, including "increasing returns" or "network effects," as described in Subsection B. Indeed, software networks manifest increasing returns, or demand-side economies of scale, more strongly than networks in more conventional industries.

The network at issue here has four components, two on the "business" side and two on the "home" side. On both the home and business sides, there is a desktop, or "client," component, and a "server" component that links the desktop into a broader network. The network as a whole can be diagramed as follows:

The following description attempts to provide some explanation for each of these boxes: the intrabusiness client, which runs on the "desktop"; the enterprise "server,"' meaning the hardware and software applications that run on a more centralized computer and that link the clients together; the home "client;'" and the home-to-business server, that similarly links home personal computers ("PCs") into a. larger network. This brief then discusses two particular technologies that play a critical role in understanding Microsoft's strategy: OLE and Windows.

1. 1. • The Business Desktop

The personal computer or "PC" was initially devised as a stand-alone device, but today it is usually used as part of a network. This is certainly the case in business, and will increasingly be the case in the home.²⁹ The PC, both stand-alone and as part of a network, is often referred to as "the desktop." The FTC Investigation and the DOJ investigation of Microsoft have focussed on the desktop.

Prior to Microsoft Windows, the intrabusiness "client side" or desktop could have been thought of as having four layers.

Figure 2

Today, the market looks more like Figure 3 below. It reflects two principal changes, each of which will be explained in Section IV, infra. First, Microsoft succeeded in forcing the market to migrate to a new operating system or "OS" (Windows), thereby inserting a new layer, the "graphical user interface" (GUI) layer (layer 3), between the operating system and the applications. Second, using its leverage in layers 2 and 3, it has become dominant as well in development tools (layer 4) and business applications (layer 5).³⁰

Figure 3

The Justice Department investigation of Microsoft has focussed primarily on operating systems (Levels 2 and 3 in Figure 3), -- but the Government's Tunney Act submission also considers the applications layers (Levels 4 and 5) insofar as they impact competition in operating systems. In order to evaluate the proposed Final Judgment, a slightly more detailed understanding of the operating system layer is necessary.

The Government's complaint defines the market as operating systems that run on the Intel chip set (known as "X86" chips). 59 Fed. Reg. at 42.847 (Complaint "¶ 13). There were formerly three principal operating system vendors for this market -- Microsoft (MS DOS and Windows), Novell (DR DOS) and IBM (PC-DOS and OS/2). Novell, as indicated above, has withdrawn from this market, and Microsoft is unquestionably a monopolist, currently enjoying a greater than 90% market share.³¹ Software written for the current version of Windows (v. 3.1) and prior versions will also run on the IBM OS/2 operating system. However, software written expressly for Microsoft's next release of Windows (Windows 95), due out in August of 1995, will not run on the IBM OS/2 operating system. Don Clark and Laurie Hays, Microsoft's New Marketing Tactics Draw Complaints, Wall St. J., Dec. 12, 1994, at B6 (Ex. 41).

There are a few other competing desktop operating systems that run on different chip sets. For example, Apple's Macintosh operating system runs on a Motorola chip set. And the UNIX operating system generally runs on a specially designed chip, such as the "RISC" (reduced instruction set) chip designed by Sun Microsystems.

Even including these other operating systems in the same market as those that run on the Intel chip, Microsoft has an overwhelming market share, with well over 85 %. As the Government's Complaint correctly points out, applications software written for an Intel chip operating system will not run on the Apple Macintosh or Sun RISC workstation without significant modification -- known as "porting." Frequently, porting application software to a new chip set and operating system entails a significant re-engineering of the software. Hence. the Government does not include operating systems for the different chip sets within the same antitrust market.

However, the Government fails to point out that the only companies in the market for developing business application software for the operating systems sold by Apple and Sun, for example, are also the business application vendors on the Windows platform -- e.g., Novell/WordPerfect, Lotus, Borland, etc., and Microsoft, itself, of course. The significance of this fact is discussed in greater detail infra. The point here, however, is that if Microsoft were able to monopolize the market for business applications software, it would severely inhibit competition from vendors of operating systems that run on other chips but nevertheless compete with the Microsoft operating system (e.g., Apple and Sun).³²

Figure 4 shows what the intrabusiness client side probably will look like once Microsoft's strategy of vertical integration of markets within the client is completely executed. It shows the completion of Microsoft's leverage from layers 2 and 3 to further its domination of all aspects of layers 4 and 5.

Figure 4

1. 2. • The Intrabusiness Server

The "server" is the direct lineal descendant of the mainframe computer. Prior to the advent of the personal computer, companies operated using a mainframe, to which "dumb" terminals were connected. Personal computer technologies now allow many computing functions to be performed on the desktop by an individual worker, but workers within a business still need to share information with each other and access a body of data simultaneously. The "server," a dedicated hardware platform with its own server operating system, allows this to happen. Indeed, increasingly, workers within a business will want simultaneous access to several bodies of data and several different application programs, so that, for example, textual documents containing spreadsheets can be prepared by a number of employees working at the same time.

There are two basic components of the server markets. The intrabusiness server is the backbone of business. Microsoft has projected that there will be 300 million servers in the business community, running everything from phone systems, to copying systems, to cash registers. J. William Semich, The Long View From Microsoft: Component DBMSs. Datamation, Aug. 1. 1994, at 40 (Ex. 10). If a Single company controls all business server markets and applications, that company has far greater market power in various sections of the economy than, say, mere control of the desktop would bestow. The second server component. home-to-business, will be described in a subsequent section.

Today, the "server" side of the intrabusiness environment has approximately eight layers. It would unnecessarily complicate this brief of amici to describe the intrabusiness server markets in great detail. There are, however, three important points about the intrabusiness server markets that are relevant for this Court's consideration. First, the most important layer in the server market is the operating system level. The two leading competitors in this market at present are Novell's "Netware" product and Microsoft's NT product.³³ The operating system is important because the other products in the server market run on top of the server operating system in much the same way as desktop applications run on top of Windows. The operating system level is also important because it is the level through which the server is connected to the business desktop and (through on-line services) to the home client.

Second, as was the case on the desktop four years ago, competition is vigorous at all levels of the server market. At each of the eight levels, there are a number of competitors, each striving to make better products at cheaper prices. This condition represents a significant (and welcome) departure from the state of the computer industry prior to the advent of personal computer and server technology. In an earlier period, there were only a few vertically integrated companies in the computer industry, such as IBM, DEC and Wang. These companies attempted to supply all aspects of computer technology -- from the underlying chips and operating systems, to applications, to distribution, and even including service and support of previously sold computers. Generally speaking, consumers have benefitted enormously by the fragmentation of the industry into horizontal layers characterized by vigorous competition. Consumers have been able to choose the technologically superior and most cost effective product at each level and combine those products into a system that addresses the consumers' needs. The pro-competitive benefits of the industries' current horizontal alignment is discussed in some detail in the Economist article (Ex. 14).

Finally, Microsoft is pursuing a vertical integration strategy on the intrabusiness server side similar to that pursued on the business desktop side. This strategy is only briefly discussed elsewhere in this paper. The Court can get further information concerning Microsoft's strategy, goals and prospects for success from the following articles found in the Appendix: Stuart J. Johnston and Ed Scannell, Server Suite Could Squeeze Market, Computerworld, Oct. 10, 1994, at 4 (Ex. 7); How Microsoft's Server Strategy Will Change The Industry - Parts I & II, Report by Summit Strategies Inc.; J. William Semich, Datamation, Aug. 1, 1994, supra, at 40 (Ex. 10). Obviously, after complete execution of this strategy, Microsoft products would be dominant or exclusive on each of the server layers.

1. 3. • The Home-to-Business Server

The second aspect of server technology is the home-to-business server market, sometimes known as "online services." Today, most online services run off mainframe computers the way LEXIS and NEXIS do. Businesses will increasingly need to sell directly into the home through online services in order to remain competitive. Control by a single company of the home-to-business server market would have significant economic ramifications.

Although there is a vigorous online services market in place, the home-to-business server does not yet exist, except in Microsoft's plans. It can be readily assumed that the home- to-business server would look much like the intrabusiness server, with only Microsoft products being vertically integrated.

Figure 5

1. 4. • Home Computer Market

The home computer market is in its incipiency. The most important applications programs on the home client are "home banking" (also sometimes known as "personal finance") and tax preparation.³⁴ The most successful company in this market, Intuit, Inc., makes the largest selling home banking ("Quicken") and tax preparation ("TurboTax") programs. The only substantial competition to Intuit's products comes from Microsoft. Yet, despite a very. substantial commitment in marketing staff and resources, Microsoft has gained only a 10% share. Microsoft has therefore elected to take over the home finance market by purchasing the leading software developer, Intuit, rather than by making better products to compete against it. The Microsoft acquisition of Intuit was announced on October 13, 1994 and is still under review by the Department of Justice. It is the largest acquisition in the history of the industry with Microsoft paying twice as much for Intuit as that company was worth in the stock market.³⁵

The Microsoft acquisition of Intuit is highly strategic. It is a key element in Microsoft's plans to dominate all of information processing and will be discussed in a subsequent section. If the Microsoft-Intuit deal is consummated, it is not difficult to project what the home client will look like given Microsoft's recent announcement concerning "Marvel" (described in a subsequent section).

Figure 6

In summary, in each of the four components of the software industry, Microsoft's overall business approach and strategy is based on the creation of technological linkages between layers within the same market (e.g., DOS to Windows on the desktop) and between layers in one market and corresponding layers in another market (e.g., Windows NT to the Microsoft Network to Windows 95 on the home client). To fully understand Microsoft's strategy and its economic implications, however, it is necessary to understand two additional strategic Microsoft technologies: OLE and Windows. This Memorandum of Amici will address each in turn.

1. 5. • OLE

OLE (object linking and embedding) is a strategic technology for Microsoft on both the client and server side. It is the Microsoft-imposed standard for sharing information both among applications, and between applications and the operating system. During the Justice Department investigation, desktop application companies complained that Microsoft seeded OLE to its own application developers before giving it to ISV's (independent software vendors), thereby giving its own applications a lengthy head start over the competition.³⁶ As set forth in a subsequent section, these charges are supported by ample evidence and constitute the clearest examples of Microsoft's use of operating system information and specifications to achieve an unfair head start in the application markets. This is precisely the issue raised by this Court.³⁷

Even more striking is the fact that Microsoft continues to exercise the very same strategy on the server side. See, e.g., J. William Semich, Datamation, Aug. 1. 1994, supra, at 40, 41-44 (Ex. 10) ("If you think OLE is everywhere in the future, the answer is yes"). Microsoft has made it clear that OLE will be strategic technology for the home-to-business server market, but Microsoft has not provided sufficient specifications to independent database server providers to enable them to release equally well-behaved products on the same time schedule as Microsoft's own products.³⁸

1. 6. • Windows

The business desktop connects to the server through the Windows operating system ("OS") and the home-to-business server ("online services") also connect to the home computer through the Windows operating system. Microsoft has several different Windows products that provide OS, GUI and networking capabilities. A brief (and superficial) description of these products is included at this point to avoid confusion.³⁹

1. 1. 1. • Desktop

Microsoft's first Windows products were targeted for the desktop and were built on top of Microsoft's dominant desktop operating system MS-DOS. Because of their DOS legacy, these products are unable to take full advantage of the capabilities of the 32-bit microprocessors they run on. Microsoft's current product in this area is Windows 3.1, which. due in part to the illegal per-processor licensing challenged by the Government, is pre-installed on most desktop systems presently sold.

Microsoft plans to proliferate Windows 95 (also known in the press as "Chicago" or "Windows 4.0") widely next year as the successor to Windows 3.1. Windows 95 is a true 32-bit operating system, but it is being targeted to the mainstream personal computer market. It also includes advanced networking features.

Windows NT was Microsoft's first true operating system for 32-bit microprocessors. NT's principal use is in the server market (discussed below) but Microsoft has also targeted its NT marketing to power users running high-end personal computers or workstations.

1. 1. 2. • Server

Windows NT can also be used as an operating system for a network server. Microsoft markets a version of NT with advanced server capabilities, called Windows NT Advanced Server, as an enterprise-wide computing solution. Microsoft offers a suite of applications for Advanced Server called "BackOffice" that includes database services, electronic mail, systems management, and connectivity to mainframe and minicomputers.

Microsoft's vision for enterprise computing is being marketed through its plans for a replacement for Windows NT currently code-named "Cairo." Cairo brings object-oriented technology into the file server and operating system. Microsoft already controls object standards through it OLE specification, discussed in the next subsection. See J. William Semich, Datamation, Aug. 1, 1994, supra, at 41-44 (Ex. 10).

2. • Free Market Forces in lncreasing Return Industries

In some industries, companies generally compete on a "level playing field." In such industries, diminishing returns to scale ensure that the forces of the free market will naturally gravitate toward an equilibrium point which maximizes the production of goods and services and results in the most efficient allocation of resources. Under such conditions, antitrust enforcers as well as business executives can count on the fact that superior products will necessarily prevail in free and open competition.⁴⁰

Free market forces in other industries -- including those at issue here -- do not exhibit such qualities. Rather, they exhibit "increasing returns." In such industries, there is more than one equilibrium point and there is no reason to expect the free market to reach equilibrium at a point that most efficiently allocates resources.⁴¹ The markets in such industries can easily be manipulated by a company with a large "installed base,"⁴² with the result that superior products of competitors are not likely to prevail in the free market.⁴³ Indeed, in "increasing returns" industries, there is every reason to believe that consumers will get "locked into" the first product that appears on a new platform, even if the product is technologically inferior.⁴⁴ Similarly, a company with a large installed base in one market can give its inferior product in a second market an insurmountable advantage over competitors in the second market by integrating the products from the two markets together technologically.⁴⁵

Some of the early economic research in the area focused on perceived anomalies -- particular standards that became locked in, notwithstanding their obvious inferiority. Stanford economist Paul David identified several such examples, the most famous of which is the layout of the common typewriter keyboard, known as the "QWERTY" configuration because of the order of the keys in the second row of the keyboard.⁴⁶ Primitive typewriters were unreliable mechanical devices and the QWERTY keyboard, at least according to the folklore, was therefore deliberately designed to be dysfunctional so that typists would not strike the keys so rapidly that the device would jam. Obviously. modem software and computers can process keystrokes far more quickly, yet consumers are locked into the QWERTY standard. There are even allegations "that the combination of constant repetitive motion and inefficient finger movements that QWERTY requires is the ticket to the most well-known [repetitive stress injury.] RSI. carpel tunnel syndrome." yet we go right on teaching it in elementary schools.⁴⁷ Superior keyboard layouts were developed years ago but were unsuccessful in dislodging the clearly inferior design that established itself as an early standard.⁴⁸

By the late 1980's, economic analysis was finally able to explain such situations more clearly. Economists at Stanford and the University of California at Berkeley published " leading articles demonstrating that market characteristics long viewed as anomalous were, in fact, widespread in high technology industries.⁴⁹ By the mid-1990's, increasing returns economics has become widely accepted as mainstream economic analysis.⁵⁰ There is now extensive theoretical literature with direct empirical application to many leading industries, including telecommunications, broadcasting, computers, and ATMs.⁵¹

Increasing returns are present in industries throughout the economy, but two high technology market situations, in particular, give rise to increasing returns. First, users of high technology products are frequently electronically connected in a network. Networks exhibit and produce certain important economic results. Because the purpose of a network is to enable communication with others, the value of the network increases with the total number of users who join the network.⁵² Consequently, once a network such as a telephone network is in place. a competing network would have to enter the market with at least as large a number of nodes in order to displace (or even compete meaningfully with) the first network.⁵³

A second factor that gives rise to increasing returns is referred to as "compatibility" in the economic literature. Unlike more conventional industries, the value of the technology to end users in increasing returns industries increases with the number of users who use compatible technology. While the "network" feature draws its force from physical interconnection, the "compatibility" factor arises from a dependency of mutual use by consumers without regard to actual physical interconnection.⁵⁴ For example, although manual typewriters were not connected in a physical network, new users adopted the QWERTY keyboard because it was in wide use by others.⁵⁵

Economic analysis demonstrates that superior products do not necessarily prevail in markets and technologies that exhibit increasing returns. Rather, these markets are easily susceptible to "tipping" -- once moved off of equilibrium by an event, the market tends quickly toward a single standard that dominates the market:

• [N]etwork markets are 'tippy': the coexistence of incompatible products may be unstable, with a single winning standard dominating the market. The dominance of the VHS videocassette recorder technology and the virtual elimination of its Betamax rival is a classic case.

See Stanley M. Besen and Joseph Farrell, Choosing How to Compete, J. of Econ. Perspectives, Spring 1994, at 118; see also Michael Katz and Carl Shapiro, J. of Econ. Perspectives, Spring 1994, supra, at 106. Once a market is "tipped" in favor of a particular competitor, it would take truly massive forces to return the market to a state of equilibrium (i.e., competition). See, e.g., W. Brian Arthur, Increasing Returns and Path Dependence in the Economy, supra, at 2, 10-11.

Creating a large installed base is the key to dominating such an increasing returns market.

• Because of the compatibility and network benefits, all else equal, a new user prefers a vendor with a larger total installed base of users. Thus installed bases have a tendency to be self-perpetuating: they provide the incentive for the provision of products (software and hardware) that is compatible with the installed base which in turn attracts new users to the installed base further swelling its ranks ....

Garth Saloner, Econ. Innov. New Tech., 1990, supra, at 140. Indeed, "de novo entry into a market occupied by vendors with large installed bases is exceedingly difficult." Id. at 140.

The self-perpetuating nature of an installed base in an increasing returns industry causes particular products to become "locked-in." W. Brian Arthur, Scientific American, Feb. 1990, supra, at 99 (Ex. 36). The costs to a consumer of using or switching to a different system are so high that the vendor with the installed base has a substantial advantage over competitors and can, once the base is established, charge consumers supracompetitive prices.⁵⁶

Because increasing returns markets are particularly susceptible to "tipping," a company with a monopoly in one market that faces competition in a second market can use the locked-in installed base of the first market to wipe out competition in the second market by "tipping" the second market. The monopolist might achieve this result by releasing a "predatory preannouncement" with regard to a product in the second market. In markets that feature increasing returns, users will want to be on the same standard as other users, so expectations (what users believe will happen) dominate user choice in the second market -- as opposed, for example, to the inherent technological quality of competing product offerings,⁵⁷

• [A] preannouncement can sometimes secure the success of a new technology that is socially not worth adopting, and that would not have been adopted absent the preannouncement.

Joseph Farrell and Garth Saloner, Amer. Econ. Rev., Dec. 1986, supra, at 942.

Similarly, a monopolist that is cash rich from monopoly profits in the first market might also "buy off" early adopters to create a "band wagon effect" in favor of its product in the second competitive market.⁵⁸ This technique of predation is known in the economic literature as "penetration pricing."

• An installed base advantage might also be achieved by "penetration pricing," the technique of offering low prices to early customers so as to build up an installed base and influence the choice of later adopters. Penetration pricing seems a natural strategy in network industries, and appears prominently in the theory.

See Stanley M. Besen and Joseph Farrell, J. of Econ. Perspectives, Spring 1994, supra, at 122; see also Janusz A. Ordover and Garth Saloner, Predation, Monopolization, and Antitrust, supra.

Finally, a monopolist with a large installed base in one market might "tip" a second competitive market in favor of his product in that market by technologically linking the two products, or by outright bundling of the functionality of the second product into the first product, thereby eliminating the need for the competitor's product in the second market. For example, by subtly altering the tying product so that rival products in the tied market become incompatible with the monopolist's "standard," the monopolist can quickly dominate the second market.⁵⁹

The Justice Department's complaint in this case recognizes the critical importance of an "installed base." The complaint alleges that the "lack of a sizable installed base of users" constitutes a "substantial barrier to entry" for Microsoft's operating system competitors. 59 Fed Reg. at 42,847 (Complaint ｶ 15). The complaint also alleges that Microsoft used "anticompetitive contracting practices" including "per processor licenses" starting as early as 1988 to "significantly increase the already high barriers to entry." Id. at 42,847, 42,848 (Complaint ｶｶ 18, 20, 26). The complaint appears to assume that Microsoft's monopoly was lawfully acquired. Id. at 42,847 (Complaint ｶ 19). But since Microsoft's installed base of operating system users has increased six-fold since 1988, it must follow that the "anticompetitive licensing practices" with which Microsoft is charged had the result of increasing its own installed base at the same time it impeded the development of competitors' installed bases. As set forth in the next section, Microsoft has used its installed base both to preclude competitive entry into the operating system market, and to stifle competition in related markets.

MICROSOFT'S TACTICS AND PROSPECTS FOR SUCCESS

This section of the Memorandum of Amici will examine Microsoft's overall strategy, the tactics that Microsoft has used in pursuing that strategy, and the likelihood that Microsoft will accomplish its aims. Microsoft, by the admission of its own Chief Executive Officer, intends to dominate all of data and information processing.

• There's no level of performance or specific application of corporate information systems that we don't intend to go after... [and] there won't be anything we won't say to people to try and convince them that our way is the way to go. That's because this new, electronic world of the information highway will generate a higher volume of transactions than anything to date, and we're proposing that Windows be at the center, servicing those transactions.

Brent Schlender, Fortune, Jan. 16, 1995, supra, at 40 (emphasis in original).

To accomplish these aims, Microsoft has pursued licensing practices that the Government has denominated as "anticompetitive," and has engaged in classic predatory behavior by using its monopoly in one market to achieve monopolies in other markets. This section applies increasing returns economic analyses to Microsoft's behavior and concludes that, unless restrained by Governmental intervention, it is highly likely that Microsoft will achieve its goal of dominating the entire national information infrastructure.

1. • Microsoft's Strategy

Even if Microsoft's initial monopoly was lawfully obtained, its enormous market power (and particularly the power to leverage into related markets) comes from its installed base in operating systems. That installed base, according to the Complaint, was procured as a result of anticompetitive practices. Indeed, Microsoft's installed base of operating system users has increased more than six-fold (from 18 to 120 million) since 1988, when the company began its anticompetitive practices. Microsoft has used its monopoly and its installed base in a classically predatory manner. It has used its monopoly revenues in one market to drive competitors out of other markets. It has also used its operating system installed base in a predatory manner to "tip" adjacent competitive markets in the direction of its own product in those markets, to the detriment of competitors.

Microsoft's strategy at any particular point on the network (for example, at the home client or at the business desktop) can only be understood and evaluated in the context of Microsoft's overall strategy. Microsoft pursues a strategy of leverage from product markets in which it is dominant, to markets in which its competitive position is weak. It targets particular markets, establishes marketing and, in particular, technological links to those markets from established monopolies, and then leverages its power to monopolize the target markets.

As used in this brief, "leverage" means that Microsoft uses the installed base in a market it dominates (for example, the operating system) to create an installed base in a new market (for example, desktop applications). It uses predatory subsidization, and both marketing and technological linkages, to accomplish leverage, as explained in greater detail in the succeeding pages. For the sake of easy example, Microsoft's horizontal tie-ins within, a single layer represent the most trivial example of its marketing strategy. Thus, Microsoft has trundled for sale a number of desktop applications (under the name, the "Microsoft Office"), putting companies like Lotus, WordPerfect and Borland at a competitive disadvantage. Carole Patton, Bundles Are Bad News, Computerworld, Nov. 14, 1994, at 57 (Ex. 8). Microsoft is executing the same tactic on the server side by bundling its "BackOffice" products to foreclose meaningful competition at the "server applications" layer. See Stuart J. Johnston and Ed Scannell, Computerworld, Oct. 10, 1994, supra, at 4 (Ex. 7).

Microsoft also pursues other tactics. In particular, Microsoft derives leverage from its control of Windows products and logo; from its use of a consistent graphical user interface; and from its tight technical integration between interconnected machines through the control of standards such as OLE. After establishing market power on one level, Microsoft will target an adjacent layer, subsidize the creation and sale of products at that layer from the monopoly it derived on the first level, establish proprietary technological linkages to the target layer, and then leverage its market power to establish market power in the next layer. Two examples of this within the desk-top side are DOS to Windows, and Windows to desktop applications. In addition, Microsoft uses its market power from one side of the network (server or client) to leverage to the other side, again by establishing linkages. Microsoft is already attempting to leverage its control of the desktop into a control of servers. It will also use its market power in the PC-based financial and text software market, through the acquisition of Intuit, to leverage into the server.⁶⁰

Obviously, control of certain layers in the various markets of the network create greater potential for leverage than control of other layers. In particular, there are a few "gateway" layers into the network. Control of these layers represents the most effective platform for leverage (i.e., moving the installed base). Generally speaking, the operating system layers in each box represent the most powerful platforms for both horizontal and vertical leverage.⁶¹ For example, Microsoft has already leveraged control of operating systems to desktop applications. It can also leverage control of the desktop operating system (Windows 95) to the server operating system (Windows NT).

Control of the "gateway" layers provides greater possibilities for leverage because control of the architecture at those levels effectively controls all higher vertical levels, and also provides significant power at the horizontal interface between the client operating system and the server operating system. This brief uses the term "architecture" in the same way as that term is used in the Morris and Ferguson Harvard Business Review article -- namely, the complex of standards and rules that define how programs and commands will work and how data will move around the system. Charles R. Morris and Charles H. Ferguson, Harv. Bus. Rev., Mar. 1993. supra, at 88 (Ex. 16).

By owning the installed base at a gateway, Microsoft can control not only the architecture at that level but also at all higher vertical levels. For example, by controlling the desktop operating system architecture, Microsoft can easily obsolete or render inoperable Lotus 1-2-3, merely by making a minor change to the architecture. Microsoft can pretextually or otherwise claim the change to be an "upgrade" or a "bug fix," but it is the effect of the power to control architecture that is more important than Microsoft's subjective intent.

If Microsoft controls the architecture at a "gateway," it can loudly proclaim its system to be "open" while in truth its architecture remains closed. Thus, for example, Microsoft can claim that its desktop operating system will continue to work with Lotus 1-2-3 or that its server operating system will continue to work with the database products offered by Microsoft competitors (and, to that extent, its system is "open"). Because Microsoft can easily obtain competitive advantage over (or outright displacement of) vertically related competitors by upgrades to the architecture, however, its nominally "open" system does not provide for effective competition on higher vertically related levels.⁶²

All companies try to use leverage to some extent,⁶³ but Microsoft has a powerful advantage over its competitors. It has used "anticompetitive" licensing practices to acquire a huge installed base and it uses the power of this installed base against competitors in adjacent markets. Microsoft employs multiple linkages and leverage from the different markets (and, in particular, from the gateways) it controls into a single target market, so as to completely outflank and overrun existing competitors in that market.

In the beginning (for our purposes), IBM had a monopoly in computers and the market for computer products was, generally speaking, vertically integrated. (This necessary background is explained in The Economist, supra, at 3-18 (Ex. 14).) How IBM got this monopoly was the subject of much conjecture and years of litigation, but is irrelevant for our purposes. What is relevant is the fact that IBM, in its rush to get out a personal computer, did not leverage its own power from mainframes. Rather, it procured chips from Intel and an operating system from Microsoft ("DOS"), thereby transferring its market power to them as the market for personal computers expanded to displace mainframes and IBM's imprimatur established a standard. In short, IBM empowered Microsoft and Intel to control the architecture for the next generation of computers, and has been playing catch-up ever since. See Charles R. Morris and Charles H. Ferguson, Harv. Bus. Rev., Mar. 1993, supra, at 86, 92 (Ex. 16). See. also Elizabeth Corcoran. Washington Post, Nov. 13, 1994, supra, at H6 (Ex. 44).

Bill Gates, the founder of Microsoft, secured control of the personal computer market by riding IBM's coattails. The success of the IBM PC opened a lucrative market for compatible computers, or "clones." At the time, Microsoft was the sole source for a compatible operating system. Accordingly, Microsoft was able to license the operating system ("DOS") to compatible makers at significantly higher rates than those charged to IBM. Hence, as the Government's Complaint (¶ 19) explains, "Microsoft quickly dominated and gained a monopoly in the market for PC operating systems." 59 Fed. Peg. at 42,847. More precisely,

• DOS would have been worth relatively little had Gates not retained the right to license its use to IBM's rivals. This arrangement -- the source of Gates' wealth and power -- became clearer as IBM set the standard for the burgeoning PC market. By the mid 1980's every rival except Apple computer felt that the only way to compete against IBM was to sell a clone of IBM's PC. Making a clone required, among other things, licensing DOS from Microsoft. Over time DOS became a kind of annuity for Microsoft: buying DOS was the price of admission for entering the PC business.

See G. Pasquel Zachary, Showstopper: Breakneck Race To Create Windows NT and the Next Generation at Microsoft, 27 (1994).

As new technologies overcame the old mainframe market, the market for computer products formed into a number of horizontal markets that are vertically related to each other. Charles R. Morris and Charles H. Ferguson, Harv. Bus. Rev., Mar. 1993, supra, at 8 (Ex. 16). There are many competitors at each level that aggressively compete with each other to develop more powerful products at lower prices. Generally speaking, consumers have benefitted from the formation of horizontal markets. Consumers can put a system together using the best and most cost effective products at each level, even if the products are made by different manufacturers. But by using its installed base in operating systems to "tip" each of these markets in favor of its own products, Microsoft undermines the competitive process. From the initial monopoly bestowed on it by IBM and the huge installed base secured by anticompetitive practices, Microsoft has leveraged and linked a series of powerful monopolies with the intent of forming a new verticality on the market. After establishing several monopolies with enormous leverage potential, the positive feedback from the verticality imposed by Microsoft will in short order eliminate competition on all horizontal layers within the server and online markets, .just as it is eliminating competition in the horizontal layers on the desktop.

1. 1. • The Business Desktop

The Justice Department's Tunney Act filing alleges that Microsoft has monopolized "the market for PC operating systems worldwide" for "almost a decade." 59 Fed. Reg. at 42,850. As noted previously, in 1988 Microsoft had an installed base of approximately 18 million operating system users.⁶⁴ In 1988, Novell (formerly Digital Research, Inc.) entered the X86 operating system market with a competitive product, DR DOS, and it was in response that Microsoft began the "anticompetitive licensing practices" identified by the Government. Microsoft continued these practices through mid-1994, and, as noted previously, it was during this period that Microsoft was able to increase its installed base by more than 100 million users.⁶⁵ As the preceding section explains, it is the size of Microsoft's installed base, rather than merely its market share, that determines the company's true market power. Accordingly. through practices that the Government has identified as "anticompetitive," Microsoft has increased its market power many fold.

Having gained this market power, Microsoft has used it both to maintain its monopoly in operating systems (described in subsection (a) immediately below) and to obtain a monopoly in desktop applications (subsection (b)). The remainder of this section (subsections (c) through (f)) describe how Microsoft has used its market power to engage in other predatory, conduct in the desktop markets.

1. 1. 1. • Effect of the Monopoly on Operating Systems

Microsoft's strategy, which was based at the outset on an installed base created in part through anticompetitive licensing practices, succeeded in monopolizing the desktop OS and threatening desktop applications. Once Microsoft had control of the operating system, which is the key architectural technology for desktop computing, it was able to maintain its share, even with an inferior product. The introduction of DR DOS from Novell showed that Microsoft had failed to keep MS DOS abreast of leading technology.⁶⁶ Yet Novell's compatible offering in the DOS market (DR DOS) stopped selling when Microsoft made it clear that Microsoft would create versions of Windows that were incompatible with DR DOS. It is common for "better" products to fail if a competitor controls the architecture in which the product operates. See Charles R. Morris and Charles H. Ferguson, Harv. Bus. Rev., Mar. 1993, supra, at 89-91 (Ex. 16).

Microsoft was also able to raise prices for its operating system, as its monopoly position continued to solidify and its installed base increased. In the early 1980's, Microsoft licensed MS DOS for $2 - $5 per copy. By 1988, the price was up to $25 to $28. Once Microsoft drove DR DOS out of the operating system market, it was able to double the price it charged, with recent press reports indicating that it is demanding as much as $70 per copy of the forthcoming version of its operating system.⁶⁷

Overall, Microsoft's strategy has been enormously successful in maintaining its monopoly in operating systems while expanding its installed base. Microsoft's share of all desktop operating systems is a staggering 85 %. See supra note 32. Microsoft's share of the operating system market that runs on X86 chips is even larger -- more than 90%. See id.

1. 1. 2. • Effect of the Monopoly on Applications

Having entrenched its operating systems monopoly, Microsoft has aggressively leveraged this monopoly to gain a monopoly in business applications. In 1991, Microsoft's senior vice-president Mike Maples expressly stated the company's intention to monopolize the software applications market:

See Jane Morrissey, Microsoft's Application Unit Seeks Market Dominance, PC Week, Nov. 18, 1991, at 1.

Microsoft used the monopoly revenues from licensing the operating system to fund the development of applications to run on DOS, in competition with software vendors which had no operating system control (for example, Lotus, Borland. and WordPerfect). But because of the relatively open nature of DOS, competitors like Novell could make "compatible" operating systems -- operating systems that would run applications written for Microsoft's MS DOS without modification. Therefore, Microsoft could not exercise sufficient control to give its own applications a strong competitive advantage over the application programs of competitors. The competitors' products were the first developed on DOS and had therefore acquired significant installed bases, as to which powerful network externalities had attached. In order to displace these competitors, Microsoft needed to create a new operating system platform so that its own applications would reach the market on the new platform before its competitors' products.

Microsoft "solved" this problem by (1) developing a new operating environment (Windows) that it totally controlled, (2) targeting a function performed in the application layer that it could either embed in the operating system (for example, the "graphical user interface" or "GUI" feature) or link with the operating system, and (3) using its power over DOS to migrate users to Windows. Microsoft thereby got more control over the OS, added value to the OS it controlled, and forced independent application publishers to rewrite all of their applications twice (once for Windows and a second time for OLE, as described below). The forced migration that Microsoft effected with the GUI and Windows may be depicted as follows:

Figure 7

Microsoft, in effect, added a new layer to the architecture of the desktop, moving the industry, from Figure 2 to Figure 3 above. Controlling architectures is the key to dominating competition. See Charles R. Morris and Charles H. Ferguson, Harv. Bus. Rev., Mar. 1993, supra (Ex. 16).

Microsoft leveraged its control over the operating system to control desktop applications, following a carefully crafted plan that utilized the market power of its installed base. First, Microsoft emulated the application program of the market leader in that application (e. g., Lotus, WordPerfect or Borland), breaking the network externality of the installed base by providing file and keystroke compatibility. Microsoft funded the development, marketing, and below-market pricing of its applications from the profits it reaped on the six-fold increase in the installed base of its operating system. Microsoft's

• stronghold in operating system software . . . financed Microsoft's push into applications software.

Victor F. Zonana, $l4-Million Deal Microsoft Buys Software Competitor, L.A. Times, July 31, 1987, at 4.⁶⁸ For years, Microsoft funded "many versions" of applications programs before they "were good enough to grab substantial market share."⁶⁹ But

• [b]ecause Windows gives Microsoft a "pervasive presence on any desktop that matters, Microsoft can subsidize its loss leaders [in applications] and leverage its desktop heritage".

Barbara Darrow, Developers Brace for Shakeout, Computer Reseller News, Feb. I, 1993 at 28 (quoting Don DePalma, senior industry analyst for Forrester Research). ACCESS, Microsoft's database program, is a case in point. It

• cost a staggering $60 million to develop .... By contrast, the [entire 1992 development] budget at Borland was $50 million. At Lotus, it was $35 million. That's not all. Microsoft also had the money to offer an introductory price of $99 for ACCESS -- less than one-third the retail price for similar packages. Result: Microsoft sold 700,000 copies in just three months. The entire market in 1992 was only 1.2 million units.

Kathy Rebello, et al., Business Week, March 1, I993, supra, at 88.

1. 1. 3. • Unfair Early Access

Moreover, because of Microsoft's installed base in operating systems, it was able to provide an unfair advantage to its applications in a variety of other ways, as well. For example, Microsoft based its own application programs on components m the operating system that it had unique or early access to. Microsoft claimed it was "open," but actually used hidden features and functions to gain a competitive advantage. Brian Livingston. InfoWorld, Nov. 16. 1992, supra, at 98 (Ex. 19). That is, Microsoft provided a proprietary architecture with a supposedly "open" system. See Charles R. Morris and Charles H. Ferguson, Harv. Bus. Rev.. Mar. 1993, supra. The most well-known such example involves Microsoft's "OLE" (object linking and embedding) standard.

Microsoft created interoperability among its own applications, and between its applications and its operating system, by creating a new standard, OLE, which copied functionality from Hewlett-Packard's product New Wave. Stuart J. Johnston, Dangerous Liasons, InfoWorld, April 8, 1991, at 44. With market power on both sides of the interface (i.e., in both the applications and the operating system), Microsoft easily displaced the existing standard in favor of OLE. It embedded OLE functionality into both its operating system and applications, and it heavily marketed this new functionality using profits from its market position in operating systems.⁷⁰

During the very same time period that the Government contends Microsoft was using "anticompetitive licensing tactics" to harm OS competitors, applications competitors repeatedly complained that Microsoft was using its knowledge of new operating system features to give its own applications programs a head start and performance advantage over applications competitors. As stated in Section II of this memorandum, throughout the 1980's and early 1990's Microsoft responded to this criticism by asserting that it had erected a "Chinese Wall" between its operating system developers and applications developers. According to Steve Ballmer, the senior vice-president for Microsoft's system divisions:

• [T]here is a very clean separation between our operating system business and our applications business .... It's like the separation of church and state.

Business Week, Nov. 21, 1983, supra, at 114 (Ex. 2).

In the face of mounting criticism, Microsoft executives adhered to the party line. For example, in 1989, Steve Ballmer again disputed "the charge that his people gave their counterparts in applications previews of their upcoming systems products."⁷¹ Microsoft executives repeatedly told the press that a "Chinese Wall" was in place. See, e.g., Laurie Flynn and Rachel Parker, Infoworld, Aug. 7, 1989, supra, at 43. Indeed,

• Gates insisted that Microsoft kept the playing field level by erecting an imaginary barrier between the company's operating systems group and its applications division.

Hard Drive, supra, at 308. Even into early 1991, Microsoft executives were claiming that the company had an "ISV-independent program" that treated Microsoft applications "the same as any other ISV [independent software vendor]."⁷² Although the FTC began investigating Microsoft in 1990, Microsoft continued to maintain that it had a "Chinese Wall" well into 1991.⁷³

But Microsoft's head start in using OLE in 1991 to the detriment of applications competitors put the lie to such claims. Microsoft incorporated OLE into its Windows operating system and shipped its first completed application incorporating OLE, Excel 3.0, in February. of 1991. at the very same time it was releasing a "beta version" of OLE -- not suitable for commercial distribution -- to ISV's. Indeed, the February 1, 1991, issue of Byte Magazine reports the two events in the same issue.⁷⁴ Microsoft's applications competitors suffered delays of many months as they were forced to rewrite their own applications to make them perform under Windows as well as Microsoft's Excel, which had a head start in using OLE. It was not until many months later that the first third-party implementation of OLE appeared on the market.⁷⁵

Microsoft's unfair advantage obtained from prior knowledge of operating system functionality created a significant head start for its own applications on the new Windows platform. As the prior economic analysis demonstrates, the advantage of being first to market in an "increasing returns industry" is enormous -- it permits a competitor to begin to generate an installed base, reap the benefits of "positive feedback," and otherwise drive its own products to "lock in" before competitors even reach the market. Microsoft used its operating systems information to secure these unfair benefits for its applications.

Confronted with their obvious untruths, Microsoft executives did an abrupt corporate-wide about-face at the end of 1991. Microsoft senior executive Mike Maples stated in December of 1991:

• There is no Chinese Wall. We don't want there to be a Chinese Wall, and I don't think we've ever claimed that there is a Chinese Wall. Microsoft is a single company .... We don't try to pretend that there is a Chinese Wall ....

Stuart J. Johnston, 'No Chinese Wall' at Microsoft, Infoworld, Dec. 30, 1991, at 107 (Ex. 18). And since early 1992, Microsoft has freely and openly given its applications developers an advantage over ISVs. In November of 1992:

• at least half a dozen cases in which Microsoft allegedly withheld information on its DOS or Windows functions from outside developers, for periods ranging from six months to several years. During these periods, Microsoft's own developers appear to have used these functions in applications or utilities that competed with those eventually developed by independent software vendors, according to programmers who have examined the code.

  [I]n each case, the lack of documentation of the functions may have given Microsoft applications a time-to-market lead of six months or more before similar features could be incorporated into competing developers' applications ....

Brian Livingston, InfoWorld, Nov. 16, 1992, supra, at 98 (Ex. 19).

1. 1. 4. • Predatory Bundling

Since dropping all pretense of a "level playing field," Microsoft has increasingly used the power of its operating system installed base to gain advantages over applications competitors. It has attempted to monopolize the market for the development tools (also known as programming languages) used to create applications by predatorially preannouncing its products (as documented in the introduction to this brief) and by bundling versions of its own programming language products into its operating systems so that users will have a powerful disincentive to purchase a competitor's programming language separately.⁷⁶

Microsoft has also conducted a lengthy "campaign" to bundle business software applications into the operating system so that it can "mop up competitors that sell stand-alone applications, resulting in more limited user choice down the road."⁷⁷ Microsoft has steadily increased the price of its operating system to cover its own loss of revenue from the diminished sales of free-standing applications that it bundles into the operating system. Although free- standing applications generally cost more than Microsoft's increases in operating system licensing fees, the unit sales of each application are far fewer than the number of users that upgrade to each new release of the OS -- because of the huge installed base that Microsoft has procured by "anticompetitive practices." Hence, even a modest increase in operating system fees more than offsets Microsoft's loss of revenue from diminished applications sales.

Applications competitors, of course, do not fare as well -- when Microsoft bundles the functionality of their products into the operating system, they lose their only source of revenue. After the competitors go out of business, Microsoft is free to unbundle the applications from the operating system and charge, in the absence of competition, whatever price the market will bear. Microsoft initiated this strategy with the introduction of Windows, by bundling word processing, calculations, communications and "paint" business applications software directly into the operating system.⁷⁸

Microsoft has even bundled technology into its operating system that it misappropriated from its competitors. When Microsoft wanted to add data compression capabilities to DOS, for example, it approached Stac Electronics, developer of the industry's leading data compression software. Microsoft demanded a worldwide license to use Stac's software as part of DOS, but "steadfastly refused . . . to pay Stac any royalty for [its] patented data-compression technology."⁷⁹ When Stac refused Microsoft's demand, Microsoft simply incorporated Stac's intellectual property directly into DOS. Id. Stac brought suit and a federal jury found Microsoft guilty of infringing Stac's data compression patents and awarded Stac $120 million in damages.⁸⁰ Microsoft thereafter settled the case by acquiring a 15 % interest in Stac, and obtained a license to Stac's vital data compression technology for a fraction of the jury's verdict.⁸¹ Because Microsoft's conduct in the Stac case "underscore[s] the sort of allegations that have kept the [Government's antitrust investigation] alive for years," some observers have suggested that the timing of Microsoft's settlement with Stac m late June 1994 was calculated to "remove [Stac president Gary] Clow as a hostile witness in the Justice investigation."⁸²

1. 1. 5. • Predatory Unbundling

Microsoft has also unbundled technology from its operating system in order to render other companies' products uncompetitive. For example, the DOS operating system contained, in version after version, a portion of code known as the "debug kernel." Both Microsoft and competitors like Borland created development tools that used the functionality of the debug kernel in order to run.

With the introduction of Windows 3.1 in April, 1992, Microsoft removed the debug kernel from the operating system and bundled it with its own language application program. If a user wanted to run the competitive Borland program, it had to buy the debug kernel separately from Microsoft, at a price Microsoft set to make the Borland product less competitive. Microsoft even conspicuously advertised the fact that its own product was cheaper than the Borland product because the user had to buy the debug kernel separately from Microsoft. Byte, May 1992, at 159 (Ex. 6). Whatever pro-competitive benefits Microsoft might advance to justify its bundling of new functionality into the operating system, it is difficult to imagine any justification for unbundling operating system technology, other than harming competition.

1. 1. 6. • Other Uses of Leverage

Microsoft further exploited its leverage, both vertically and horizontally. Horizontally, within the desktop applications layer, Microsoft introduced additional applications, touting and exploiting the benefits and advantages of its vertical linkage (to the operating system): for example, word processing ("Word"), database ("Fox Pro" and "Access"), and presentations ("Power Point"). Microsoft also employed horizontal leverage in the applications layer through its marketing practice of bundling a group of applications into a "suite," which is sold at low price points. And, all the while, Microsoft used its profits from its monopoly position in OS for (1) massive marketing to promote the linkage features of the OS. and (2) sustaining a protracted battle with independent applications vendors in a new market that. without the profits from the leveraged market, could not be sustained.⁸³

As noted in the introduction to this brief, Microsoft has been spectacularly successful in leveraging its installed base in the operating system market to dominate the business applications market. In four years, Microsoft "went from an also ran in the business applications market to the industry leader." Inside Telecom, Sept. 26, 1994. Although Microsoft has not yet fulfilled Mike Maples' goal of "I00 percent" market share, it is by far the leading supplier in each individual applications product category. Microsoft Domination, San Jose Mercury News, Dec. 21, 1994, supra, at 1F (Ex. 35). Moreover, suites are the fastest growing category of business applications software and Microsoft accounts for an astounding 85% of all suites sold. See supra note 16.

Microsoft's success in monopolizing business applications is, absent effective Government intervention, only a taste of things to come. Having succeeded in dominating the desktop operating system and applications markets, Microsoft has begun to leverage its installed base to monopolize both the intrabusiness server and on-line systems, as set forth m subsequent sections.

1. 2. • The Intrabusiness Server

Microsoft intends to displace all of the competition on the enterprise server, just as it did on the desktop, by employing multiple linkages and leverage. Its leverage will come from the large installed base of the PC operating system monopoly. Using this base, Microsoft will employ three strategies: (a) vertical linkages similar to those that worked in the desktop markets, (b) horizontal linkages from desktop to intrabusiness server, and (c) horizontal linkages from home-to-business server to intrabusiness server.

Microsoft began the implementation of its strategy by creating a new server OS ("Windows NT") that horizontally leverages from the monopoly position of DOS/Windows in the client market. Microsoft has increasingly placed server functionality into Windows and Windows applications (for example, with the Microsoft products, Access, Fox Pro, and Excel).

Zachary, Showstopper, supra, at 3.

In addition, Microsoft is nakedly leveraging its market power in the desktop operating system market to the enterprise server by requiring software developers who want to use the logo for "Windows 95," the forthcoming version of Microsoft's desktop operating system, to make their desktop application products also run on "Windows NT" (Microsoft's server operating system). See William Brandel, Developing for Next Generation of Windows May Mean Running on NT, Computerworld, November 18, 1994, at 4. There is no technical reason to require an application to run on both Microsoft's desktop and server: indeed, a user would not even expect (nor perhaps even want) a "Windows 95" application program to run on the server. Microsoft's requirement is simply another way of leveraging:

Brian Livingston, Will 'Windows' Compatible Really Mean What It Says?, InfoWorld, November 14, 1994, at 40 (Ex. 20) (quoting Andrew Schulman, Unauthorized Windows 95). Microsoft is using its operating system power to force independent application vendors to establish the linkage between the desktop and the server that Microsoft has been trying to establish through its own products. In affect, Microsoft is using independent software vendors to establish Microsoft's power in servers.

Microsoft also enhances its power in the server applications layer by horizontally bundling these products into a suite (the "BackOffice") in the same way Microsoft bundled desktop applications into a suite. Just as with the desktop applications, there is also vertical leverage to enforce the horizontal bundle by making all server applications OLE-enabled. See Stuart J. Johnston and Ed Scannell, Computerworld, supra, Oct. 10, 1994, at 4 (Ex. 7); J. William Semich, Datamation, Aug. 1, 1994, supra, at 4144 (Ex. 10).

1. 3. • The Home-to-Business Market (Server and Client)

Increasingly, business will need to communicate with personal computers in homes in order to sell products or services and in order to provide information, for work or other purposes. Obviously, businesses that exploit this channel will have a strong advantage over competitors that do not, with the result that all businesses will seek entry. This market is currently known as "online services." There are three principal competitors in this market -- America Online, CompuServe and Prodigy.

Control of the home-to-business market by a single company would produce an enormous windfall. First, of course, the monopoly would be able to extract a toll for a large percentage of consumer financial and product transactions. More strategically, a company that controlled the home-to-business market could leverage that control back to the intrabusiness, or enterprise, server market. Control of both sides of the server market, intrabusiness (enterprise) and home-to-business, would place enormous power (financial services, information, education, etc.) in the hands of a single company. Microsoft has this power within its grasp. Microsoft is pursuing its policy of targeting, linking and leverage from the operating system installed base to seize control of the architecture of the home-to-business market, just as Microsoft gained domination of the desktop.

On November 14, 1994, Microsoft announced its own online service known as "Marvel" or the "Microsoft Network." Microsoft will use Windows NT as the home-to-business server for the Network. Adam Gaffin & Peggy Watt, Microsoft, Lotus Baffle Shifting to On- Line Services, Network World, Nov. 21, 1994, at 1. More importantly, Microsoft will use the market power from its' installed base in operating systems in a number of ways to displace existing on-line competitors and dominate the home-to-business market.

1. 1. 1. • Predatory Bundling

First, Microsoft intends to leverage its installed base in operating systems to give its own on-line service an unfair advantage over existing competitors. Microsoft has already announced that the next upgrade of its PC operating system, Windows 95 (due out later this year), will have a connection to the Microsoft Network already bundled in. According to Bill Gates, "We'll give you access to [the Microsoft Network] with Windows 95... If (the software notices you have a modem, it will ask you if you want to register."⁸⁴

This tactic will instantly displace existing on-line competition. Windows 95 will be pre-installed on virtually every PC sold in the United States in the coming year⁸⁵ and approximately 20 million copies will be in use within a year of its release. Amy Bernstein. Microsoft Goes Online, U.S. News & World Report, Nov. 21, 1994, at 84. This "potent plan for spreading Marvel" will dwarf the competition. Id. America On-Line, by comparison, has an installed base of 1.25 million subscribers. Elizabeth Corcoran, Washington Post, Nov. 12, 1994, supra, at H6.

Industry analysts and commentators have repeatedly raised concerns that Microsoft's bundling of its own on-line service "tilts the playing field in its direction," likening Microsoft's bundling practice to the utility company selling appliances or the local phone company automatically connecting the user up with AT&T's long distances:⁸⁶

• In essence, OEMs will be forced to distribute MSN [The Microsoft Network] if they want to access Windows 95 -- even if that distribution is to the OEM's detriment.

Jesse Bent, Microsoft's On-Line Rivals Could End Up In `Cyberia', PC Week, Dec. 12, 1994, at 120 (Ex. 30). Microsoft's conduct is a textbook example of an attempt to use market power in one market (operating systems) to "tip" a competitive adjacent market (online systems).

1. 1. 2. • Unfair Use of Information

Microsoft is also using its power over the operating system installed base to dominate the content of the home market -- CD ROMs -- the same way it used leverage from the operating system installed base to dominate business applications. For example, as a condition to obtaining information about how to run on the multimedia portions of Microsoft's operating system, independent CD ROM developers were required to fill out a form, designated "Microsoft Confidential." In other words, in order to obtain necessary operating system information, the form required Microsoft's CD ROM competitors to disclose to Microsoft confidential business information necessary to make successful CD ROM products. This form is a remarkably glaring example of the open exercise of market power. It required, inter alia, the following disclosures:

Provide proposed product areas.

Current key software products (in order of market share and importance to your company).

Who is the target audience for your products?

What is the price of your products?

What is your supply date for retail distribution?
What competition do you perceive for this product?

How will you differentiate this product from its competition?

How is this project funded?

(The "Microsoft Confidential" form is found in the Appendix as Ex. 22.) Armed with all of this confidential information about its competitors plans and products, Microsoft has successfully entered the CD ROM business itself, and is "churning out about one new CD ROM title per week." Washington Post, Nov. 13, 1994, supra, at H6 (Ex. 44).

1. 1. 3. • Unfair Head Start

Microsoft will also ensure domination of the content of on-line services by using OLE-based tools as the standard for business developers and users to create object-oriented documents that can be transmitted over the Microsoft Network. Mary Jo Foley, Microsoft Lays Foundation For On-Line Network, PC week, Nov. 14, 1994, at 1; Doug Barney, Microsoft to Announce New On-Line Service at Comdex, InfoWorld, Oct. 24, 1994, at I, 140. According to a PC Week article, the Microsoft network employs OLE technology and uses the "standard Microsoft Exchange E-mail client included with Windows 95.... "In short, "Microsoft Network's on-line services are well-integrated into the Windows 95 user interface." Eamonn Sullivan & Matt Kramer, Microsoft Marvel Beta Leverages WIN 95 Desktop, PC Week, Nov. 7, 1994, at 169 (Ex. 28).

And, as if Microsoft's use of leverage to dominate the home and on-line markets is not sufficient, Microsoft announced on October 13, 1994⁸⁷ its intention to buy Intuit. Inc., paying a 100% premium to market. See supra note 36. Intuit publishes the personal finance and tax planning software programs that dominate their respective markets. Intuit's product controls 80-85% of the personal finance markets.⁸⁸

Personal financial software is generally regarded as the "killer app[lication] of the 90's" for the home computing market,⁸⁹ Personal financial software has broad consumer appeal in that everyone has a bank account. It requires the integration of several sources of data including bank accounts, brokerage accounts, and credit information. Because of Intuit's commercial success, there is a strong network externality ("lock in") attached to a user's viewing his personal financial information through the Intuit user interface. Accordingly, Intuit provides tremendous leverage into the home banking market.

The Intuit acquisition is currently under Justice Department scrutiny. If the deal is consummated, Microsoft can be expected to leverage Intuit's installed base to further lock in its own products. For example, Microsoft will bundle Intuit's products with its next release of the operating system to increase the number of users who will upgrade to Windows 95.⁹⁰ Microsoft can also provide an enormous market edge to its own on-line service by making Intuit available exclusively (as among on-line services) on the Microsoft Network. See Michael J. Miller, The World According to Microsoft, PC Magazine, Jan. 24, 1995, at 80 (Ex. 25).

Domination of home banking and personal finance provides the optimum platform from which to dominate other on-line services, including, for example, shop-at-home. Businesses that want to provide financial information to Intuit users, or who want to provide other on-line services, will want to choose server software for interacting with the Microsoft Network. Microsoft will be able to use all of its vertical integration skills developed in the desktop and enterprise server marketplace to ensure that businesses choose Microsoft home-to- business server software.

Based on the leverage potential from its operating system installed base, Microsoft has been able to consummate deals that will ensure that Microsoft Network dominates the market. For example, on November 8, 1994, Microsoft and VISA (the credit card company) announced the provision of a standard and secure method "for executing electronic bankcard transactions across global public and private networks.' Visa News Release, Nov. 8, 1994 (Ex. 39). In the question and answer session following the press release, the VISA spokesperson said that the driving force in VISA's decision to do the deal with Microsoft was the fact that Microsoft had an installed base of 60 million copies of Windows. The significance of Visa's agreement with Microsoft is not lost on industry observers. See, e.g., Elizabeth Corcoran, Washington Post, Nov. 12, 1994, supra, at H6. Nor is it likely to be the last such agreement: the Post reported, for example, that "four telecommunications companies are expected to announce on Monday [November 14, 1994] that they are working with Microsoft to make dialing into Marvel a local call for many subscribers." Id. And, on December 21, 1994, Microsoft announced that Tele-Communications, Inc. purchased a 20% stake in the Microsoft Network for $125 million. The deal implies a value of $625 million for an on-line service that doesn't exist yet .... "Jim Carlton & G. Pascal Zachary, Microsoft Sells A 20% Interest In Planned Unit, Wall St. J., Dec. 22, 1994. Once again, Microsoft is controlling the architecture and using a nominally open standard.

If Microsoft is successful in establishing the standard for the home-to-business market, it will be able to leverage into the enterprise server market both from the desktop, which it already controls, and the home market. Once a business decides that it should use the Microsoft server to communicate with customers, there is no point in having a different. probably incompatible, server for intrabusiness needs. After all, the operating system for the server side of Microsoft's home-to-business server is Windows NT. Why have a different business server operating system? This connection between the home server and the business server is clearly in Microsoft's contemplation because Microsoft has already announced that Marvel (the Microsoft network) will connect directly to a company's server. Doug Barney, Microsoft to Announce New On-Line Service at Comdex, InfoWorld, Oct. 24, 1994, supra, at I.

The inevitable result of Microsoft's monopoly leverage will be to transform Microsoft into a "middleman" or rent collector for every transaction processed in an all- encompassing information economy. Whether writing a letter, placing an order, or paying a bill, every consumer and business connected to the information highway will pay Microsoft's toll. As noted in Fortune, "[t]his isn't just a gleam in Bill Gates' eye -- [by purchasing Intuit, entering a joint venture with Visa, and bundling the Microsoft Network] -- its already starting to come together, and in Microsoft's typically orchestrated fashion.⁹¹

MICROSOFT'S NETWORK-WIDE MONOPOLY

It is readily apparent that Microsoft's strategy of targeting, linking and leveraging from the desktop operating system has been successful in seizing control of the business desktop. It is also apparent that Microsoft is leveraging from the business desktop to the business server and is vertically integrating within the business server so as to seize control of the critical server operating system gateway. The Intuit acquisition is intended to control the gateway on the home computer and leverage toward the home-to-business market.

Application of "increasing returns" economic analysis would reasonably predict that, given the present situation, Microsoft will succeed in monopolizing the entire information infrastructure (just as it has monopolized the desktop) and that the monopoly will remain in place for a very long period of time.⁹² Indeed, the monopoly on the enterprise and home-to-business server markets is likely to be so vast that Microsoft will be able to extract monopoly rents on not only financial transactions, but also the transmission of information and data.

• Some fear that as the digital future of the information superhighway emerges, an unchallenged Microsoft and Intel will wind up in total, undisputed control of the technology upon which the country's citizens and economy will depend . . . "Increasingly, I'm believing it's all over, and we're going to be locked into Microsoft and Intel forever," said Dataquest analyst Kimball Brown.

Rory J. O'Connor, Microsoft, Intel Set to Define Technology, San Jose Mercury News, Nov. 13, 1994, at 1-A. (Ex. 34).

Notwithstanding the Government's conclusion that Microsoft has increased its installed base in operating systems six-fold using "anticompetitive practices," and ample evidence that Microsoft has leveraged that installed base to attempt to monopolize business applications (as well as other markets), the Government's Tunney Act filing does not require divestiture of any part of its operating system installed base, nor does it prevent Microsoft from using that installed base to monopolize other markets, including business applications. The Government has articulated no economic rationale to justify its failure to act in the face of such clear evidence of anti-competitive intent and effect. These Amici can identify four possible economic justifications for the Government's inaction, but none of the four is persuasive.

1. • Leverage of the Installed Base by Competitors

Although the Government has not articulated an economic rationale for its position, the Justice Department may have concluded that a monopoly of the X86 operating system market by Microsoft is inevitable -- either because MS DOS is already locked-in or because an "increasing returns" market will cohere around a standard in any case. Following this approach, the Government may have concluded that the best hope for competition in the operating system market is through an operating system program compatible with MS DOS, but made by a Microsoft competitor. Arguably, a vendor of such a program could tap into Microsoft's huge installed base and attempt to displace Microsoft by "migrating" users to subsequent versions of the competitor's operating system.

If such was ever in the Government's contemplation, events since the announcement of the settlement between the Justice Department and Microsoft have shown that such a scenario is unrealistic. Novell has withdrawn its MS DOS compatible operating system from the market entirely. See, supra note 14. And Microsoft's market is so strong that IBM selected Microsoft's MS DOS program for pre-installation on a new line of IBM personal computers, instead of IBM's own PC-DOS (compatible) program -- notwithstanding the fact that IBM's product is technologically superior to MS DOS and is less expensive.⁹³

IBM's technologically advanced OS/2 is faring no better. OS/2 is capable of executing both DOS and Windows 3.1 applications, and according to Microsoft executive Steve Ballmer, IBM is "offering computer makers OS/2 for free and may be even paying some to take it."⁹⁴ However, Microsoft's market power has resulted in IBM getting few if any takers, even on these terms. As one potential customer, a computer manufacturer, stated:

• Microsoft can kill us ..... I worry more about my dealings with Microsoft than I do about my competitors.⁹⁵

2. • Alliances

Alternatively, the Government may have concluded that other operating system competitors might combine with application developers in alliance-type combinations to prevent Microsoft from extracting monopoly rents from the business desktop. But alliances among companies rarely work in the best of circumstances -- i.e., in more conventional markets. Here, the alliances would have to produce or blend complex software technologies in order to make a competitive offering equally useful and reliable to that marketed by a single vertically integrated competitor, which is better able to guarantee seamless integration.⁹⁶ Similarly, from the economic perspective, the possibilities of real competition from an alliance-based product line are highly remote, at best. Microsoft's installed base and share of the applications market is so large that its products are "locked-in" and true competition can be restored only through truly massive forces or structural relief. See, e.g., W. Brian Arthur, Increasing Returns and Path Dependence in the Economy 2, 10-11 (1994).

Most importantly, although there are companies that make operating systems that run on different chips, no Microsoft competitor or group of competitors controls the operating system gateway to the network in the way that Microsoft does. Control of the "human interface" gateway on the home computer through the acquisition of Intuit will only heighten Microsoft's control throughout the market. In short, the prospects of an alliance to compete effectively with Microsoft, in the current market where the gateways are controlled by Microsoft, are extremely remote. Competitors would have to produce a competing information infrastructure through a different paradigm (e.g., cable television), something that is years, if not decades, away. Microsoft is, moreover, already committing substantial resources -- reportedly 500 employees by next June -- in anticipation of this paradigm shift. See Elizabeth Corcoran, Washington Post, Nov. 13, 1994, supra, at H6 (Ex. 44). It therefore is clearly preparing now to be in a position to control this new paradigm as well.

3. • "Tiered" Monopoly

Third, the Justice Department might have concluded that, although Microsoft has achieved a monopoly in the operating system market, there is no need for governmental intervention because Microsoft would prefer competition in business and home applications software. In other words, the Government might argue that Microsoft has no economic incentive to monopolize the applications market intentionally and has acquired its dominant position in the market only because of superior products. According to this approach, although Microsoft has a monopoly on X86 operating systems, it would actually prefer that the applications (and development tools) market be fully competitive in order to maximize monopoly profits from the operating system market. A schematic representative of the "desktop," Figure 3, is reproduced below for reference:

Figure 8

This type of economic thinking would suggest that if Microsoft truly had a monopoly at the second level (operating systems), it would prefer competition at all higher levels so as to maximize its ability to extract monopoly profits through the operating system level. And, according to this economic argument, there would be no point in Microsoft expending the resources to monopolize applications (level 5), since it would derive the same benefit by monopolizing the operating system (level 2).

Indeed, according to this approach, because of the presence of demand side economies of scale, there would be a need for Microsoft to control the X86 operating system (level 2). There is a network externality that must be solved by a single firm with control of both level 2 and all of the levels above it (3-5). All other factors being equal, according to this argument, consumers would be better off with the greatest possible variety of level 5 competition and the greatest possible adoption of one operating system standard.⁹⁷ Hence, if Microsoft controls the operating system, it would have an incentive to price it low because it could extract the profits through the applications (level 5). (Or, alternatively, Microsoft might price the applications low and take the profits out through the operating system.) Indeed, Microsoft might be willing to price below cost.

On the other hand, according to this economic approach, if a Microsoft competitor gained control of applications, Microsoft and the competitor would fight over the division of profits. This would be wasteful, would lead to higher total costs for the system because of "double marginalization" and would not lead to as great adoption of the overall system. Given that Microsoft controls the X86 operating system, so the argument would go, its profits would be maximized if the market for applications were made as large as possible. Hence, it would follow that Microsoft would want to control applications to make this market as large as possible and would do this by pricing applications at a low level, and by making the inter-connection between its applications and operating system as efficient as possible.

This economic approach is unpersuasive for three reasons. First, although Microsoft monopolizes the market for operating systems that run on the X86 chip, there are competitive operating systems that run on other chips -- Apple and UNIX, for example. These competitive operating systems, like the Microsoft operating system, run business applications. Hence, so long as these competitive operating systems exist, Microsoft can extract "monopoly rents" by monopolizing a layer above operating systems -- business applications.

Second, as the Government's complaint in this case points out, there must be "a variety of high quality applications" that run on an operating system if that operating system is to be successful. 59 Fed. Peg. at 42,847 (Complaint ¶¶16-18). Accordingly, control of applications enables Microsoft to maintain and increase barriers to entry in the operating system market, thereby solidifying and maintaining Microsoft's operating system monopoly.

Finally, control of the application layer enables Microsoft to price discriminate more effectively, thereby maximizing its monopoly returns. For example, because Microsoft also monopolizes business applications, it has the ability to selectively bundle some word processing functionality into operating systems, while at the same time offering a higher priced, more fully functional word processing program to users who need greater functionality. This enables Microsoft to extract greater revenues than would be possible merely by uniform operating system prices -- i.e., if Microsoft only monopolized operating systems, but not applications.

In short, Microsoft has ample economic incentive to monopolize business applications. To the extent Microsoft is concerned at all about actual or potential competition for operating systems, gaining control of applications will ensure overall control of the desktop. regardless of what might transpire in the future with respect to operating systems.

A complete comparison of consumer welfare in a world with uniform dominant- firm pricing in operating systems and competition in applications on the one hand, with monopoly price discrimination on the desktop (operating system and application together), on the other hand, is beyond the scope, of this Memorandum. However, economic theory would strongly suggest that with respect to pricing, competition in applications, coupled with imperfect competition in operating systems - or at least the presence of potential competition in operating systems -- is preferable to monopoly of the entire desktop. Moreover, in terms of technology, it is considerably more likely that the best technology will emerge in applications if there is open competition for the technology, rather than if it is dominated by the firm that monopolizes operating systems. That is especially true if the reason that Microsoft is able to monopolize applications is because it can leverage its operating systems monopoly and not because of any superiority of its technology.

4. • Efficiencies of Integration

Finally, the Government might justify its failure to act on the belief that the benefits Microsoft is providing by vertical and horizontal integration outweigh any anti- competitive effects. Microsoft will point out that it seamlessly integrates new technologies into new markets, and it will argue that unless it is permitted to link and leverage, these markets will not be opened in a way meaningful for consumers. It will further argue that if markets are opened by less efficient alliances, the services are bound to cost more because Microsoft competitors will not enjoy the efficiency benefits of integration. Indeed, according to this argument, allowing Microsoft to leverage Windows from one market to the other amortizes the research and development costs over a broader base of potential customers, with the result that Microsoft can charge less for the product in the first instance.

Furthermore, Microsoft presumably will argue that because these markets and technologies exhibit increasing returns, they will gravitate toward a standard (i.e., a monopoly) anyway. According to this argument, it would be economically wasteful to require two networks that do the same thing. And, if there is only going to be one standard, that standard should be chosen by the market, as opposed to by Government intervention.

There are two important responses to this argument. First, software is not similar to many conventional products in an important way. With software it is possible to achieve virtually all of the benefits of integration without excluding competitors. There is no reason why an application developed by an ISV cannot work just as well with the operating system as a Microsoft application, provided Microsoft provides necessary information to application competitors on a timely and complete basis.

Second, while there are benefits to vertical and horizontal integration that Microsoft will point out, there are also very substantial costs. The enterprise server market, for example, is currently organized into a number of horizontal layers, each of which is characterized by strong competition. Generally speaking, consumers prefer this horizontal competition. See, e.g., The Economist, Feb. 27-Mar, 5, 1993, supra, at 11 (Ex. 14). Microsoft is attempting to impose a verticality on the enterprise market so that it can extract monopoly rents.

Benefits of vertical integration, as opposed to horizontal competition at each layer. both on the desktop and the server, should be evaluated on the basis of product quality and incentive to innovate, as well as product cost. It is clear that vertical integration will allow Microsoft to displace even superior technologies. As PC Magazine recently observed:

Michael J. Miller, PC Magazine, Jan. 24, 1995, supra, at 79-80 (Ex. 25). Similarly, if Microsoft controls the operating system gateway layer, its vertical integration will permit the displacement of superior products at the applications (and development tools) layer merely because of the vertical integration. The displacement of superior products is clearly a cost that should be evaluated, offsetting Microsoft's claim that its products would be lower-priced to the consumer.⁹⁸

Moreover, once Microsoft achieves dominance in a market, it has little incentive to innovate.⁹⁹ So the negative effects of vertical integration include both the displacement of superior products, as well as the diminution of the incentive to advance technology that has become a standard. The latter cost should be evaluated as well.

Nor is it altogether clear that vertical integration will necessarily produce efficiencies (that translate into lower prices) over, say, horizontal competition at each layer. There is not yet empirical research on point, but there is certainly theoretical research suggesting that there are benefits to horizontal competition in the vertical layers.¹⁰⁰ Hence, while there is theoretical literature that documents the efficiency of the horizontal competition model, the real challenge is maintaining the horizontal model in the world. Increasing return economics indicates that there is no reason to believe that the market, as currently structured, will choose the "best" product at a particular level. Rather. there is every reason to believe that Microsoft. through leverage from control of the operating system, will be able to impose verticality, with its associated costs -- notwithstanding the fact that users appear to desire the benefits of horizontal competition. See, e.g., The Economist, Feb. 27-Mar. 5, 1993, supra (Ex. 14). In short, Government intervention is necessary merely to provide a sufficiently level playing field for the horizontal model to have a reasonable chance of succeeding.

ANTITRUST ENFORCEMENT

This section of the brief identifies the deficiencies of the proposed Final Judgment and compares the relief sought by the Government in this case to the relief sought by the Government in comparable situations involving pharmaceutical, computer and telecommunications monopolies. Finally, the section analyzes the relevant case law that would support similar relief in this case, particularly a preclusion on the use of leverage from an installed base that was procured by "anticompetitive practices."

1. • Deficiencies of the Proposed Judgment

Manifestly, the proposed judgment has failed to achieve its stated purposes. Instead of saving consumers money and providing them with greater operating system choices as the Attorney General promised, the settlement has permitted Microsoft to run yet another competitor out of the operating systems market (Novell) and raise its own prices to resellers. From an economic perspective, this was to be expected. The relief proposed by the Government will neither maintain nor restore competition in the operating systems market. More ominously. the settlement clears the way for Microsoft to use its unfairly acquired installed base to run competitors out of other software and networking markets, as well.

According to the Government's complaint, Microsoft used anticompetitive licensing practices from at least 1988 to 1994. As noted earlier, during that period, Microsoft maintained its greater than 90 % share of the X86 operating system market.¹⁰¹ thereby increasing its installed base six-fold.¹⁰² Contrary to the assertions of the Assistant Attorney General, the relief proposed by the Government, a cessation of further anticompetitive practices, will not restore competition to the X86 operating system market because of the "network effects" present in the market.

Because Microsoft now has a huge installed base and an overwhelming market share of X86 chip operating systems, thousands of applications have been written for the Microsoft operating system. Microsoft products, in economic jargon, are "locked in." New purchasers of computers with X86 chips have every incentive to demand Microsoft operating systems -- and no incentive to demand the operating systems of its competitors. Given the huge installed base, OEM's will therefore preinstall the Microsoft operating system in order to meet consumer demand -- whether Microsoft continues to pursue "per processor" licenses or not.

This conclusion is demonstrable from the economic literature cited in earlier sections. It is also obvious to the journalists, analysts and commentators who follow the computer industry. For example, following announcement of the settlement, PC Week wrote:

• According to computer manufacturers, industry analysts and end users, the outlook is grim for Novell's DOS and IBM's PC-DOS and OS/2. They say there is not much motivation for PC manufacturers to pre-install a competing product, since Windows has millions of users and thousands of software applications.

See Jeff Bertolucci, Microsoft Settles: Business As Usual, PC World, Oct. 1994, at 72 (Ex. 31).¹⁰³ Furthermore, Microsoft has adopted new marketing incentives that violate the spirit if not the letter of the consent decree by rewarding OEMs for activities designed to prevent them from doing business with competing operating system vendors. Don Clark & Laurie Hays, Wall St. J., Dec. 12, 1994, supra, at B6. In short, Microsoft's new practices achieve substantially the same effect as those banned by the Judgment.

More importantly, Microsoft remains free to leverage its installed base -- apparently with the Government's blessing -- to put competition out of business in scores of new markets: business applications, entertainment software, personal finance software, on-line systems, server technologies, etc. This key issue is simply not mentioned in the Government's Tunney Act filings, but, as with "lock-in," the significance of the issue is not lost on the industry:

• The settlement did not specifically address what many competing companies consider the antitrust issue. Microsoft, they say, has used its control of DOS and Windows to extend its hold on the software sector.

See David Einstein, Microsoft Unscathed by Settlement, S.F. Chronicle, July 18, 1994, at A1 (Ex. 32).¹⁰⁴ As explained in Section V.C., supra, Microsoft's use of leverage against application competitors damages competition in the operating systems market, the very market the Government purports to address.

The pernicious use of leverage is well known to the Justice Department. Decrees sought by the Antitrust Division in comparable circumstances over the past forty years have prohibited leveraging of monopoly power to dominate related markets.

2. • Comparable Consent Decrees

It is hardly aberrational for the Department of Justice to settle monopolization cases in high technology industries by securing consent judgments that prohibit the use of leverage from a monopolized market to a market in which competition is present. Some of the largest monopolization cases in history were settled on such a basis.

1. 1. • Parke, Davis Decree (Pharmaceuticals)

The decree entered in United States v. Parke, Davis and Co. and Eli Lilly and Co., 1951 Trade Cas. (CCH) <20> 62,914 (E.D. Mich. 1951), prevented Parke, Davis and Eli Lilly from using their market power in the primary market for pharmaceuticals to exert leverage into the secondary market for gelatin capsules (used to contain individual doses of particular drugs). The decree did not foreclose the defendants from competing in the capsule market, but it imposed severe restrictions designed to ensure competition:

Mandatory Licensing of Patents Pertaining to Secondary Market: Defendants were required to grant to "any applicant" (except the other defendant) royalty-free, unrestricted licenses under all Defendants' existing capsule-related patents. Defendants also were required to grant licenses to all of their future capsule-related patents in return for a "reasonable and non-discriminatory royalty." (An equivalent Microsoft decree would require, at minimum, that Microsoft grant royalty-free licenses on all its existing application and server software patents.)

Publication of Documentation to Enable Competition in Secondary Market: Defendants were required for five years to provide to all applicants "a written manual.., describing the methods, processes, materials and equipment used by [Defendants]" in the commercial manufacture of capsules. (A provision that would have the same effect in the Microsoft decree would require, at minimum, that Microsoft immediately provide all competitors or potential competitors all operating systems documentation and specifications necessary to create a well-behaved application program. Going forward, Microsoft would have to provide the information necessary to place each of its competitors in the applications program market on an equal footing with Microsoft itself.)

This decree remained in effect until 1987. See United States v. Parke, Davis and Co. and Eli Lilly and Co., 1987-2 Trade Cas. (CCH) . 67,834 (E.D. Mich. 1987).

1. 2. • International Business Machines Corp. (Computers)

In 1956, the Justice Department settled its monopolization case against IBM with the entry of a comprehensive decree, United States v. International Business Machines Corp., 1956 Trade Cas. (CCH) . 68,245 (S.D.N.Y. 1956). That decree still remains in effect.

The IBM decree prevents IBM from utilizing its power in a primary market (the market for "tabulating systems" and "electronic data processing systems") to create a monopoly in secondary markets (the markets for service on IBM machines). Unlike the Microsoft settlement, however, the IBM decree makes a comprehensive effort to prevent leveraging of the primary market monopoly. Rather than prohibiting a small number of specific practices (e.g., per-processor licensing), the IBM decree fundamentally restructured IBM's method of operation in the primary market to eliminate leverage opportunities.

A similar decree against Microsoft would have included (at minimum) provisions requiring that Microsoft: (1) train its customers and competitors in the use and structure of Windows, (2) disclose to all developers, customers and competitors the same details about Windows that it discloses to its own employees and at the same time, (3) make public Microsoft technical documentation and tools used in Windows development, and (4) create a separate corporation for developing application programs, with a true "Chinese Wall" between the applications and operating system development personnel.

1. 3. • American Telephone and Telegraph (Telecommunications)

In January of 1982, the Department of Justice filed a Final Judgment breaking up the AT&T monopoly. In its response to comments on the proposed final judgment, the Government explained that it sought broad relief to prevent the type of leverage that Microsoft is currently employing:

47 Fed. Reg. 23,320, 23,335 (1982). Microsoft is not a regulated monopolist, but its monopoly in operating systems is no less thorough and its use of leverage to dominate related markets no less pervasive. Yet according to newspaper interviews given by the Assistant Attorney General following announcement of the settlement with Microsoft, the Justice Department "never considered" breaking up Microsoft. Viveca Novak, Antitrust's Bingaman Talks Tough in Microsoft Case, Wall St. J., July 19, 1994, at B5.

3. • Case Law

Had the Justice Department sought to prevent Microsoft from leveraging its installed base of "locked-in" operating system users, its position would have found support in the case law. Cases in which leveraging claims have been denied involve factual situations in which the plaintiff conceded that monopolization of the target market was impossible, even with the leveraging. See, e.g., Alaska Airlines, Inc. v. United Airlines, Inc., 94.8 F.2d 536, 54.6 (9th Cir. 1991), cert. denied, 112 S. Ct. 1603 (1992).

This is not such a case. Here, both Microsoft and the Government concede that Microsoft has a monopoly in the operating system market and that Microsoft used "anticompetitive practices" to increase its installed base in operating systems six-fold. Microsoft then clearly expressed its intention to monopolize the business application market and thereafter succeeded by leveraging. Now, Microsoft's executives have clearly expressed their intention to monopolize every "specific application of corporate information systems." Brent Schendler, Fortune, Jan. 16, 1995, supra, at 40. Microsoft's tactics, coupled with the economics of the markets at issue, would lead inexorably to the conclusion that Microsoft will succeed.

A number of courts, including the Supreme Court, have evaluated conduct in one market based upon conditions in an adjacent, related market. Relevant decisions have reflected increasing returns-type analyses. For example, in Eastman Kodak Co. v. Image Technical Services, Inc., 112 S. Ct. 2072 (1992), the Supreme Court held that factual issues regarding consumer "lock-in" in the after-market for replacement parts constituted a proper basis on which to deny motions for summary judgment in a tie-in case. Similarly, a plaintiff's use of leverage in lock-in situations has frequently been cited in the lower courts as a principal basis for the denial of summary judgment motions in both tie-in and monopolization situations.¹⁰⁵

One good example of such thinking is Grappone, Inc. v. Subaru of New England. Inc.. 858 F.2d 792 (1st Cir. 1988). There the First Circuit (Breyer, C. J.) provided what it referred to as a more "refined analysis" for tie-in situations. This analysis begins to consider the anti-competitive consequences of actions that require competitors to enter the market on two levels (rather than a single level) of business. Id. at 795-96.

PROPOSED PROCEDURES UNDER SECTION 16(f)

Reflecting its emphasis on the importance of court review of decrees agreed to by the Justice Department, Congress in 15 U.S.C. . 16(f) has expressly authorized a wide variety of procedures that the Court may use in making its determination regarding the public interest. These procedures include, inter alia, taking the testimony of Government officials or experts, or other expert witnesses (. 16(f)(1)); appointing a special master or court expert (. 16(f)(2)); examining documentary materials (. 16(f)(3)); or "taking such other action in the public interest as the court may deem appropriate" (. 16(f)(5)).

In this action, some information is relatively well-documented in the public record, and hence is less pressing significance to the Court's ability to engage in a meaningful public interest analysis. By way of comparison, in United States v. Yoder, 1989-2 Trade Cas. (CCH) . 68.723, at 61,797 (N.D. Ohio 1986), the Department provided the court with an affidavit identifying the number of competitors, distributors and customers in the industry, whom it had contacted about a proposed modification to a consent decree, and described the responses and concerns of those contacted. See id. at 61,797 n. 10. Here, the Department has simply asserted orally that "by and large I think we got positive feedback" from competitors and customers, then adding (in response to a comment by the Court) "there were clearly some people who wished that we had done more." Tr. of Status Call, Sept. 29, 1994., at 13:16-22. These observations certainly do not give the Court the full flavor of industry concerns, but critical reports in the media amply document the true reaction in the industry to the proposed decree.¹⁰⁶ It is, therefore unnecessary to further burden the Court with affidavits or the testimony from those in the industry, regarding these concerns.

Similarly, the nature of the allegations regarding Microsoft's conduct are well- established. Media reports, publications such as Hard Drive, this brief, and the Government's own submissions all document what the alleged illegal conduct is claimed to be: undocumented calls; early disclosure of operating systems information to Microsoft's own applications engineers; predatory preannouncements; predatory bundling and unbundling of operations and applications functionality; restrictive licensing practices; and the use of subsidized pricing to leverage into the applications market using monopoly profits from operating systems. See supra text at notes 69-70. It would therefore appear unnecessary to hold hearings in which various independent software vendors, OEM manufacturers, and other industry participants recount particular instances of such alleged conduct.

Instead, these amici submit that what is missing from the record before the Court are two categories of information, neither of which should require unduly protracted hearings. but which together should provide the Court with a sufficient record to make a determination under Section 16(e). First, in the course of its investigation, the Government has reviewed large quantities of documents from Microsoft, and these amici believe that a very. small group of these documents have been identified by the Government as "key" documents. These documents largely should answer questions regarding Microsoft's intent and use of various illegal practices. They should be turned over to the Court for its review.

Second, the Government should be required to submit affidavits from its economic experts that set forth in detail what those experts anticipate the operating systems and applications software markets will look like in five years, assuming that the present proposed decree were implemented. Such a submission should indicate whether, under the present decree, the Government's experts anticipate that competition will have been restored in the operating systems market by that time. If the Government's experts believe that competition is not likely to have returned to the market by that time, they should be required to indicate what effect different alternative proposals might have on restoring competition to the market. And, if they believe under "increasing returns" theory that it is simply too late to restore competition -- that the operating systems market "runs to scale," and having been permitted to establish dominance through its illegal practices, that Microsoft cannot now practically be unseated -- the Government should be required to indicate what alternatives it has considered to minimize adverse consumer consequences resulting from this monopoly.

These amici submit that the affidavits from the Government's economists also should address the extent to which they anticipate that Microsoft will have been able to leverage its operating systems monopoly into secondary software markets. Because Microsoft's installed base monopoly (and the resulting monopoly profits) were illegally acquired, the Government's economists should explain why it is unnecessary from an economic point of view to implement provisions such as those present in the IBM and Eli Lilly consent decrees. This analysis would include, for example, the effect of alternatives such as prohibiting Microsoft from acquiring stock in companies that make or sell application programs (Eli Lilly): spinning off its applications division into a separate subsidiary, and enjoining it from giving any benefit to the subsidiary, that is not also provided to third-party applications providers (IBM); and making public Windows technical documentation and tools used in Windows development (IBM). In the event that such alternatives were not viewed as sufficient to ensure a "level playing field" in the applications markets, given Microsoft's now-dominant installed base, the economists should address whether divestiture (such as in AT&T) is the appropriate remedy.

Based upon the information made available to the Court as a result of this analysis, these amici believe that the Court would be in a position to accept or reject the Government's current proposed decree, or to identify those modifications that would be necessary to bring the decree within the public interest standard. Cf. AT&T, 552 F. Supp. at 153 & n.95, 212-13. At a minimum, such submissions would provide a factual record which the Court's own economist expert could review in considering the economic issues raised by the proposed decree, or to which economists could respond on behalf of other interested parties.

Given the extreme importance of these proceedings to the future of the American software industry, and hence to the economy as a whole, the Government should be permitted to do no less. As documented in previous Sections, economic theory predicts that, even without resort to its ongoing (and unchecked) illegal practices, Microsoft would very likely be able to leverage its unlawfully acquired installed base in operating systems to monopolize the entire business and home software network in the United States. The Government's decision to do nothing to restrain Microsoft's ability to engage in such monopoly leveraging, or even to curtail Microsoft's use of blatantly predatory and unlawful practices in furtherance of that end, requires explanation. Absent such explanation, these amici submit that the Court has no choice but to reject the proposed consent decree as plainly outside the bounds of the public interest.

I. Sharon S. Kelly, declare:

I am employed in the City of Palo Alto, County of Santa Clara, State of California. I am over the age of 18 years and not a party to the within entitled cause. I am readily familiar with Wilson, Sonsini, Goodrich & Rosati's practice for collection and processing of correspondence for overnight delivery by courier. In the ordinary course of business, correspondence would be consigned to a messenger service on this date.

On January 9, 1995, I served the attached NOTICE OF MOTION, MOTION, AND MEMORANDUM IN SUPPORT OF MOTION TO FILE MEMORANDUM OF AMICI CURIAE IN OPPOSITION TO PROPOSED FINAL JUDGMENT and MEMORANDUM OF AMICI CURIAE IN OPPOSITION TO PROPOSED FINAL JUDGMENT as well as the APPENDIX TO MEMORANDUM OF AMICI CURIAE IN OPPOSITION TO PROPOSED FINAL JUDGMENT on the parties listed below by placing the documents described above in an envelope addressed as indicated below, which I sealed. I consigned the envelopes to an overnight courier service by placing it/them for collection and processing this day following ordinary business practices at Wilson, Sonsini, Goodrich & Rosati, 650 Page Mill Road, Palo Alto, California 94304-1050, to be personally served on the following:

• Honorable Stanley Sporkin
  United States Courthouse
  Third Street & Constitution Ave., N.W.
  Washington, D.C. 20001

  Anne K. Bingaman
  U.S. Department of Justice
  Civil Division
  Tenth Street & Constitution Ave., N.W.
  Room 3101
  Washington, D.C. 20530

  Donald J. Russell
  U.S. Department of Justice
  Antitrust Division
  555 Fourth Street, N.W.
  Room 8249
  Washington, D.C. 20001

FOOTNOTES

1 This memorandum amici curiae is submitted by Wilson, Sonsini, Goodrich & Rosati on behalf of certain clients that prefer to retain their confidentiality. Hence, they are not identified in this submission.

2 James Wallace & Jim Erickson, Hard Drive: Bill Gates and the Making of the Microsoft Empire (1992).

3 Exhibit numbers refer to selected supporting documents which have been included in the Appendix to this Memorandum of Amici, filed herewith. For the Court's convenience, documents in the Appendix have been organized alphabetically by publication title.

4 Microsoft Corp. Employee Performance Review, dated May 4, 1987, at 3 (Ex. 21). (Although this review has become a public document, these amici have redacted the review to safeguard the employee's privacy interests.)

5 Microsoft Corp. Employee Performance Review, dated Nov. 2, 1987, at 8 (Ex. 21). (Although this review has become a public document, these amici have redacted the review to safeguard the employee's privacy interests,)

6 Id. at 6.

7 Microsoft Corp. Employee Performance Review, dated May 4, 1987, supra, at 3 (Ex. 21).

8 Examples of such "undocumented calls" will be described in Section IV, infra.

9 According to industry consultant Jerry Schneider, Microsoft's installed base in March 1988 was only nine to twelve million. Dump DOS? No Way, Not Yet, Computer Decisions, March 1988 at 50 ("between nine and twelve million DOS machines"). Indeed, according to Business Week, no more than twelve million PCs had been sold by April 1988. Will Sun Melt the Software Barrier, Business Week, April 18, 1988, at 72 ("Sun aims to coax a portion of the 12 million owners of PCs and clones into the UNIX camp.") The more expansive measure taken by industry analysts at International Data Corp. indicated there were "approximately 18 million IBM PCs and compatibles worldwide," in March 1988. Alan Radding, IBM PC Orphans Hang On To A Good Thing, Computerworld, March 7, 1988, at 81. Therefore, even under the assumption that Microsoft's operating system software had been installed in every IBM PC or compatible sold by 1988, Microsoft's installed base at that time was no larger than eighteen million. Cf. Christopher O'Malley, The New Operating Systems, Personal Computing, October 1986, at 181 ("better than 95 percent [of then-existing] PC's and compatibles use] Microsoft's disk operating system.").

10 Amy Cortese, Next Stop, Chicago, Business Week, Aug. 1, 1994, at 24 C120 million MS-DOS customers (including 55 million Windows users)"). See also OS Overview, Computer Reseller News, Aug. 22, 1994, at 223 (International Data Corporation table) (DOS and Windows installed base of 110.1 million).

11 Amy Cortese, No Slack for Microsoft Rivals, Business Week, Dec. 19, 1994, at 35 (Ex. 5).

12 Attorney General Janet Reno, Department of Justice Press Conference Transcript Microsoft Settlement (July 16, 1994) at 2 (Ex. 12).

13 Amy Cortese, Business Week, Dec. 19, 1994, supra, at 35 (Ex. 5) ("Computer makers have been startled to learn that they will be asked to swallow a huge price hike for their use of Windows 95 -- to as much as $70 per PC, vs. roughly $35 today.").

14 Larry Campbell, Novell to Introduce SuperNOS Strategy, South China Morning Post, Sept. 20, 1994, at 1 (Ex. 37) (quoting Robert Frankenburg speech to Networld + InterOp '94 conference). See also Bob Lewis, Ten Troublesome Trends in Computing That Are Sure to Spook You, InfoWorld, Oct. 31, 1994, at 82 ("Let's all admit that NextStep and QNX should have all of the market if there was any justice," but Microsoft's "Windows and DOS have more than 80 percent market share, so the wax is over! ").

15 Nor has the irony of this withdrawal been lost on the computer industry. As one observer noted: "July [of 1994] saw Microsoft in full agreement with the Justice Department. Microsoft agreed to withdraw the 'per processor' option that most PC suppliers found the cheapest way to buy DOS [in order to] encourage firms to offer alternatives to Microsoft's operating systems. Shortly afterward, Novell announced that it was stopping development of DR-DOS." Jack Schofield, Computing 94: Processor Wars and Rumors of Delays, Guardian, Dec. 29, 1994, at T14.

16 Indeed, in 1990 Microsoft began to bundle its application products together into so-called "suites." These mites are the fastest growing segment of the applications market, and Microsoft commands more than 85% of the suite market. See Personal Computing Software Worldwide, Dataquest, June 27, 1994, at 20 (selected pages at Ex. 11) (unit shipments of suites grew more than 350% in 1993); id. at 27 (Microsoft's 1993 market share for suites is 85.4%); Doug VanKirk, Integrated Office Suites, InfoWorld, Feb. 7, 1994, at 51 ("Microsoft owns a 90 percent share of the suite market .... ").

17 Hard Drive, supra, at 398-99.

18 Tr. of Status Call, Sept. 29, 1994, at 25-28.

19 Ron White, Microsoft Gives the New Word, PC Week, Oct. 20, 1987, at 95.

20 See. e.g., Brent Schlender, Bill. Gates: What Doesn't He Want, Fortune, Jan. 16, 1995, at 36.

21 Id. at 47.

22 As the sponsor of the Act, Senator Tunney, declared: "Specifically, our legislation will ï½· . . make our courts an independent force rather than a rubber stamp in reviewing consent decrees, and it will assure that the courtroom rather than the backroom becomes the final arbiter in antitrust enforcement." The Antitrust Procedures and Penalties Act: Hearings on S. 782 and S. 1088 before the Subcommittee on Antitrust and Monopoly of the Committee on the Judiciary, 93rd Cong., 1st Sess. (1973).

23 Accord AT&T, 552 F. Supp. at 148 (Congress had "found that consent decrees often failed to provide appropriate relief, either because of miscalculations by the Justice Department or because of the 'great influence and economic power' wielded by antitrust violators").

24 An initial difference between that case and the present one, of course, is that the initial decree in that case was entered after the District Court had already heard approximately 11 months of trial testimony from roughly 350 witnesses. See AT&T, 552 F. Supp. at 140.

25 See, e.g., A Fierce Battle Brews Over the Simplest Software Yet, Business Week, November 21, 1983, at 114 (Ex. 2) (quoting Microsoft executive Steve Ballmer) ("There is a very clean separation between our operating system business and our applications business... It's like the separation of church and state").

26 See, e.g., Paul Andrews, Can Microsoft Just Do It?, Seattle Times, March 18, 1991, at B1 (Microsoft "repeatedly' asserted "that a 'Chinese Wall' exists between its applications and systems divisions"); Microsoft and IBM Under Investigation by FTC, Technical Computing, Apr. 1, 1991 ("Microsoft maintains that it does not take unfair advantage of advance knowledge of operating systems in designing its consumer products. It says there is a 'Chinese Wall' between systems and applications"); Michael Stroud, ITC Widens Probe of Microsoft Dominence, Investor's Daily, Apr. 15, 1991, at 1 ("Microsoft maintains that it keeps a 'Chinese Wall' between its operating system and applications divisions to prevent such an unfair advantage from occurring"); Sean Silverthorne, AMD Files $2 Billion Antitrust Suit Against Intel, Investor's Daily, August 30, 1991, at I (Microsoft responds to charges that its application developers receive "inside knowledge" about the company's operating systems by claiming that Microsoft "has erected a 'Chinese Wall' between the two operations.").

27 The Computer Industry Survey: Reboot System and Start Again, The Economist, Feb. 27-Mar. 5, 1993, at 3 (Ex. 14).

28 Charles R. Morris and Charles H. Ferguson, How Architecture Wins Technology Wars, Harv. Bus. Rev., Mar. 1993, at 86 (Ex. 16).

29 See, e.g., All Things Considered (NPR broadcast, Nov. 17, 1994) ("if there's a sub- theme to this whole [Comdex] conference, it's networking, and Microsoft is the company that wants to connect all those different boxes that are going to be in your house."); Elizabeth Corcoran, Microsoft Heads Home: Software Giant Targets Huge Consumer Market With a Host of High-Tech Innovations, Washington Post, Nov. 13, 1994, at H1 (Ex. 44).

30 Layer 5 has been broken out into two parts to reflect the development of what are known in the industry as "client-server" applications: applications that run partially on the desktop, and partially on server hardware connected to the desktop by a computer network.

31 PC Week, Feb. 21, 1994, at 39 (Paine Webber, Inc. Table) (excluding sales of Macintosh -- which does not use X86 chips -- Microsoft's 1994 market share was 92.4%). See also Computerworld, Dec. 6, 1993, at 99 (International Data Corp. Table) (Microsoft 1992 market share is 92.5%).

32 The situation with respect w UNIX is slightly more complex, but in the final analysis, the situation is the same. UNIX has a strong following among technical engineering (as opposed to business) users of computers. There are companies that have written technical engineering application programs (such as "computer aided design" programs) to run on UNIX. But, as with Apple, the business applications vendors for the UNIX platform are the same companies that write applications for Windows. Hence, by controlling business desktop application programs, Microsoft can keep UNIX from penetrating the business desktop market.

33 Laura DiDio, NetWare, NT Server to Divide Lion's Share, Dec. 26, 1994, at 77 ("The network operating system arena looks like a two-horse race in 1995, with Novell, Inc. 's NetWare 4.1 and Microsoft Corp. 's Windows NT Server 3.5 locked in a battle for first place.").

34 See, e.g., Michelle Flores, Probe of Microsoft is Extended -- Justice Dept. Asks For More Information, Seattle Times, Nov. 22, 1994, at B11 (electronic banking is the "killer app. of the '90s").

35 Prior to rumors of the acquisition, Intuit's stock traded at 40 3/4. John Eckhouse, Giant Microsoft Buys Intuit for $1.5 Billion, San Francisco Chronicle, Oct. 14, 1994, at A1, A19. Each Intuit share is to receive 1.336 Microsoft shares at the closing. Id. Based on Microsoft's January 3, 1995 closing price of 60 3/16, each Intuit share receives over $80.

36 See Brian Livingston, Undocumented Windows Calls, InfoWorld, Nov. 16, 199"2, at 98 (Ex. 19); Doug Barney and Ilan Greenberg, ISVs Dampen Microsoft Furor for OLE, InfoWorld, July 18, 1994, at 1.

37 Tr. of Status Call, Sept. 29, 1994, at 25-28.

38 Microsoft has made numerous presentations around the country that specifically make this point and written documentation from these presentations has been provided to the Justice Department.

39 For a more thorough discussion, see Miles B. Keyhoe, The Winds of Change, HP Professional, Aug. 1994, at 40 (Ex. 17). See also Microsoft Corporation, Microsoft Windows NT and Client-Server Computing, May 1993.

40 W. Brian Arthur, Positive Feedback in the Economy, Scientific American, Feb. 1990, at 92, 93 (Ex. 36).

41 Id. at 92 (Ex. 36).

42 "Installed base" in the economic literature "means the number of owners of a good who may be dependent on the manufacturer of the good for the provision of complementary goods." Joseph Katten, Market Power in the Presence of an Installed Base, 62 Antitrust L.J. 1, 4 (1993).

43 Joseph Farrell and Garth Saloner, Installed Base and Compatibility: Innovation, Product Pre-Announcements, and Predation, Amer. Econ. Rev., Dec. 1986, at 940; Janusz A. Ordover and Garth Saloner, Predation, Monopolization, and Antitrust, in Handbook of Industrial Organization 537, 565 (R. C. Schmalensee and R. Willis eds., 1989).

44 W. Brian Arthur, Scientific American, Feb. 1990, supra, at 92-93 (Ex. 36).

45 See, e.g., Garth Saloner, Economic Issues in Computer Interface Standardization, Econ. Innov. New Tech., 1990, at 140-142.

46 See, e.g., Paul A. David, Clio and the Economics of QWERTY, Amer. Econ. Rev., May 1985, at 332; David A. Harvey, Ergonomic Issues Have Taken a Backseat to Performance, Resulting in a Growing Tide of Computer-Related Injuries. Change is Needed - Now!, Byte, Oct. 1, 1991, at 119.

47 See David A. Harvey, Byte, Oct. 1, 1991, supra, at 120.

48 Joseph Farrell & Garth Saloner, Amer. Econ. Rev., Dec. 1986, supra, at 942; Jean Tirole, The Theory of Industrial Organization at 405, n.40 (1988)

49 W. Brian Arthur, Scientific American, Feb. 1990, supra, at 93.

50 See W. Brian Arthur, Increasing Returns & Path Dependence in the Economy, 1994, at ix (forward Kenneth J. Arrow).

51 For the theoretical literature see, for example, the recent Symposium on Network Externalities in the Journal of Economic Perspectives, Spring 1994, the Symposium on Compatibility, edited by Richard Gilbert in the Journal of Industrial Economics, March 1992, and the survey by Paul David and Shane Greenstein in the Economics of Innovation and New Technology, 1990. For an application to telecommunications, see Stanley Besen and Garth Saloner, The Economics of Telecommunications Standards, in Changing the Rules: Technological Change, International Competition, and Regulation in-Communications 177 (1989); for applications to broadcasting, see Stanley Besen and Leland Johnson, Compatibility Standards, Competition, and Innovation in the Broadcasting Industry (1986); for applications to ATMs, see Garth Saloner and Andrea Shepard, forthcoming in the Rand Journal of Economics, and Steven Salop, Deregulating Self-Regulated Shared ATM Networks, Econ. of Innov. and New Tech., 1990; and for computers, see Garth Saloner, Econ. Innov. New Tech., 1990, supra.

52 This "network effect" has been described by numerous authors. In a recent Symposium in the Journal of Economic Perspectives, Michael Katz and Carl Shapiro write, "Consequently, as has long been recognized, the demand for a network good is a function of both its price, and the expected size of the network." See also Jeffrey Rohlfs, A Theory of Interdependent Demand for a Communications Service, Bell J. of Econ., Spring 1974, for an early reference, as well as Michael Katz and Carl Shapiro, Network Externalities, Competition, and Compatibility, Amer. Econ. Rev., June 1985; Joseph Farrell and Garth Saloner, Amer. Econ. Rev., Dec. 1986, supra: and other papers cited in Michael If, am and Carl Shapiro, Systems Competition and Network Effects, J. of Econ. Perspectives, Spring 1994.

53 See Julio J. Rotemberg and Garth Saloner, Interfirm Competition and Collaboration, Strategic Options, 1991, for an example of the power of network size.

54 For early examples in the economics literature, see Joseph Farrell and Garth Saloner, Standardization, Compatibility, and Innovation, Rand J. of Economics, Spring 1985 and Michael Katz and Carl Shapiro, Amer. Econ. Rev., supra; Jean Tirole, supra, at 405.

55 Jean Tirole, supra, at 404-406.

56 Garth Saloner, Econ. Innov. New Tech., 1990, supra, at 137-138; Joseph Farrell and Carl Shapiro, Dynamic Competition with Switching Costs, Rand J. of Econ., Spring 1988, at 123-137.

57 Stanley M. Besen and Joseph Farrell, J. of Econ. Perspectives, Spring 1994, supra, at 118; Joseph Farrell and Garth Saloner, Amer. Econ. Rev., Dec. 1986, supra, at 946.

58 Joseph Farrell and Garth Saloner, Rand J. of Econ., Spring 1985, supra; Joseph Farrell and Garth Saloner, Amer. Econ. Rev., Dec. 1986, supra.

59 See Garth Saloner, Econ. Innov. New Tech., 1990, supra, at 141-142.

60 For a detailed review of Microsoft's server strategy, see How Microsoft's Server Strategy Will Change The Industry, supra, (Ex. 38).

61 There was clearly the potential for at least some leverage from the chip or hardware level, when the OS level was more fragmented. This possibility is not treated in this brief for a number of reasons, including the widely publicized alliance between Microsoft and Intel that makes separate treatment of the hardware layer irrelevant.

62 The operating system gateways are the most effective layers for leverage. But the system can also be leveraged from other access points as to which strong network externalities attach. For example, on the home client, Intuit has leverageable power from the strong network externalities that have attached to that product at the computer-human interface. (This is described in greater detail elsewhere in this brief.)

63 In many respects Microsoft's strategy of targeting, linking and leverage is little different from that employed by MITI and Japanese keiretsus to target and capture American markets. Microsoft's leverage comes from technical ties in markets it dominates, while Japanese companies' leverage comes from the installed base of buyers it creates in Japan. In both cases, the leverage can be applied by forward-pricing into the target market to damage competition in that market. Cf., L. D. Tyson, Who's Bashing Whom.? Trade Conflict in High-Technology Industries, at 55-57, 99-101 (1992).

64 See supra note 9.

65 Amy Cortese, Next Stop, Chicago, Business Week, Aug. 1, 1994, at 24 C 120 million MS-DOS customers (including 55 million Windows users)"). See also OS Overview, Computer Reseller News, at 223 (DOS installed base of 110.1 million).

66 See Start Miastkowski, Digital Research Creates a Better DOS, Byte, Nov. 1991, supra, at 68.

67 See Amy Cortese, Business Week, Dec. 19, 1994, supra, at 35 (Ex. __) ("Computer makers.., have been startled to learn that they will be asked to swallow a huge price hike for their use of Windows 95 -- to as much as $70 per PC vs. roughly $35 today.").

68 See also O. Casey Corr, IBM vs. Microsoft -- Software Superbowl -- IBM to Kick Off New Version of OS/2, but will Microsoft Make Winning Goal, Seattle Times, March 29, 1992, at C1 (system sales are "the cash cow that has fueled Microsoft's aggressive entry into nearly every field of personal computing"); id. ("DOS, which comes installed on computers at the factory, has provided profits to finance Microsoft's development of applications such as the Excel spreadsheet and Word, a writing program."); Laurie Flynn & Rachel Parker. Extending its Reach, InfoWorld, August 7, 1989, at 43 ("the Microsoft strategy has been to fund expensive applications development and marketing with its profits from the recurring DOS royalties it receives.").

69 Kathy Rebello, et al., Is Microsoft Too Powerful, Business Week, March 1, 1993 at 88 (Ex. 4).

70 See Cara A. Cunningham, IBM and Microsoft Wage Open Doc vs. OLE Find, InfoWorld, Aug. 15, 1994, at 25 (Microsoft has an "army of evangelists.., that goes out and sells the [OLE] technology and swarms over developers").

71 Richard Brandt, Microsoft Is Like an Elephant Rolling Around, Squashing Ants, Business Week, Oct. 30, 1989, at 148 (Ex. 3).

72 Ray Weiss, Windows Stars at SD 91, Electronic Engineering Times, Feb. 18, 1991 (Ex. 15).

73 See supra note 27.

74 Compare Andrew Reinhardt, First Impressions: New Extras for Excel, Byte, Feb. 1, 1991, at 136 with Microbytes, Byte, Feb. 1, 1991, at 20.

75 See, e.g., Start Levine, Lotus Embraces 'Competition As It Aims for Identity, LAN Times, June 17, 1991.

76 Ethan Winer, BASIC, Yes; Feeble, No, PC Magazine, Oct. 30, 1989, at 187 (Because "the BASIC [programming language] interpreter [is] bundled with DOS... at no extra cost, [it] is known and used by more people than any other programming language for personal computers.").

77 Michael Csenger & Adam Griffin, Microsoft Free At Last?, Ruling Still Lets Firm Incorporate Apps Into Its OS'es, Network World, July 25, 1994, at 4 (Ex. 23); see also John Markoff, Microsoft's Future Barely Limited, N.Y. Times, July 18, 1994, at D1 (Ex. 24) (describing Microsoft's 14 year "campaign[] to expand the definition of what computing functions belong inside the computer operating system.").

78 Paul Andrews, Windows Is No JFK, But Its Visual Appeal Is Outstanding, Seattle Times, May 22, 1990, at C2 ("Windows 3.0 comes with a suite of mini-applications including Write, Paintbrush, Clock, Recorder (a macro utility), and Terminal (telecommunications). ").

79 O. Casey Corr, A Look Behind Stac Deal, Seattle Times, June 26, I994, at FI (quoting Stac's complaint).

80 Id.; Charles McCoy, Microsoft to Pay Stac Judgment of $120 Million, Wall St. J., Feb. 24, 1994, at A4.

81 Stuart J. Johnston, Microsoft Settles for Piece of Stac, Computerworld, June 27, 1994, at 30 (Microsoft paid $39.9 million for 155'o of Stac, and an additional $43 million over 43 months for a license to Stac's data compression technology); Doug Barney, Microsoft, Stac Resolve Dispute; Microsoft Finally Pays Up, InfoWorld, June 27, 1994, at 14.

82 O. Casey Corr, A Look Behind Stac Deal, supra, at Fl.

83 As explained in Section V.C., infra, the superficially irrational behavior of undermining the application vendors that produce programs that run on Microsoft's operating system is logical specifically because Microsoft has an independent economic incentive to monopolize the market for business application programs.

84 Elizabeth Corcoran, Washington Post, Nov. 12, 1994, supra, at H6.

85 Amy Cortese, Business Week, Dec. 19, 1994, supra, at 35 (HP, Compaq and other big U.S. PC makers plan to bundle Windows 95 into their machines).

86 See Lawrence J. Magid, Microsoft: Not So Marvelous, Bay Area Computer Currents, Dec. 1, 1994, at 98, 101 (Ex. 1); Carole Patton, Computerworld, Nov. 14, 1994, supra, at 57 (Ex. 8).

87 Lee Gomes, Microsoft to Acquire Intuit, San Jose Mercury News, Oct. 14, 1994, at 1D.

88 Don Clark, Microsoft to Buy Intuit In Stock Pact, Wall St. J., Oct. 14, 1994, at A3 (86% of retail store sales); Karen Epper, Software Deal Shakes Up Home Banking, Amer. Banker, Oct. 17, 1994, at 1, 25 (80-85%).

89 Michelle Flores, Probe of Microsoft is Extended - Justice Dept. Asks For More Information, Seattle Times, Nov. 22, 1994, at B11; Michael Schrage, Microsoft Can Make Lots of Money; Can It Shape the Management of It?, Washington Post, Oct. 21, 1994, at B3; Brent Schlender, Fortune, Jan. 16, 1995, supra, at 36.

90 Gina Smith, Merger Misgivings: Will Intuit Go 'Soft?, S.F. Chronicle, Dec. 4, 1994, at B5, B14.

91 Brent Schendler, Fortune, Jan. 16, 1995, supra, at 4748; see also, Michael I. Miller, PC Magazine, Jan. 24, 1995, supra, at 80 (Ex. 25) ("Microsoft could require just a small service charge on each transaction. Or it could make money on the float -- the interest in the few seconds it takes to move money from one place to another. Or both.").

92 For example, leading industry analyst Rick Sherlund of Goldman Sachs predicted that with the settlement, Microsoft "should dominate the market for desktop software for the next 10 years." And another leading analyst, Richard Shaffer concluded that "It]he operating system wars are over -- Microsoft is the winner .... Microsoft is the Standard Oil of its day." Andrew Schulman, Microsoft's Grip On Software Tightened By Antitrust Deal, Dr. Dobb's Journal of Software Tools, Oct. 1994, at 143 (Ex. 13).

93 See John M. Goodman, The DOS Heavyweights Go Another Round, InfoWorld, Aug. 29, 1994, at 87 (rating PC-DOS version 6.3 above MS-DOS version 6.22) and Earle Robinson, DOS-version Madness? Integration Coping with DOS, Windows Sources, Oct. 1994, at 163 ("my choice would be the IBM . . . it's cheaper") and Yael Li-Ron, PC DOS 6.3: DOS and DOS: Separated At Birth, PC-Computing, July 1994, at 94 (IBM's Ambra computers ship with MS-DOS).

94 Don Clark & Laurie Hays, Microsoft's New Marketing Tactics Draw Complaints, Wall St. J., Dec. 12, 1994, at B6 (Ex. 41).

95 Id.

96 All of these problems are discussed in Rory O'Connor, San Jose Mercury News, Nov. 13, 1994, supra, at 1A, 28A (Ex. 34).

97 See Michael Katz and Carl Shapiro, Systems Competition, supra.

98 Joseph Farrell and Garth Saloner, Installed Base, supra; Paul David, Amer. Econ. Rev., May 1985, supra.

99 Indeed, Microsoft's operating system "lock-in" has permitted it to bring demonstrably inferior products to market (products that did not enjoy any appreciable consumer acceptance) without negative consequences to the company. See Michael Morris, Microsoft Deal: Too Little, Too Late, S.F. Examiner, July 24, 1994, at C-5. (Ex. 33)

100 Joseph Farrell, Hunter K. Monroe and Garth Saloner, The Vertical Organization Of Industry and Systems Competition Versus Component Competition, October 1994 (working paper).

101 See, e.g., supra, note 32. (Microsoft presently holds greater than 90% of the X86 operating system market share); Christopher O'Malley, Personal Computing, October 1986, supra, at 181, 183 ("Microsoft's operating system" has "better than 95 percent" share of the X86 systems.)

102 Department of Justice Press Conference (July 16, 1994), at 3-11 (by Asst. Attorney General Anne Bingaman).

103 See also Stuart J. Johnston, Decree: Deal or Dodge?, Computerworld, July 25, 1994 ("Interviews with PC hardware vendors last week indicated few are likely to switch to a competing system any time soon. 'Customers have already voted with their dollars in a very strong way for DOS and Windows. I don't see that changing,' said Howard Elias, a vice president at AST Research, [a leading OEM].") Jane Morrissey, DOJ Accord Fosters 'Too Little, Too Late' Perception, PC Week, July 25, 1994, at 1 ("[O]bservers doubt the consent decree agreed on will have much effect on the company or its competitors," because it is "too little, too late. "); Jesse Berst, Behind The Smoke: Microsoft Wins Again, PC Week, July 25, 1994, at 106 ("Does the agreement really change anything? No .... If the decree had come five years ago, when there were viable MS-DOS clones, it might have had some immediate impact. Now, in a world where MS-DOS is on the way out and Windows has no real clones, it will have no short-term impact") (Ex. 27); Andrew Schulman, Dr. Dobb's Journal of Software Tools, Oct. 1994, supra, at 143 ("the change from per-processor to per-copy licensing probably comes about four years too late"); Claudia Maclachlan, Software Makers Mull Over Microsoft Legal Challenge, National Law Journal, Aug. 1, 1994, at B1 ("They can't do [original equipment manufacturer] pricing, but they don't need it anymore.")

Indeed, even Microsoft's supporters concede that, "[a] year from now, [the proposed decree] will be" no more than "a blip on the radar screen of computing history." William Casey, Let's Stop Beating On Microsoft, Washington Post, July 25, 1994, at F15. "Issued five years ago, the ruling would have had an effect.., users were open to alternative environments, even if it meant migrating from [Microsoft's products]." Id. "Those choices, and the years in which they could have been made freely, are ancient history .... It's a fact that [today] the operating environment of choice on Intel-based processors is DOS and Windows." Id.

104 See also John Markoff, N.Y. Times, July 18, 1994, supra, at D1 (Ex. 24) ("The agreement leaves untouched what many computer industry executives say is Microsoft's principal advantage -- that it develops both the basic operating system software that makes personal computers run.., and applications software.., that performs specific tasks."); id. ("The other important issue not specifically addressed in the consent decree is whether Microsoft has been able to leverage its virtual monopoly in operating systems into domination of applications software -- a far bigger and more lucrative market"); Claudia Maclachlan, National Law Journal, Aug. 1, 1994, supra, at B1 ("As long as [Microsoft has] a dominant position in operating systems ... it allows them to leverage that into applications. This agreement does nothing to the sums quo.") (internal quotation omitted).

105 See, e.g., Digidyne Corp. v. Data General Corp., 734 F.2d 1336, 1340-43 (9th Cir. 1984), cert. denied, 473 U.S. 908 (1985); (software); Ortho Diagnostic Systems, Inc. v. Abbott Laboratories, Inc., 822 F. Supp. 145, 155-56 (S.D.N.Y. 1993) (blood screening technology); Viacom International, Inc. v. Time Inc., 785 F. Supp. 371,377 (S.D.N.Y 1992). See also Lee v. Life Ins. Co., 829 F. Supp. 529, 537-39 (D.R.I. 1993), aff'd, 23 F.3d 14 (1st Cir.), cert. denied, 1994 U.S. LEXIS 7596 (1994).

106 See, e.g., David Einstein, S.F. Chronicle, July 18, 1994, supra, at A1 (Ex. 32) (Ernie Simpson, president of a software company which develops programs for use with Windows, called the decree "a waste of time"); Quote of the Week, InformationWeek, Aug. 1, 1994, at 10 (Reacting to the proposed decree, Gordon Eubanks, CEO of software firm Symantec Corp., said simply, "That's it?"); John Markoff, N.Y. Times, July 18, 1994, supra, at D1 (Ex. 24) (quoting Martin Goetz, cofounder of Applied Data Research, the nation's first software company, as saying of the decree, "The Justice Department hasn't listened to the cries of the software companies"); Jane Morrissey, PC Week, July 25, 1994, supra, at 1 (Ex. 26) (quoting Mitchell Kertzman, chairman of Powersoft Corp., as saying the proposed decree will have "close to zero impact," and that "to the extent that Microsoft's behavior prevented other operating systems from succeeding, the war is over ... DOS is it and Windows is it"); Andrew Schulman, Dr. Dobb's Journal of Software Tools, Oct. 1994, supra, at 143 (Ex. 13) (quoting spokesman for Compaq as saying "Windows is the standard--not much will change"). See also David Einstein, S.F. Chronicle, July 18, 1994, supra, at A1 (Ex, 32) (quoting a leading industry analyst as concluding that "[t]he operating system wars are over -- Microsoft is the winner ... Microsoft is the Standard Oil of its day"); Claudia Maclachlan, National Law Journal, Aug. 1, 1994, supra, at B1 ("As long as [Microsoft has] a dominant position in operating systems ... it allows them to leverage that into applications. This agreement does nothing to the status quo") (internal quotations omitted).

GARY L. REBACK (Bar No. 218594)
SUSAN A. CREIGHTON
DAVID A. KILLAM
NEIL M. NATHANSON
IRWIN R. GROSS
WILSON, SONSINI, GOODRICH & ROSATI
Professional Corporation
650 Page Mill Road
Palo Alto, California 94304--1050
Telephone: (415) 493-9300

FOR THE DISTRICT OF COLUMBIA

INDEX

TAB

1. Lawrence J. Magid, Microsoft: Not So Marvelous, Bay Area Computer Currents, Dec, 1, 1994, at 98.

    

2. A Fierce Battle Brews Over the Simplest Software Yet, Business Week, Nov. 21, 1983, at 114.

    

    

3. Richard Brandt, Microsoft Like An Elephant Rolling Around, Squashing Ants, Business Week, Oct. 30, 1989, at 148.

    

    

4. Kathy Rebello, et al., Is Microsoft Too Powerful, Business Week, Mar. 1, 1993, at 82.

    

    

5. Amy Cortese, et al., No Slack For Microsoft Rivals, Business Week, Dec. 19, 1994, at 35.

    

    

6. Byte, May 1992, at 157, 159.

    

    

7. Stuart J. Johnston and Ed Scannell, Server Suite Could Squeeze Market, Computerworld, Oct. 10, 1994, at 4.

    

    

8. Carole Patton, Bundles Are Bad News, Computerworld, Nov. 14, 1994, at 57.

    

    

9. William Brandel and Ed Scannell, Developing for Next Generation of Windows May Mean Running on NT, Computerworld, Nov. 28, 1994, at 4.

    

    

10. The Long View From Microsoft: Component DBMSs, Datamation, Aug. 1, 1994, at 40.

     

     

11. Personal Computing Software, Dataquest Report (1994) (selected pages).

     

     

12. Department of Justice Press Conference (July 16, 1994).

     

     

13. Andrew Schulman, Microsoft's Grip on Software Tightened by Antitrust Deal, Dr. Dobb's Journal of Software Tools, Oct. 1994, at 143.

     

     

14. Computer Industry Survey: Reboot System and Start Again, The Economist, Feb. 27-Mar. 5, 1993).

     

     

15. Ray Weiss, Windows Stars at SD 91, Electronic Engineering Times, Feb. 18, 1991.

     

     

16. Charles R. Morris and Charles H. Ferguson, How Architecture Wins Technology Wars, Harv. Bus. Rev., Mar.-Apr. 1993, at 86.

     

     

17. Miles B. Keyhoe, Microsoft readying Three 32-Bit Operating Systems, HP Professional, Aug. 1994, at 40.

     

     

18. Stuart J. Johnston, 'No Chinese Wall' at Microsoft, Infoworld, Dec. 30, 1991, at 107.

     

     

19. Brian Livingston, Undocumented Windows Calls, InfoWorld, Nov. 16, 1992, at 98.

     

     

20. Brian Livingston, Will 'Windows Compatible' Really Mean What it Says?, InfoWorld, Nov. 14, 1994, at 40.

     

     

21. Microsoft Corp. Employee Performance Reviews, dated Nov. 2, 1987, and May 4, 1987.

     

     

22. Confidential Multimedia Systems Development Partner Program, Microsoft Corp., Aug. 1990.

     

     

23. Michael Csenger & Adam Griffin, Microsoft Free At Last?, Ruling Still Lets Firm Incorporate Apps Into Its OS'es, Network World, July 25, 1994, at 4.

     

     

24. John Markoff, Microsoft's Barely Limited Future, N.Y. Times, July 18, 1994, at D1.

     

     

25. Michael J. Miller, The World According to Microsoft, PC Magazine, Jan. 24, 1995, at 80.

     

     

26. Jane Morrissey, DOJ Accord Fosters 'Too Little, Too Late' Perception, PC Week, July 25, 1994, at 1.

     

     

27. Jesse Berst, Behind The Smoke: Microsoft Wins Again, PC Week, July 25, 1994, at 106.

     

     

28. Eamonn Sullivan and Matt Kramer, Microsoft Marvel Beta Leverages WIN 95 Desktop, PC Week, Nov. 7, 1994, at 169.

     

     

29. Mary Jo Foley and Lisa DiCarlo, PC Vendors Allege Undue Pressure From Microsoft, PC Week, Nov. 21, 1994, at 1.

     

     

30. Jesse Berst, Microsoft's On-Line Rivals Could End Up In 'Cyberia', PC Week, Dec. 12, 1994, at 120.

     

     

31. Jeff Bertolucci, Microsoft Settles: Business As Usual, PC World, Oct. 1994, at 72.

     

     

32. David Einstein, Microsoft Unscathed by Settlement, S.F. Chronicle, July 18, 1994, at A1.

     

     

33. Michael Morris, Microsoft Deal: Too Little, Too Late, S.F. Examiner, July 24, 1994, at C-5.

     

     

34. Rory J. O'Connor, Microsoft, Intel Set to Define Technology, San Jose Mercury News, Nov. 13, 1994, at 1A.

     

     

35. Microsoft Domination, San Jose Mercury News, Dec. 21, 1994, at IF.

     

     

36. W. Brian Arthur, Positive Feedback in the Economy, Scientific American, Feb. 1990, at 92.

     

     

37. Larry Campbell, Novell to Introduce SuperNOS Strategy, South China Morning Post, Sept. 20, 1994, at 1.

     

     

38. How Microsoft's Server Strategy Will Change The Industry - Parts I & II, Report by Summit Strategies, Inc.

     

     

39. Microsoft and Visa To Provide Secure Transaction Technology For Electronic Commerce, Visa News Release, Nov. 8, 1994.

     

     

40. Viveca Novak and Don Clark, Trade Group's Board Cancels Hearing On Microsoft's Plan to Acquire Intuit, Wall St. J., Dec. 5, 1994, at B6.

     

     

41. Don Clark and Laurie Hays, Microsoft's New Marketing Tactics Draw Complaints, Wall St. J., Dec. 12, 1994, at B6.

     

     

42. Elizabeth Corcoran, Microsoft Dea1 Came Down To a Phone Call, Washington Post, July 18, 1994, at A1.

     

     

43. Elizabeth Corcoran, Microsoft's Plan To Buy Intuit Raises Concern; Trade Group Calls 2 Hearings To Get Industry Opinion On Deal, Washington Post, Nov. 8, 1994, at C1.

     

     

44. Elizabeth Corcoran, Microsoft Heads Home, Washington Post, Nov. 13, 1994, at H6.

     

TO

APPENDIX TO MEMORANDUM OF AMICI CURIAE
IN OPPOSITION TO PROPOSED FINAL JUDGMENT
IN CIVIL ACTION NO. 94-1564 (SS)
SIGNED BY GARY REBACK

BAY AREA COMPUTER CURRENTS       DECEMBER 1 -- DECEMBER 12, 1994

Microsoft Chairman Bill Gates didn't get to be the richest person in America by being modest or by playing patsy with his competitors. So it came as no surprise to hear Gates belittle his competition and exaggerate the value of his offerings at the recent Comdex trade show.

Gates was introducing the Microsoft Network. Preliminary reports on this online service, code-named "Marvel," have been circulating for months.

Like everything Gates announces, The Microsoft Network is purported to be the greatest thing since individually wrapped cheese slices. In introducing the service, Gates made some not-so- subtle digs at current online providers. "There is an opportunity to bring innovation into this market," said Gates, adding that existing services "take print material and move it over.'

GOOD, NOT REVOLUTIONARY
Due that's not entirely true, While the three big online services (CompuServe, Prodigy, and America Online) each offer online magazines, newspapers, and other print material, they also offer interactive forums, live chat rooms, shareware libraries, online technical support, and other information and interactive services that you can't get from your local newspaper. All three services are also experimenting with sound and graphics, and all plan to introduce animation and full motion video when communications technology (i.e., ISDN) lets them get around the limits of today's phone system. Bill Gates and his team of developers may be smart, but they have an exaggerated opinion of themselves when compared to the rest of the world.

Besides, what did Gates show when he demonstrated the service! An icon pointing to an online edition of USA Today. This is creative? To be fair, Gates also demonstrated some interesting new technology, including an online prototype of Microsoft Bookshelf, the company's multimedia reference guide. Most commercial online services offer online encyclopedias and other reference works, but none currently include graphics as a routine part of the deal.

Gates also showed how the Microsoft Network will be integrated with Windows 95, Users will be able to drag icons directly from the service to their Windows desktop. In theory at least, information that's online will be as easy to locate as information on your computer's hard disk. Of course, your modem will have to dial into the network to retrieve the information, at least until we're all hard wired into cyberspace.

I was also impressed by the way the Microsoft Network will display complex graphics like color photos. When you enter an area that uses images, graphics will quickly reveal themselves in low-resolution form, then become sharper and more vivid as data streams over the modem. This gradual display of graphics is necessa because the Microsoft Network will initially suffer the same phone-line and modem limitations that other online services do.

As interesting as Microsoft's service may be, it's hardly revolutionary. Prodigy, CompuServe, America Online, and Interchange (Ziff-David's forthcoming service) are all capable of offering similar features. The Microsoft Network won't be available to the public for at least eight months; its competitors will have plenty of time to catch up, if not move beyond Microsoft's plans.

Three basic issues affect the popularity of online services - interface, price, and content. Prodigy, CompuServe, and America Online can compete quite successfully on all three fronts. All three have plenty of time to tweak their interfaces, all can adjust their prices, and they all have a head start when it comes to content. Prodigy will announce a major interface overhaul in early 1995, and has had several years to build up its online content. CompuServe is reportedly working on easier-to-use software and, after nearly 15 years, is a leader in online databases, shareware, and more. America Online, though lacking the content of its two major competitors, leads the way with online editorial offerings and is generally regarded as being easy and pleasant to use.

Ziff-Davis's Interchange is every bit as up-to-date as Microsoft's Network. The only advantage Microsoft has is its ability to build its online software into Windows 95. And therein lies my biggest concern.

A RIGGED PLAYING FIELD
By making its network part of the operating system, Microsoft tilts the playing field in its direction. Microsoft clearly has the right to enter the online business, but I question whether it's fair to the other players if the Microsoft Network - and only the Microsoft Network - is part of Windows 95.

Imagine that utility companies said appliances. You've just moved into an empty house and after turning on the power, the power person says she has a great refrigerator in the truck that she'd be happy to install for you. "It's as cheap as any you'll get in town and you don't have to make any payments until after you've used it for a while. Besides, our refrigerator is optimized to work best with our electricity." That utility company would sell a lot of refrigerators - and every other appliance vendor would rightly cry foul. This would never happen in real life, because utilities are regulated monopolies. Microsoft, despite the Justice Department's recently rulings, is a virtual monopoly, controlling nearly 80 percent of the personal computer operating system business.

Gates claims that his bundling the Network with Windows 95 is no different than IBM's bundling Prodigy software with some of its machines. And IBM does own half of Prodigy. But IBM also offers America Online on some of its machines. More germane, IBM controls only a fraction of the personal computing market. Nobody, except Microsoft, has a grip on more than about 10 percent of the market.

Microsoft's bundling scheme has caused America Online president Steve Case to cry foul, accusing Microsoft of creating an "unlevel playing field." Others in the online industry agree. Robert D. Mainor, vice president of Product Marketing for CompuServe, didn't go as far as Case in criticizing the Microsoft announcement, but he did say that "Microsoft enjoys a distribution model that no one else has access to." He added that his service, in business for about 15 years, is in a good position to compete with Microsoft. If CompuServe's claim of 2.4 million members is accurate, it is currently the largest online service.

Prodigy's president, Ross Glatzer, said that Microsoft's entry will help expand the total online market. However, he agrees that Microsoft has its thumb on the scale. Glatzer would welcome the opportunity to include Prodigy and other online service software with Windows 95 so that users would have free choice of services.

I think Microsoft is a great company. It produces some excellent programs and it improves America's trade imbalance. But it doesn't have the right to run roughshod over the entire computer industry. Microsoft's practices affect its competitors and, ultimately, its customers. The computer industry needs competition and a balance of power. Right now, that power is tilting toward Redmond, WA. It's time for the folks in the other Washington - the one between Maryland and Virginia - to wake up and start taking a hard look at Microsoft's anti-competitive behavior. *

© 1994 Lawrence J. Magid. All rights reserve.
Larry Magid is the author of Cruising Online: Larry Magid's Guide to the New Digital Highways (Random Housc. 1994) and The Little PC Book: A Gentle Introduction to Personal Computers (Peachpit Press. 199J). He is also an internationally syndicated columnist for the L.A. Times. You can reach Larry on the Internet at magid@latimes.com, via CompuServe at 75.100,2105. via Prodigy at KPVN58A, via America Online as LarryMagid, or care of Computer Currents.

TO

APPENDIX TO MEMORANDUM OF AMICI CURIAE
IN OPPOSITION TO PROPOSED FINAL JUDGMENT
IN CIVIL ACTION NO. 94-1564 (SS)
SIGNED BY GARY REBACK

Copyright 1983 McGraw-Hill, Inc.
Business Week

November 21, 1983

SECTION: INFORMATION PROCESSING; Software; Pg. 114

LENGTH: 1565 words

HRADLING: A FIERCE BATTLE BREWS OVER THE SIMPLEST SOFTWARE YET

BODY:
The all-out fight for supremacy among the hardware makers -- Apple, Digital Equipment, IBM, and 150 others -- has been getting the headlines recently in the mushrooming personal computer market. But while it may never attract as much attention, an equally important battle is about to explode among the leading companies that write software for the small machines. Prompting this latest free-for-all is the emergence of an entirely new class of product called "environment" software Environment software has no specific application such as word processing or financial analysis. It is designed to make such jobs easier for users who are not technically trained -- a group that is rapidly becoming the majority of personal computer users. The new generation of software, which was pioneered with such machines as Xerox Corp.'s Star and Apple Computer Inc.'s Lisa, splits the computer screen into sections, or "windows." Users can run different applications software simultaneously in each window. Normally, a computer displays only one program at a time.

The new software more or less replicates tne desktop on the computer screen. In effect, a business executive or professional can put the equivalent of a letter, financial spreadsheet, or Rolodex file in the different windows on the computer display, or electronic desk.

The software battle pits three leading developers -- Microsoft, Digital Research, and VisiCorp -- against one another. Each wants its product to become the industry standard. The competition is especially fierce, because windowing programs are expected to be standard on every personal computer -- a market potential of as many as 5 million units in 1984 alone. "Environments used to feature, but now they have become a fundamental part of the [personal comter] system," says Esther Dyson, president of Rosen Research Inc.

In the fight to get their new software running on the largest variety of computer brands, the competitors are wooing the hardware makers for endorsements. The outcome will shape not only the future of the current $1.5 billion annual market for personal computer programs but also sales of the equipment itself, since the machines tnat run the most popular software will be among the best sellers. As a result, leading hardware makers -- most notably Apple and International Business Machines Corp. -- are being drawn into the fray.

The battle lines are forming rapidly. On Nov. 10, Microsoft Corp. was expected to announce that 23 computer makers -- including Apple, Digital Equipment Honeywell, Tandy, and Texas Instruments -- have signed up to muse its version of the new software, which it says will be ready by April. Meanwhile, VisiCorp, the first software company to offer this next-generation product to the industry, is racing to get its VisiOn program out the door (page 115). Not to be left out, Digital Research Inc., which has not yet demonstrated its product, is leaking word that it will begin delivering its version of environment software to as many as 10 computer makers before the end of the year. "It's a real battle of the software developers," says Steven A. Ballmer, a vice-president at Microsoft.

Competition will be fierce because there is not enough room in the personal computer marketplace to support several versions of environment software. Applications programs for specific tasks such as word processing and financial analysis will have to be rewritten to work with each environment package.

"The battle is to establish whose [environment software] is going to win, because software developers don't want to write programs for 18 different systems," says Rosen Research's Dyson. Adds Moize Adney, manager of internal software development at Texas Instruments Inc.: "The pressure to standardize will be there."

INDUSTRY STANDARD. Companies racing to market environment software are placed in something of a catch-22 situation. They must convince computer makers that many writers of applications software will develop useful programs to operate with their environment packages. But they must convince the software writers at the same time that their environment programs will be used on the largest number of machines.

Microsoft hopes to parlay the popularity of its MS/DOS operating system -- the housekeeping software that controls the basic functions of a computer -- into a marketing edge. Its new environment program, Microsoft-Windows, is actually just an extension of its operating system. An impressive 40% of personal computers sold -- including the best-selling IBM PC -- are controlled by the Bellevue (Wash.) company's operating system software. This penetration, Microsoft maintains, provides a ready market for Windows. If every computer that runs Microsoft's operating system adds Windows, Microsoft is well on its way toward becoming an industry standard, says the company's Ballmer. 'CHURCH AND STATE.' On the other hand VisiCorp is stressing its success in appli cations programs. The San Jose (Califcompany, which was made famous by the VisiCalc financial modeling and spreadsheet program, has developed a set of applications for its VisiOn environment package. "What will make a windowing system successful is the quality of the applications under it," says Danie Fylstra, VisiCorp chairman.

While archrival Digital Research has not formally announced its environment package, the company is working hard to line up applications software companies to write programs for use with its new windowing software. Digital Re search says its environment package is the safest choice, because the company does not write applications software. "We're not in the applications business like VisiCorp or Microsoft," says Digital Research President John R. Rowley. "We do not present a threat."

Rowley contends that competitors can use their own environment software to bring out applications packages before anyone else can. Microsoft, for one, denies that offering an environment program gives its own applications programs an advantage. "We have shown in the past that there i8 a very clean separation between our operating system business and our applications software," says Ballmer. "It's like the separation between church and state. Axed if you don't play it straight, you can't expect to get the business." Microsoft is expected to hit sales of $70 million this year, double its figure for 1982.

Some applications software writers are concerned, however, that Rowley may be right. "VisiCorp is a competitor," says Fred M. Gibbons, president of Software Publishing Corp., which sells personal computer applications software. "Why should I trust them?" To cover their bets, some powerful independents could decide to go with more than one environment package. "There are many valid reasons to support more than one environment and let the marketplace decide," says Mitchell D. Kapor, president of Lotus Development Corp., which for now has gone with Microsoft for 1-2-3, its popular integrated spreadsheet and graphics package.

One variation on the environment theme is Quarterdeck Office Systems, a small Santa Monica (Calif.) startup. By the end of the year, Quarterdeck will begin shipping a $399 environment package called DESQ. But instead of persuading software writers to modify their programs, the company has designed DESQ for use with several existing applications programs. "With DESQ you just buy it and run it totally without having anything modified," says Therese E. Myers, president and founder.

PREEMINENT POSITION. As the battle begins to heat up, no company has produced the supporter that could carry the day: IBM. "IBM has established such a preeminent position in the marketplace that the supplier that has its environment on IBM will have the greatest success," says John R. Keifer, senior analyst at InfoCorp. Since the IBM PC was introduced in 1981, the computer giant has won more than 26% of the market. As many as 75% of personal computers, industry observers agree, are expected to follow the IBM PC design by 1985 (BW -- Oct. 3).

Few are willing to predict IBM's strategies in this key software market. "I don't expect IBM to endorse one environment in the near term," says Rosen Research's Dyson. "It will probably make them all available." But some observers say IBM will bring out its own environment software, and it is not clear where such a move would leave the independent software companies. IBM already has shown some of its own windowing software on an enhanced version of the PC, and there are reports of another IBM environment program, called Glass, that is being considered as a product. "With IBM's announcement of its own windowing capability, it looks to us that the big guy is starting his own standard," says Dennis V. Vohs, executive vice-president of Management Science America Inc., which owns Peachtree Software Inc., a personal computer software company.

The victors in the battle over environment software may not be obvious for a year or more. Despite an impressive array of endorsements, Microsoft will not begin shipping Windows until April. At that time, VisiCorp and Digital Research will have had their products on the market for only a few months. "There will be a lot of bandstanding and claiming victory," says Digital research's Row- ley. "But you won't really know what will happen for at least 6 to 12 months."

GRAPHIC: Picture, MICROSOFT'S BALLMER AND A DISPLAY DIVIDED INTO "WINDOWS"

LANGUAGE: ENGLISH

TO

APPENDIX TO MEMORANDUM OF AMICI CURIAE
IN OPPOSITION TO PROPOSED FINAL JUDGMENT
IN CIVIL ACTION NO. 94-1564 (SS)
SIGNED BY GARY REBACK

'MICROSOFT IS LIKE AN ELEPHANT ROLLING AROUND, SQUASHING ANTS'
As the company's dominance grows, so have the complaints of other suppliers of software.

There's no doubt about it. Microsoft Corp. wants to dominate the world--the personal computer software world, that is. And it isn't very far from doing so: It already supplies the core software for just about all of the world's 25 million-plus IBM PCs and their clones. It has done well, too, in many sectors of the huge market for PC applications programs--spreadsheets, word processors, and the like. All in all, it's the leader in total PC software sales--Wall Street expects revenues of $1.1 billion for the year ending next June, up 40% from the year before.

Now, Microsoft is beginning to suffer the slings and arrows that often come with such fortune. Other suppliers of PC software are downright angry over its dominance. The company, they say, is just too powerful and its products too pervasive. Its virtual monopoly-in PC operating systems--the software life-support systems that all other programs call upon for access to the PCs memory, disk drives, and display screen--means that Microsoft's even' technical change, strategy shift, or mistake can adversely affect, producers of applications software. They argue, moreover, that Microsoft is abusing its systems software edge to put them at a disadvantage-- and win greater control of the market.

INTIMATE TIES. This, critics say, will make it harder for Microsoft's small competitors to prosper. And that hints at less innovation in software, the one part of the world computer market in which U.S. companies still hold an unassailable edge. Says Fred M. Gibbons, president of $100 million-plus Software Publishing Corp.: "Microsoft is like an elephant rolling around, squashing ants."

William H. Gates III, Microsoft's CEO, argues that such fears are misplaced. He contends that his company is so influential simply because it knows more than any other about how the pieces of a PC fit together, from chips to other components to software. Microsoft's inti??s with leading companies such as IBM, Compaq, and Intel bode well for the U.S. computer industry, he argues. By virtue of those relationships, Microsoft can establish coherent technical standards--in graphics, communications, or computer languages, for instance-- that if followed by everyone would speed up the process of writing new programs. Those would help sell machines, fulfilling Gates's vision of a PC on every desk and in every home.

What worries other software makers is where they fit into this vision. While tightening its grip on the $1.4 billion systems software market, where its MS-DOS and OS/2 operating systems are king, Microsoft has pushed harder than ever into the $4.4 billion market for applications packages. Its Microsoft Word text- processing program, Excel spreadsheet, and other such products now account for 47% of total revenues--almost equal to its systems business. And competitors say they're getting squeezed.

Recently, for example, Microsoft stopped providing them with lists of customers that use Windows, its graphical extension to MS-DOS. Instead, it offered to place ads for their Windows-compatible software in a booklet shipped with each copy of Windrows. Competitors suspected that Microsoft's own applications group was still getting the lists. So they complained--and got the lists back.

VOCAL CRITIC. More unsettling are suspicions that Microsoft doesn't keep its systerns and applications groups as separate as it promises--that church and state tend to mingle. Competitors figure that if Microsoft's applications people' get peeks at unanmounced systems software, they should, too. Otherwise, they're at a disadvantage. Microsoft' fuels suspicions by sometimes-shifting workers between its groups.

[Photo of Maples, Gates, and Ballmer. Caption: Does Microsoft Use Its Monopoly Unfairly?]

And it Agenda 90, a recent trade conference outsiders were angered to see an Excel specialist demonstrating new operating system features that they hadn't been briefed on. Apple Computer Inc. solved such conflicts in 1987 by spinning off its applications group into an independent company, called Claris Corp. Gates says that's not necessary at Microsoft.

 [D]

Micrografx, a tiny graphics software company, might disagree. Recently, it approached Microsoft with a program it thought the larger company might want to use. But it showed it only to Microsoft's applications developers--not to its systems people, who it feared would copy its proprietary ideas. Micrografx President J. Paul Grayson says that one person who saw his program was soon transferred to Microsoft's systems division. Eventually, Gates placated Grayson with a cross-licensing deal, which Microsoft concedes was unusually generous. Still, Grayson says he was "manipulated by Microsoft," which insists it did nothing wrong.

[Photo of unidentified man]

PAINFUL P.S. Whatever the case, Microsoft's tactics have strained relations even with partners. This fall, John Warnock, chief executive of Adobe Systems Inc., had an emotional, public falling out with Gates. Adobe's top product called Postscript, is a key program for desktop publishing. Earlier this year, Apple, Adohe's best customer, said it would replace Postscript in Apple computers. Microsoft continued to do business with Adobe. Then, in September, Apple and Microsoft surprised Warnock by announcing at an industry conference that they would collaborate in competing with Adobe. Says Warnock: "We used to be a strong ally of Microsoft." Now, "it's easier to help their competitors."

The biggest gripes have been with Microsoft's moves in operating systems. Like Microsoft, its competitors use those basic programs as "platforms" upon which to construct applications software. But if the platform is shaky, late to market, or just not selling well, writing software for it can be risky--as the tale of Windows shows.

Starting in early 1983, Microsoft tried to supplement MS-DOS with Windows, a program that makes PCs act much like Apple's Macintosh. But outside developera were wary of writing programs for Windows, which was 16 months late to market, because of its many early technical problems. They say Microsoft also gave them mixed signals: It positioned Windows as a program mainly for low-end PCS, while it worked on a more ad- vanced--but incompatible--operating system called OS/2 for more powerful computers. And IBM threw its weight behind OS/2.

Much to the industry's surprise, however, OS/2 has caught on slowly. And Windows has taken off. Microsoft has shipped 2 million copies of it, compared with only 150,000 of OS/2. And next year, it will bring out a major revision of Windows that will be easier to program and more functional than the original-- enough so, in fact, to do many of the same jobs that OS/2 was supposed to handle. Windows, says David G. Bayer, an analyst at Montgomery Securities, "has become the platform of choice."

DUPLICITOUS? Guess which company is poised to exploit that platform? While most competitors concentrated on writing for OS/2, Microsoft has been readying a slew of applications for Windows as well. They include a fancy new word processor, a project management program, and a long-rumored dam-base program called Omega. That's leading com- panies such as Lotus Development and Software Publishing to call Microsoft duplicitous. They charge that Microsoft enhanced Windows just to help its own applications group. And, they claim, the more powerful Windows will further hurt OS/2 "It's irresponsible of Microsoft to do that," says Software Publish ing's Gibbons.

Even discounting the effect of a re vived Windows, Microsoft has disappointed those counting on OS/2. Introduced in 1987, that program still can't do all it promised, such as use all the power of Intel Corp.'s popular 80386 chip Worse, perhaps, is that Microsoft still offers no aids for modifying Windows programs to work with OS/2. A recent poll shows that software executives don't expect OS/2 to really catch on until 1993--two years later than what they predicted last year. Gates's answer: Microsoft is devoting the maximum feasi- ble engineering talent to OS/2 and Windows, favoring neither.

'SLIDEWAR' On top of all this are wilder accusations--for instance, that Microsoft peddles nonexistent products to scare off competition. Michael J. Maples, the company's vice-president of applications software, shows slides at trade shows that list the software markets Microsoft intends to enter--programs for desktop presentations, for instance. One competitor calls that "slideware. They have slides saying they're going to be involved in every conceivable area of innovation five years from now," he says. "It slows the pace of innovation" by intimidating smaller competitors.

Gates laughs off the idea of software companies quaking in their boots. "So what are they doing instead, starting fast-food restaurants?" he quips. "I've never heard anyone say, 'we're chicken, we can't compete with you.'" WordPer- fect Corp., for example, is beating Microsoft in word processing, with a 40% share of the market, up from 16% three years ago. And companies such as Micrografx and Atlanta-based Samna Corp. have drawn technical praise for their applications programs for Windows.

In fact, many of Microsoft's critics helped create their own problems when they ignored its pleas to develop applications for Windows. "Even when Gates makes a mistake, people turn it into a Machiavellian plot," says Gordon E. Eubanks Jr., president of software house Symantec Corp. And Steven A. Ballmer, senior vice-president for Microsoft's systems division, disputes the charge that his people give their counterparts in applications previews of their upcoming systems products.

Since Microsoft earns more from systems than from applications programs, Ballmer says, he would be foolish to jeopardize his market just to boost applications sales. Indeed, he recounts an oc- casion when Microsoft's developers of Excel accosted him in the company cafeteria for revealing their work to Lotus, which confers often with Microsoft on changes in its operating systems. "Telling me is as good as telling Lotus," he says, as if to prove his independence.

So, the tension mounts. But what can Microsoft's rivals do? Their dependence on its PC operating systems puts them at a disadvantage. But no company--not even ISM--has been able to avoid that. They might try to subvert Microsoft's efforts to win control over every critical software standard in the PC market. "If people are feeling mishandled, they're going to look for other [partners]," warns Lotus CEO Jim P. Manzi. A likely one would be the group of suppliers backing American Telephone & Telegraph Co.'s Unix operating system, which rivals OS/2 in scope and function.

But Unix' base of existing customers is minuscule compared with MS-DOS's. And Microsoft already has the best-selling version of Unix for personal computers, called Xenix. Perhaps, for competitots, there's just one choice: Learn to dance with the elephant.

By Richard Brandt in San Francisco

TO

APPENDIX TO MEMORANDUM OF AMICI CURIAE
IN OPPOSITION TO PROPOSED FINAL JUDGMENT
IN CIVIL ACTION NO. 94-1564 (SS)
SIGNED BY GARY REBACK

Special Report

IS MICROSOFT TOO POWERFUL?
HOW THE INDUSTRY'S LEADER IS WIELDING ITS CLOUT

[Photo of Bill Gates, hands behind head]

It's a chilly November night in Las Vegas, but 10,000 technoids are in full fever pitch. They're in town for 1992's Fall Comdex, the computer world's biggest convention cum celebration. Tonight is the annual Chili Cook Off, a charity event for the National Center for Missing & Exploited Children. Each year, the crowd pours in for kegs of beer, vats of chili, and live music. For one night, archrivals in the industry are expected to put aside their bitter feuds and just goof off.

But not this year. The Grayson brothers, Paul and George, founders of software house Micrografx Inc. and organizers of the event, are thanking com- panies that ponied up money. Each gets a round of applause. That is, until one of the hosts offers "a special thanks to Bill Gates and Microsoft," donors of $30,000. The crowd's reaction: scattered cheers, drowned out by a round of boos.

BIG GREEN. The fear and loathing on display in Las Vegas--as well as envy and a grudging respect--are the natural responses to Microsoft Corp. these days. Long a power in personal computer software, Microsoft has now emerged as clearly the most important single force in the entire computer industry. Where Microsoft leads, computer makers and customers follow. Where it stakes a claim, rivals steer clear. And as it springboards from its dominance in operating systems into a commanding position in applications programs, Microsoft leaves less and less territory for its software rivals. Many venture capitalists these days say they won't consider funding a software startup that looks like it might wind up competing on Microsoft's expanding turf.

Such a concentration of clout and power has not been seen in the computer industry since the glory days of IBM. Even Intel Corp., whose microprocessors are as pervasive as Microsoft's software, does not have the leverage of Microsoft, in part because Intel now must respond to chip clones (page 8652). Some software executives refer to Microsoft, headquartered amid the evergreen trees of Redmond, Wash., as "Big Green." Says Alan K. McAdams, the chief economist in the Justice Dept.'s fruitless antitrust suit against IBM in the 1970s: "It sure sounds familiar. Microsoft is using its power in ways that are just like IBM's."

But does that mean Microsoft is too powerful? Does its dominance really inhibit competition in the software market, and does it hamper advancement of the computer industry itself? And, perhaps most worrisome, will it ultimately lead to fewer competitors and less innovation in an industry founded on the latest, the greatest, and never-before-thought-of? Those questions are critical because computer software has become one of the driving forces in the economy. Not only is the software industry a key area for job creation, but it also produces the tools other industries need to boost productivity. Is such a vital industry best served by having a single dominant company?

FTC PROBE. Microsoft's competitors answer no. Software rivals insist that Microsoft's hyperaggressiveness--its use of every trick at its disposal to gain an edge, enter a new segment, or eke out one more iota of market share--has started to edge out innovation itself as the force that determines the shape of the industry. Microsoft Chairman William H. Gates III says such charges are ridiculous. "Our success is based on only one thing: good products. It's not very complicated," he says. "We're not powerful enough to cause products that are not excellent to sell well." Still, complaints from other software makers helped spur a 2 1/2-year investigation by the Federal Trade Commission into Microsoft's tactics. FTC sources say the nonpublic probe was completed at the close of 1992 and focused on allegedly unfair tactics used to squelch competition (table).

 [D]

According to a confidential outline obtained by BUSINESS WEEK, the FTC investigated practices ranging from the way Microsoft prices software to the way it allegedly uses tying arrangements to force customers who want one Microsoft product to also buy others. Sources close to the investigation say that FTC staffers recommended a number of actions, including a preliminary court injunction, ordering Microsoft to cease the offending practices immediately, pending the outcome of the case.

NECESSARY EVIL? That they would even contemplate such an injunction--rather than wait for the outcome of a commission proceeding--is an indication of how serious the situation appears to the FTC staff, says Terry Calvani, a former FTC commissioner. "The reason the staff went into this uncharted area was the concern that there are companies in business today that may no longer be" by the time the FTC could finish trying a case against Microsoft, he says. But an injunction was only one staff recommendation among many and, so far, the FTC commissioners have not acted. On Feb. 5, they considered the recommendations and split 2-2 on what action, if any, to take. They are expected to meet again in a few weeks, but Calvani says the tie does not bode well for competitors who were hoping to see dramatic action.

Even if the FTC does nothing, the dominance of Microsoft will remain a maelstrom of controversy. Interviews with more than 60 industry executives and customers and a review of still secret FTC documents point to one overriding concern: Microsoft's methods and its growing control over the computer industry could choke the life out of any company that stands in its way. Steven P. Jobs, chairman of NeXT Computer Inc. and an outspoken critic of Microsoft, has publicly called for the breakup of Microsoft into two companies: one for operating systems and one for applications programs. That move--considered, then rejected by the FTC staff--would keep Microsoft from using its operating-systems business to give its applications business an extra edge, as now alleged.

For the most part, customers can't see what all the fuss is about: Most seem happy with what they're getting and with what they're paying for it. And even if computer makers grouse about how much influence Microsoft now exerts over their business plans, they concede that the standards Microsoft sets are helping to keep their industry vibrant. Says an executive with a top-tier PC maker: "Microsoft is not just a necessary evil at this point. It's necessary for the industry to proceed."

For many customers, Big Green has already taken on the role that had been Big Blue's. The saying among computer managers used to be: "Nobody ever got fired for buying IBM." Now, says the information-technology manager of a major French manufacturer: "If you put all your marbles in the Microsoft hat, you're safe--like the old IBM."

Even Gates, who pooh-poohs comparisons with the mighty IBM of the 1970s, agrees that his company has partially taken on the leadership role Big Blue has lost. "Who's there to fill that vacuum? Microsoft, more than anyone else," he says. Adds Roger McNamee, a partner in technology investors Integral Capital Partners: "Microsoft has been anointed the industry tsar. When that happens, people make it very, very rich."

[Photo of person looking at a document while standing by a server. Caption: Big Blue Meets Big Green. As IBM ruled the 1970s with its mainframe, Microsoft dominates today with its operating system software. Microsoft used IBM's own tactic against it: By "preannouncing" Windows NT, it stalled sales of Big Blue's OS/2, Version 2]

WINDOWS AND ORPHANS. Rich indeed. Microsoft's MS-DOS operating system is used by 81% of the 22 million IBM-compatible PCs built every year, according to Sanford C. Bernstein & Co. Microsoft Windows, which gives MS-DOS a graphical "look and feel," is selling at the rate of 1 million copies a month. And because it has been first to market with top-notch applications packages for Windows, Microsoft is now the king of that white-hot growth segment. Lotus Development Corp., the king of spreadsheets in the MS-DOS world, has just 20% of the $756 million Windows spreadsheet market, while Microsoft's Excel now claims 73%, says market researcher Dataquest Inc. In word processing, the MS-DOS leader, Wordperfect, has 31% of the Windows market, compared with 53% for Microsoft Word.

In short, Microsoft is cleaning up big time--at the expense of its smaller rivals. While other software makers were announcing shrinking market share, losses, or layoffs in 1992, Microsoft tacked on $975 million in calendar-year revenues--more than 90% of all the revenue growth in the PC software industry, according to preliminary Dataquest figures. Microsoft's share of the world desktop PC software industry reached 44% last year, Dataquest figures. And if, as analysts project, Microsoft sales rise 36%, to $3.75 billion, in the fiscal year ending June 30, Microsoft will have more revenues than its seven closest publicly held rivals combined. And at nearly $1 billion, it will have more than twice their net income (chart).

All that money, rivals fear, will soon translate into even greater power for Microsoft. Without healthy profits, other software makers may find it impossible to fund new development or finance upgrades of complex programs such as data bases, which comprise millions of lines of code. Borland International Inc. Chairman Philippe Kahn blamed pressure from Microsoft's foray into Borland's data-base turf when he laid off 15% of his 2,200 workers in December. Borland then reported a $61.3 million loss for the quarter and put on the back burner a word processing project that had been two years in development. Gates says Borland suffered mainly because its products were late to market.

Lotus, once No. 1 in PC applications programs, had its first-ever layoffs in 1992. Now, it's concentrating its resources where Microsoft isn't--yet: Programs such as Notes, which helps groups of workers collaborate.

'TOTAL UNDERDOG.' Such a sharp contrast between one have and many have-nots worries industry executives. They fear there will be few major players, more consolidation, and less money for everybody except Microsoft. They also warn of a chill on software startups. John M. Grillos, who manages technology investing for Robertson Stephens' venture-capital arm, says that there are still new opportunities for startups and scores are on the drawing boards--in promising new areas such as multimedia. But he has a long list of phone numbers at Microsoft and checks the behemoth's plans before going ahead with an investment. Does he call very often? "You bet," he says. "I'm not crazy."

Gates, the billionaire mastermind of the Microsoft empire, says such worries are nonsense. Is Microsoft too powerful? "The answer is simply no," he says. He points out that Microsoft still lags in some important markets. "Take networking. We're the total underdog." And, he asserts, in markets such as spreadsheets and word processing, Microsoft's presence has prodded the competition to improve their wares.

Gates also points out that his commanding position does not guarantee him success in the next generation of software: operating systems that will let networks of personal computers take on the big computing jobs now done by mainframes, minicomputers, and workstations. Microsoft's entry, Windows NT, will square off with Novell's UnixWare, Sun Microsystems' Solaris, IBM's OS/2, and NeXT's NextStep.

Still, none of those competitors has the momentum that Microsoft gets from Windows. That should help Gates reach his stated goal of selling 1 million copies of NT the first year. But he insists that doesn't mean NT is already the winner. "This is a hypercompetitive market," Gates says. "Scale is not all positive in this business. Cleverness is the positive in this business."

To be sure, competitors such as Lotus and Borland have contributed to the myth of Microsoft's invincibility through their own less than clever moves. Equally true, there are examples of software companies that have kept well ahead of Microsoft. Many, such as Intuit Inc., a maker of personal finance software, are masters of lucrative niches (page 8854).

The biggest player to successfully fend off Microsoft so far has been Novell Inc., the $933 million Provo (Utah) maker of networking software. But Microsoft is aiming for this key software market by building some features similar to Novell's NetWare into Windows NT. Says Kanwal S. Rekhi, a Novell executive vice-president: "Microsoft will keep us on our toes." Then, half-jokingly, he adds: "I hope they don't kill us."Novell can afford to joke. For now, it still holds 70% of its market. But the rest of the industry isn't laughing. Rival software companies give Microsoft credit for building good products and marketing them cleverly. But many software executives also are fuming about what they say are Microsoft's unnecessarily tough, sometimes downright mean-spirited tactics. Says the CEO of a rival software company: "If you were in my shoes, you would probably want to go and shoot them. It's not a level playing field. IBM was the most opportunistic and ruthless in the 1970s. And that's exactly what Microsoft is today."

VAPOR TIGERS. Indeed, industry veterans say there's a striking parallel between how Big Blue behaved back then and how Microsoft acts now. Computer executives say that just like the IBM of yore, Big Green bullies partners, withholds vital information, disparages competitors, and stalls the market by announcing products long before they're ready. Microsoft denies such charges. While such tactics are in the playbooks of many competitors, in the hands of the richest and most powerful player, they can be lethal.

Take IBM's classic move of announcing a product long before it was ready to ship--a tactic known as "preannouncing." In software, such products are called "vaporware" and no one pays much attention--unless the company promoting vapor holds a dominant position. In that case, the market freezes. Facing upstart Control Data Corp. in the 1960s, IBM paralyzed the market for scientific mainframes by announcing it was working on machines that would be far faster than CDC's. These paper tigers, as they came to be known in a subsequent antitrust trial, prevented CDC from winning a single order in 18 months.

Microsoft preannouncements now have a similar effect. Take the case of Adobe Systems Inc., maker of software that controls how computer printers produce typefaces. In September, 1989, Microsoft and Apple Computer Inc. said they would jointly develop a rival product. Adobe's stock fell 20% in one day, and for the next nine months the company spent 90% of its time answering customer's questions and "fighting vaporware," says Chairman John E. Warnock. As it turned out, Apple backed off and Microsoft did not ship its competing product, TrueImage, for two years.

Microsoft has turned this Big Blue weapon on IBM itself. Just as IBM was getting OS/2 Version 2.0 off the ground in mid-1991, Microsoft announced plans for Windows NT. Like the IBM product, NT would be a 32-bit operating system, meaning that it would tap all the powers of Intel's fastest chips. Customers could buy the 32-bit system from IBM then or wait at least 18 months for NT.

POWER PITCH. Guess what? Most of the market is waiting for the leader. An executive at a top PC company tells of one customer that felt the squeeze after committing to buy 36,000 copies of OS/2. The way the exec tells it, Microsoft came and pitched NT, and the buyer put the OS/2 order on hold. "It used to be IBM could put orders on hold," says the executive. "Now it happens with Microsoft."

And NT? It's the toast of the tech world even though it's still not ready. After a six-month delay, it's now scheduled for shipment by June--two years after it was announced. It could be a big hit. Even Borland is developing software for it. Says CEO Kahn: "There's no choice. The issue is not whether NT is good or bad. The issue is NT is being pushed by Microsoft."

And Microsoft is already talking about an operating system beyond NT. It's called Cairo, and it's due by 1995. The company says that package will match features of Novell's most advanced networking programs and the object-oriented programming features of NextStep and Pink, the operating system due by 1995 from Taligent, the joint venture between IBM and Apple.

F.U.D. MISSILES. Gates says Microsoft preannounces systems software because customers and outside developers need details to plan ahead. And once Microsoft tells developers, word spreads fast. "We tell 100 developers," Gates says. "And believe me, that is out in the press the next day."

Whatever the legitimate purpose, preannouncing is part of a larger strategy computer makers say IBM used effectively for years. It's called F. U. D.--for fear, uncertainty, and doubt--and it really works only for the big guy. It's essentially a whispering campaign suggesting it would be terribly unsafe to bet on a competitor. Gates snorts at the notion Microsoft uses F. U. D. as a weapon. "We have a whole department in charge of F. U. D.," he jokes. Seriously, he adds that Microsoft simply gives its opinions and expects customers to judge for themselves. "We're giving our honest view of how wise it is to buy these products," he says.

Where any discussion of Microsoft's power gets dead serious is when rivals--and the FTC--consider the power stemming from Microsoft's dominance in operating software. Like IBM, whose aggressive tactics for preserving its dominance in mainframes led to the Justice Dept.'s 1969 antitrust suit, Microsoft seems most bare-knuckled when perpetuating its position in operating sytems.

Microsoft's most controversial tactic is a "per-processor" discount plan for MS-DOS, which it offers to the highest-volume PC makers. On average, PC makers pay $13 to $14 per copy. For the steepest discounts, the PC maker must agree to pay for a copy of MS-DOS for each PC it ships, whether or not the software is actually installed. That makes it "undesirable for a manufacturer to ship anything but MS-DOS," says a PC executive. Microsoft says that PC makers are offered a number of ways to buy MS-DOS. But with other plans the discounts are smaller, and PC makers locked in a bloody price war can ill afford to pass up the steepest discounts.

DOS & DON'TS. When pricing isn't inducement enough, Microsoft allegedly uses other means. One PC maker says it told Microsoft that it planned to ship DR-DOS, Novell's clone of MS-DOS, on about 10% of its machines. By shipping MS-DOS on 90% of its PCs, the company figured it would still get the best discount. Microsoft's response: It doubled that customer's price on MS-DOS, which quickly forced the PC maker to drop the idea of offering a choice to customers. Says a company executive: "In my opinion, any monopoly situation is not good for the customer." A senior Microsoft executive says he wasn't aware of this charge but says it would not be common practice.

Such alleged tactics may seem a tad over the top, but maintaining dominance in PC operating systems is critical. Like IBM's dominance in mainframes, it gives Microsoft an extremely reliable, enormously profitable revenue stream. "Microsoft's mainframe is its operating system," says one software executive. Analysts estimate that between 1989 and 1992, MS-DOS and Windows generated revenues of $2.3 billion, with $998 million of that in 1992 alone. Net profits on those sales last year were $278 million, according to Sanford C. Bernstein & Co. Such profits have helped fund forays into almost every major software market. Microsoft's new data-base program, Access, cost a staggering $60 million to develop--and it was just one of a dozen products Microsoft brought to market last year. By contrast, last year's entire R&D budget at Borland was $50 million. At Lotus, it was $35 million.

That's not all. Microsoft also had the money to offer an introductory price of $99 for Access--less than one-third the retail price for similar packages. Result: Microsoft sold 700,000 copies in just three months. The entire market in 1992 was only 1.2 million units.

Gates shrugs off the notion that operating systems are his cash cow. "That's the biggest joke I ever heard," he snaps and points out that products such as Word and Excel are his most profitable. Yet in the next sentence, as he elaborates on the returns from operating systems, he says: "If you just took the cash cow business and did not factor in the development costs of NT and Cairo, yes, you'd get a huge profitability."

Gates is accurate when he points out that his applications business now generates more profits--about 50% of net income--than operating software. But it took years to reach that point--years during which Microsoft funded many versions of Word before it was good enough to grab substantial market share. Only when the Windows 3.0 version appeared, in 1990, did it take off.

The operating system business does more than spin profits. Competitors charge that because Microsoft writes operating systems, it also has an unfair edge in writing the applications programs that work with them. They say Microsoft's applications developers get a peek at the inner workings of new operating systems early so they can write programs to take advantage of new features first. In the FTC document, investigators referred to this as Microsoft's "fake Chinese Wall" and listed a dozen other ways Microsoft allegedly abuses its position. Microsoft denies any unfair crossover or inside knowledge.

Software developers also complain that Microsoft is slow or even reluctant to deliver needed information about operating systems. Perhaps the most ironic such charge comes from Claris Corp., Apple's software subsidiary. Executives there say they tried for a year to get information for writing Windows applications from Microsoft, to no avail. Claris says Microsoft was worried there were cracks in the Chinese Wall between Claris and Apple's operating system team--just what rivals say occurs at Microsoft. But after executive meetings and assurances of no cracks, the situation was resolved. Microsoft's head of developer relations says he wasn't aware of the Claris problem but does concede a general "concern about giving information to our operating system competitors."

Microsoft says it's doing its best to get information out to thousands of companies and that it doesn't withhold information to favor itself. Says Pat Bellamah, a manager in Microsoft's developer group: "It's ironic to us that people feel they're having a hard time getting information when that's all we're putting out there." Gates estimates Microsoft spends $80 million a year disseminating information to developers.

One reason Microsoft draws so much criticism is simply that wherever it competes, it seems to play a particularly hard-core game of hardball. Take its dealings with Logitech Inc. Until last June, Logitech had a license to buy Microsoft Windows 3.0 at a discount, then sell it together with Logitech's mice. But Microsoft abruptly canceled the deal, saying that it was losing money on such "bundles" involving inexpensive hardware, according to Logitech President Pierluigi Zappacosta. Only Microsoft still continued to sell Windows bundled with its own mice--for about $10 more than Logitech had been charging.After Zappacosta publicized his situa- tion in September, Microsoft relented. But there was a catch: The new license fee would be 30% higher. Zappacosta says that priced him out of the market, depriving his company of about $20 million annually. Microsoft continues to sell its Windows-and-mouse bundle. Says Zappacosta: "Microsoft is extremely aggressive in using everything it can to its advantage." Microsoft denies that it forced Logitech out of the market but declines to discuss its pricing.

[Photo of Philippe Kahn. Caption: "Microsoft is the IBM of the '90s and uses exactly the same marketing tactics IBM used to", Philippe Kahn, Borland International]

STAC ATTACK. Occasionally, Microsoft's hardball tactics have resulted in civil suits. The latest was filed in January by Stac Electronics, a maker of data-compression software. In its suit, Stac claims that Microsoft violated its patent by including Stac's technology in test versions of MS-DOS 6.0 without permission. Stac says it was negotiating with Microsoft to license the technology, but talks broke down when Microsoft did not offer a sufficient royalty. The suit claims that Microsoft executives then showed Stac a spreadsheet, detailing the "adverse impact on sales of Stacker" if Microsoft opted for another company's technology. Microsoft denies the claim, saying it bargained in good faith and offered "real money" for a license.

As the stories multiply, it also becomes clear that Microsoft long ago became everybody's favorite whipping boy. There's certainly resentment on the part of bright young software entrepreneurs who may never see millions, much less Gates's billions. And for all the companies that grouse about their dealings with the industry giant, there are dozens that are ardent admirers. Says Morton H. Rosenthal, CEO of software distributor Corporate Software: "We all live in a Microsoft-centric world. Working with Microsoft is like skiing behind the Queen Mary. It's a good ride. But getting up is a little rocky."

Indeed, with Big Blue's waning influence, there's a genuine need for a leader. Customers want good software and good prices. They also want a relationship with a software maker that's going to be around for the long run. They uant a new IBM. "If I were a software company, I'd be complaining about Microsoft, too," says Greg Chetel, director of systems planning and research at Gillette Co. "But I don't care who wins. I just want quality products."

In the end, that may be the key to assessing whether Microsoft does indeed have too much power. Software makers are right to cry foul when they think Microsoft's practices have been anticompetitive. They have done so, and the FTC has listened. But as long as Microsoft's dominance stems from keeping customers like Gillette satisfied, it is hard to argue that its power, per se, is harmful.

The danger is that Microsoft will start to use the power of its position, rather than the appeal of its products and services, to stay on top. "If Microsoft runs out of bandwidth," says McNamee of Integral Partners, "then there will be a problem." That's when there will be reason to fear that competition will be stifled and innovation squelched.

If the history of Big Blue is a guide, Microsoft's dominance will be in danger of waning long before it can distort the market with nefarious practices. When the Justice Dept. began its antitrust suit in 1969, IBM's hold on the mainframe market made it seem invincible. By the time federal prosecutors withdrew their suit in 1982, however, the market had taken care of the problem: New technologies such as minicomputers and PCs had made IBM's near-monopoly in mainframes largely irrelevant.

History could repeat itself: Says Joe Guglielmi, a former IBM executive, now CEO of Taligent: "Today, everyone is in fear of Microsoft." "But in the end, everyone will compete. There are thousands of Bill Gateses out there who will find pieces of this market and win them." Just the way Microsoft won its place in the sun.

WHY ALL THE FUSS? THE POINTS OF CONTENTION BETWEEN MICROSOFT AND COMPETITORS

EARLY PRODUCT ANNOUNCEMENTS

CHARGE To preempt competing products, rivals say, Microsoft sometimes announces products years before they actually exist. Even if a rival's product already has the features' that Microsoft promises, many customers are reluctant to buy it, preferring instead to wait for the "safe choice"--Microsoft

RESPONSE Microsoft says it is important to let outside software developers know Microsoft's directions in system software so they can develop application programs. In fact, software developers demand it. And. Microsoft says. it is important to let customers know where it's headed so they can plan accordingly

INSIDE KNOWLEDGE

CHARGE Makers of applications programs allege that Microsoft's applications programmers have advance details of its operating system software, and the company is slow to share vital information. They say Microsoft uses this edge to bring out better applications sooner. This, rivals complain, is a big reason Microsoft has more than 60% of the market for programs that work with Windows

RESPONSE Microsoft says it freely shares its knowledge with the industry and enjoys no substantial advantage in developing applications that work with its operating systems. The company says its software sells well because it's good

THE DOS TAX

CHARGE Rivals say that the way Microsoft licenses MS-DOS and Windows to major PC manufacturers makes it nearly impossible for them to compete. Under some Microsoft licensing contracts, PC makers pay a fee to Microsoft for every PC they ship, even if they don't install the Microsoft software on each machine. Because of this, PC makers are less likely to substitute a competing operating system

RESPONSE Microsoft says PC makers [? word unreadable] can, and do, choose several different ways to license MS-DOS. The cotroversial "per processor" licensing arrangement offers a lower price for higher volume

PRICING

CHARGE Microsoft can offer low-ball prices in two ways: by including extra programs with its operating systems and by using profits from operating system sales to support low pricing of applications programs. For instance, because it has not made much headway so far. against Novell in sales of networking software, Microsoft is now building networking into Windows and MS-DOS

RESPONSE Microsoft says it is an industrywide trend that. as operating system software is improved, more features, such as networking, communications and graphics, are included to make computing more seamless for customers

THE F.U.D. FACTOR

CHARGE As the dominant force in PC software, Microsoft uses its unique position to spread "fear. uncertainty, and doubt" about its rivals to stop customers from buying rival products. Microsoft, competitors say, warns buyers that if they buy IBM's OS/2 or Novell's DR-DOS-- both of which claim advantages over Microsoft's operating systems---they will be throwing away their money because those products may wind up incompatible with Windows or may not be around in a few years

RESPONSE Microsoft says customers ask for advice on many products, and, when it comments it is just responding to questions

BRAIN-PICKING

CHARGE Several companies charge that Microsoft has, in effect, stolen their ideas in the course of exploring collaborative agreements. Go Corp., for example, says that Microsoft expressed interest in writing applications for Go's operating system for pen-based computers. After Microsoft programmers examined Go's technology, however, Microsoft said it was no longer interested, Go says. Then, Microsoft announced plans for a competing system, developed, in part, by those who visited Go

RESPONSE Microsoft says it is upsetting that companies accuse it or imply it stole from them. Microsoft says it always honors nondisclosure pacts

By Kathy Rebello in Redmond, Wash., with Evan I. Schwartz and John W. Verity in New York, Mark Lewyn in Washington, Jonathan Levine in Paris, and bureau?? reports

For information on reprints of this Special Report, call Business Week Reprints of 609 426-5494, or write Business Week Reprint, P.O. Box 457, Hightstown, N.J. 08520.

TO

APPENDIX TO MEMORANDUM OF AMICI CURIAE
IN OPPOSITION TO PROPOSED FINAL JUDGMENT
IN CIVIL ACTION NO. 94-1564 (SS)
SIGNED BY GARY REBACK

NO SLACK FOR MICROSOFTS RIVALS
They complain it hasn't been reined in at all by Justice

[Photo of Bill Gates. Caption: Microsoft Chairman Bill Gates]

When Microsoft Corp. signed a consent decree in July with U.S. Justice Dept. trust-busters, it emerged virtually unscathed from the feds' five-year probe. Still, the investigation was a protracted--and expensive--headache for Chairman William H. Gates III. And the settlement banned some of Microsoft's most aggressive licensing practices. The experience, rival executives figured, surely would leave Microsoft chastened.

No such luck. "The consent decree seems to have set [Microsoft] free," gripes Robert J. Frankenberg, chief executive of Microsoft rival Novell Inc. "Now, they are running rampant over everything." There is little doubt that Microsoft is competing aggressively: Even while the software giant presses its market-share advantages in operating systems and applications programs, it is bolting into new consumer markets with its own on-line service and a plan to buy Intuit Inc., the top maker of personal-finance software. The $1.5 billion deal requires approval of Justice, and rivals once again are regaling Justice staffers with tales of Microsoft's alleged anticompetitive behavior.

WINDOWS PAIN. What really stirs fresh fear and loathing in the computer business, however, is Windows 95. Microsoft plans to begin shipping the upgrade of Windows by mid-1995, and the industry already is complaining about the software giant's pricing and marketing plans for the software. Computer makers, for example, have been startled to learn that they will be asked to swallow a huge price hike for their use of Windows 95--to as much as $70 per pc, vs. roughly $35 today. At the same time, Microsoft has established more rigorous technical requirements for hardware and software makers who want to claim their products are compatible with Windows. "Prices are going up and terms are becoming more restrictive," says John B. Landry, senior vice-president at Lotus Development Corp.

There are ways pc makers can lower their costs--if they agree to shipment goals and marketing tactics designed to give Windows 95 an early boost. Indeed, a new "Market Development Agreement" that Microsoft has distributed to pc makers spells out a dozen ways to cut the Windows 95 license fee. For example, a company can save $3 per system by preloading Windows 95 on at least 50% of its personal computers in the first month Windows 95 is available. In a business with ever-shrinking margins, that's a deal many pc makers can't afford to pass up, ensuring Microsoft lots of promotional help.

In Europe, where Windows' grip on the market isn't as firm as it is in the U.S., Microsoft's pricing has prompted a minirebellion. Vobis Microcomputer, the No.1 pc maker in Germany, announced in late November that it plans to bundle ibm's OS/2 operating system, rather than Windows, with its machines starting Jan. 1. Says Theo Lieven, Vobis' ceo: "Every penny counts."

 [D]

CONTINENTAL DRIFT. Lieven contends his rebellion already is working. He says sales have jumped since mid-November, when Vobis began offering OS/2 in addition to Windows, and "we think OS/2 helped" contribute to the increase. Other European computer makers, including Peacock Computer, have also quietly begun shipping OS/2 on their machines.

U.S. pc makers aren't likely to follow the Vobis lead--partly because the American market is less receptive to OS/2. But that doesn't mean they're all happy about the Windows 95 pricing. Hewlett-Packard Co. executives, for example, say they are concerned that the higher cost of Windows 95 may cause a pricing differential between Windows pcs and those equipped with OS/2. Still, says a spokesman, hp expects to bundle Windows 95--and not OS/2--into its machines. And other big U.S. pc makers also remain loyal. "We plan to move to Windows 95 as quickly as we can," says Lorie L. Strong, a vice-president at Compaq Computer Corp.

Still, with Microsoft on the offensive again, some rival software companies believe the Justice Dept. should use the Intuit inquiry to look once again at broad questions about Microsoft's dominance of the software market. Indeed, rivals say Justice has been asking them probing questions about Microsoft's potential dominance of new distribution channels such as on-line services. But others call another move from Justice wishful thinking. "We're just going to need to slug it out in the marketplace," says a resigned Frankenberg at Novell. The way things are going, that's just going to get tougher and tougher.

By Amy Cortese in New York, with Rich- ard Brandt in San Francisco. Gail Ed- mundson in Paris, and bureau reports

TO

APPENDIX TO MEMORANDUM OF AMICI CURIAE
IN OPPOSITION TO PROPOSED FINAL JUDGMENT
IN CIVIL ACTION NO. 94-1564 (SS)
SIGNED BY GARY REBACK

BYTE   May 1992        Volume 17, Number

Introducing Microsoft C/C + +.

* [Terms of sale]

The best C/C++
tools for Windows are
from the company that
makes Windows.

TO

APPENDIX TO MEMORANDUM OF AMICI CURIAE
IN OPPOSITION TO PROPOSED FINAL JUDGMENT
IN CIVIL ACTION NO. 94-1564 (SS)
SIGNED BY GARY REBACK

Server suite could squeeze market
Microsoft product linking plans point to another bid for dominance

By Stuart J. Johnston and Ed Scannell

Microsoft Corp.'s recently announced BackOffice server suite is the first step in an evolution designed to accomplish much tighter integration during the next few years between the company's enterprise building blocks of servers and its operating systems.

In fact, by the time Microsoft's Cairo version of Windows NT arrives in late 1995, the fit may be so tight that a competitor's knife blade will not fit between the blocks.

Problems could arise for competitors if Microsoft shares information only with its own developers on how to tightly integrate with the object-oriented Cairo file system, suggested Warren Smith, a certified public accountant and certified information systems auditor in Pacific Bell's auditing department.

If Microsoft puts shortcuts into Cairo that turn out to be better than the industry standard implementation of Cairo, Smith said the situation could be a return to the days when other third-party vendors complained about Microsoft using application programming interfaces ``that no one else knew about in some of their applications.''

At least one other observer agreed.

``All of this is an inevitability,'' predicted Jerry Schneider, president of Schneider Associates, Inc., a consultancy in Burke, Va., and former president of the Capitol PC User Group. ``The [operating system] is always going to be getting more and more aggressive. No one is safe anymore.''

The very thought may further unhinge competitors, some of which are still smarting from the recent Justice Department antitrust settlement with Microsoft. However, many large users do not appear concerned. In fact, some said they welcome a model along the lines of the old IBM that positions Microsoft as the new empire builder.

``Where Microsoft is at right now reminds me of where IBM was in the 1970s and 1980s, [and] if it continues to do things right, the users will benefit,'' said Scott Piper, a network analyst at Public Service Co. of Colorado in Denver.

``Generally, I don't find Microsoft's proprietary elements to be an impediment, [and] by making life simpler, it's going to be positive,'' said Colin Carpi, president and founder of Chartwell Advisory Services, Inc. in Penn Valley, Pa., which is developing a large on-line financial services system.

``Big is usually good [for users] because if you're going to have things work, then you [must] have standards, and that takes one [dominant] company,'' said Briscoe Stephens, coordinator for space sciences in the Advanced Scientific Information Systems Group at NASA in Huntsville, Ala.

Enhancing that vision of dominance are recent acknowledgments by Microsoft officials that over time, the line between server applications and systems software will begin to blur. The first step will be to provide tighter integration among the components in Microsoft's recently announced BackOffice server suite.

A major jumping-off point will come, however, when Microsoft ships the next major release of Windows NT, code-named Cairo, which will include a new file system that will store information as objects instead of files.

[D]

Total control

Cairo's Object File System will provide many core functions that users currently think of as database functions functions that can become part of a standard computing architecture that Microsoft controls from top to bottom. Cairo is scheduled to ship late next year, but many analysts and industry observers said they do not believe it will be out until 1996, at the earliest.

By the time the entire strategy unfolds, users may depend on Microsoft for virtually all their computing needs, which Amy Wohl, editor of the ``TrendsLetter'' industry newsletter in Narberth, Pa., suggests may not be a good thing.

``Microsoft is becoming [like] IBM, [and] the downside [for users] is the more they do that, the less open they're going to be [so that] it becomes harder to swap in your favorite database,'' Wohl said.

Microsoft officials deny their plans will make their systems more closed. Many users agree, arguing that competitors will always be able to come up with innovative products to help keep the systems open.

TO

APPENDIX TO MEMORANDUM OF AMICI CURIAE
IN OPPOSITION TO PROPOSED FINAL JUDGMENT
IN CIVIL ACTION NO. 94-1564 (SS)
SIGNED BY GARY REBACK

Desktop Computing

Commentary
Carole Patton

Bundles are bad news

[Photo of Carole Patton]

The bundling strategy is a prescription for destroying the Windows service applications market.

Windows 95 is not just an operating system. When it arrives next spring, this tour de force from Microsoft will replace all those (formerly) separate utilities you probably have on your PC right now, such as fax software, E-mail and communications capabilities. Especially neat here is that all these built-in Windows applications will be tightly integrated into a single common interface and even a central database of names and addresses.

In fact, Windows 95, the next generation of Windows, is such a complete operating environment that you may never have to purchase another Windows utility again. Nice for you. Not so nice for software developers such as Lotus, Delrina or Symantec, whose Windows products are about to become ``buggy whips'' in the name of progress. Microsoft is even including a somewhat feature-limited version of its best-selling suite, Microsofts Office, in Windows 95.

This strategy is a prescription for destroying the Windows service applications market and damaging (if not terminating) the market for core business software. For example, what if you buy a laptop preloaded with Windows 95 and Microsofts Office. Will you then go out and buy SmartSuite from Lotus or PerfectOffice from Novell? Probably not. Consumers aren't interested in replacing ``good enough'' with ``great.'' Most new car buyers keep the standard radio their car came with. Only a handful are willing to shop around and pay a premium for better audio quality.

Bundle bandwagon

Microsoft is not alone in pursuing a bundling strategy. IBM's OS/2 Warp Version 3.0 ships with a Bonus Pak that includes a word processor and a spreadsheet (IBM Works), plus a host of third-party software.

These ``free'' goodies help sell the product and anoint those vendors whose software is bundled It's in there; it must be good, users think. But such a strategy also leaves out in the cold any vendor whose software was overlooked.

While bundling is arguably anticompetitive, the issue has expanded with Windows 95. IBM's Warp cannot claim the same high level of integration of Microsoft's standards, such as the internal communications process embodied in Object Linking and Embedding (OLE).

Take, for instance, Lotus SmartSuite 3.0 for Windows, released in September. For all intents and purposes, SmartSuite is an office in a box. You get a word processor (Ami Pro), Lotus' famed 1-2-3 spreadsheet, a database (Approach), a calendar program called Organizer and Freelance Graphics presentation software. Lotus' SmartCenter tool for switching among these applications is nifty.

However, Lotus SmartSuite programmers were able to provide support foronly part of the OLE 2.0 specification: the drag-and-drop among 1-2-3, Approach and SmartCenter. Microsofts Office, on the other hand, supports OLE 2.0 across the board. (It is Microsoft's standard, after all.) It is so tightly integrated with Windows 95 that removing it to make room for SmartSuite may not be practical.

I've long thought that by selling both systems and applications, Microsoft would gain advantages that could eventually terminate competition among Windows vendors. This is especially evident now. With hardware prices dropping so fast, Windows (and Windows applications) could become ``disposable'' thanks to Windows 95. You won't ever have to replace or upgrade programs. Just buy a new PC that comes complete with all the software anyone is ever likely to need.

Sees their side

I can't blame Microsoft or IBM for trying to create a complete operating environment. We are leaving the era when users cared about software and took the time to learn a variety of different packages and understand the differences. Most computers today are being bought by novice PC users, and these newcomers require software that is easy to use. They want Windows point-and-click software, not arcane commands.

But I believe competition is, for all practical purposes, ``locked out'' when operating systems developers can integrate their own applications and so much ``free'' third-party software in a single, seamless package. If this doesn't raise a red flag in the offices of the U.S. Justice Department, then our regulators are asleep at the switch.

Here's the bottom line: Will you be better off five years from now without Lotus SmartSuite for Windows? Without PerfectOffice? Without WinFax? Without any choice?

[Carol Patton's biography unreadable]

COMPUTERWORLD NOVEMBER 14, 1994

TO

APPENDIX TO MEMORANDUM OF AMICI CURIAE
IN OPPOSITION TO PROPOSED FINAL JUDGMENT
IN CIVIL ACTION NO. 94-1564 (SS)
SIGNED BY GARY REBACK

Developing for next generation or Windows may mean running on NT

By William Brandel and Ed Scannell

While software developers are enthusiastically preparing their wares for Microsoft Corp.'s rollout of Windows 95 next year, they are not thrilled about the strings attached to Windows NT.

A number of software developers last week expressed both reluctance and frustration over their efforts to codevelop Windows 95 applications that also run on Windows NT. Microsoft has required that software developers who want to use the Windows 95 logo meet a standard of seven different features, one of which is that the application also run on Windows NT [CW, Nov. 14].

A number of vendors said developing for the two operating systems is difficult, as each targets a different market. Windows 95 is the 32-bit version of Windows mostly likely to run on Intel Corp.-based desktops, while Windows NT targets applications that have typically run on Unix platforms, such as server and workstation applications.

``We are not going to delay our Windows 95 applications so that we can deliver Windows NT applications at the same time,'' said Peter Cohen, a spokesman for Lotus Development Corp.'s spreadsheet business. ``We just don't see the demand for desktop applications on Windows NT.''

In addition to the timing issue and a perceived lack of demand for Windows NT applications, Cohen cited the need to conduct extra rounds of quality assurance testing for applications that run on NT. Those resources could be better applied elsewhere, according to Lotus.

This sentiment was echoed by a number of developers, including officials at WordPerfect, the Novell, Inc. Applications Group. WordPerfect is trying to meet the NT compatibility requirement.

``Windows NT has [more than a dozen] different APIs, a different shared memory space and different bugs to test for,'' said Gary Gibb, director of development for WordPerfect's PerfectOffice. ``You have to go through your entire testing cycle on a number of different hardware configurations and then repeat these tests on NT. And none of this makes the Windows 95 applications run any better.''

However, while Microsoft has listed running on NT as a Windows 95 compatibility standard, it has also left the door open for vendors to opt out. Microsoft will allow certain vendors to ``degrade gracefully,'' or choose not to comply with some of the compatibility requirements, if the applications do not mislead the end user into believing certain functions are available when they are not, said Brad Struss, manager of Win32 developer relations at Microsoft.

``We'll let some vendors gray out an item menu if they don't support a feature,'' Struss said.~~ He added that Microsoft's aim is to ensure that users do not run into application errors from trying to run a Windows 95 application on NT.

As for the alleged difficulty in developing to the Windows NT application programming interfaces (API), Struss blamed confusion over which API developers should be using.

Few choices for small vendors

While companies such as Lotus and Symantec Corp. have opted not to let the NT compatibility requirement bog down their Windows 95 application deliveries, smaller companies with limited resources are facing tougher choices.

``We have to measure the investment for Windows 95 development vs. our revenue streams for Windows 3.1,'' said Bruce Shafer, president of PC-Kwik Corp. For a smaller vendor, ``it is a tightwire act,'' Shafer said.

Resources aside, other small vendors said their products are not conducive to the NT platform.

``The logo requirement is extreme,'' said Richard Smith, president of Phar Lap Software, Inc. ``It makes no sense for us to have our Front Runner under NT.'' Front Runner is a shell that enables an end user to toggle between Windows and DOS applications.

However, it is risky to be a software developer rebelling against the mandates of Microsoft, cautioned industry watcher Jeffrey Tarter, editor of ``Soft Letter'' in Watertown, Mass.

``There is a long history of developers who thumbed their noses at Microsoft and later regretted it,'' Tarter said. ``As an [independent software vendor], betting against Microsoft does not seem to be a high-payoff game.''

Senior editor Michael Fitzgerald contributed to this report.

COMPUTERWORLD NOVEMBER 28, 1994

TO

APPENDIX TO MEMORANDUM OF AMICI CURIAE
IN OPPOSITION TO PROPOSED FINAL JUDGMENT
IN CIVIL ACTION NO. 94-1564 (SS)
SIGNED BY GARY REBACK

Can Microsoft get SQL Server on everything from big iron to steam irons? If you think OLE everywhere is the future, the answer is yes. By J. William Semich

The LONG VIEW from Microsoft: Component DBMSs

THEY JUST DON'T GET IT. Informix, Oracle, Sybase- all the top database-system companies. They watch in a dither for a year while Microsoft messes up their space with its dirt-cheap SQL Server For NT database technologies, and they think that Microsoft is being random (as Bill Gates might put it) with pricing that just doesn't make sense. Dangerously random.

But the UNIX database oligopoly is only haft right. They got the danger part right. Because it turns out Microsoft has a plan. When the company announced its future SQL Server 95 last June. the SQL Server crew had its three-year strategy all mapped out. `Course, they didn't show that map at the announcement-- only Microsoft's top management had seen and approved it. But we got an indepth look after hours, and we'd like to share it with you.

Microsoft is taking a three-pronged strategy with its SQL Server technology. For symmetry's sake. We'll label the prongs "Three Hundred Million Servers.' "Three Hundred Processors," and "Three Hundred Objects." If Microsoft succeeds ma all three fronts, the face of computing will change, and so will the trajectories of the high-flying database vendors.

THREE HUNDRED MILLION SERVERS

First. there's the Three Hundred Million Servers strategy. That's basically a price strategy. Microsoft thinks it can push the price of the powerful database server software central to enterprisewide distributed computing so low that all your future computing systems will be based on database servers running on super powerful, cheap boxes. This is no: a 'servers attack the mainframe" strategy, though, cautions Microsoft's director of enterprise computing. David Vaskevitch. It's more of a "servers run the business: approach. "There are 11 million places of business in the U.S. alone. They're 'all doing things right now that servers could help them do better." Vaskevitch explains. "And there are other things they never dreamed of being able to do; servers can make those things happen, too." That blue-sky approach means, for example, that SQL servers could run your phone system, copying system, cash registers, all that stuff. Not hard to get to 300 million that way, eh?

This won't happen overnight, adds Vaskevitch. but he sees it as inevitable over the long term.

THREE HUNDRED PROCESSORS

Second. there's Microsoft's high-end corporate-computing strategy. By this time next year. when Microsoft ships the next upgrade of SQL Server For NT (code- named SQL Server 95, now officially named Microsoft SQL Server)," its database server will be able to run on the most powerful mainframe-class multiprocessor computers and virtually match the pow- er. Features. and functions of the latest multiprocessor and parallel-processing products From the UNIX vendors.

You could call this the Three Hundred Processors Strategy and you'd not be half wrong. Well. maybe not 300 processors- at least not right away.

The Three Hundred Processors Strategy. is actually one way Microsoft plans to become a major player in large-scale, mission-critical computing technologies. Microsoft recently restructured itself to better Focus resources on making it happen. Last year. prior to restructuring, Microsoft sold just under S5 billion worth of PC software in a market that totals barely S10 billion. With its high-end corporate strategy, Microsoft intends to move into the $70 billion+ market for business software, so it can grow lickety split to something like $20 billion.

"We'll need to be selling into a $ 100 billion software market to Bet to that S20 million." says Richard Tong, marketing director for Microsoft's new IS-focused Business Systems Division.

Of course, sitting squarely in the middle of that enterprise computing market are the likes of Computer Associates. Oracle. Sybase-you know the names. They won't let on that they're really concerned about the competition From Microsoft. They say Microsoft's big talk is just smoke. So how does Microsoft intend to prove them wrong?

Early prowess toward the long-term Three Hundred Million Servers kraal will help Microsoft achieve its aggressive revenue growth forecast, but performance improvements will help more. By 1996. when Microsoft ships its even more advanced SQL Server For Cairo. it expects it to actually outperform Informix- Online Dynamic Server 6.0. Oracle7.1. and Sybase System l0 (see "MS-SQL, Server: Good Today, Better Tomorrow").

Think you'll ever seriously consider swapping your mission-critical DB2- based financial management system, or oracle 7-based airline reservation system, or even your Sybase SQL Server 10-based loan approval system for a realty complex Excel spreadsheet with a SQL Server engine? Hah!

But Bill Gates is betting the company that you will. Not exactly an Excel spreadsheet, of course- but a whole new kind of mission- critical, high-availability, heavy- duty, object-oriented, component- based enterprise system that includes the next generations of Microsoft's Windows NT operating system (dubbed Calro), the new SQL Server 95 (and the unannounced SQL Server For Calro), and its coming distributed OLE technology.

All this will happen in the next year or two, a not-so-distant future that Microsoft internally refers to as "the Calro timeframe."

At right is the techno-time line Microsoft hopes will turn into reality:

THREE HUNDRED OBJECTS

Third, there's the Three Hundred Objects prong of Microsoft's strategy, the component-software-system pieces. In order to build enterprise-computing systems From reusable mix-and-match software components, you need more than the object-oriented operating system Cairo, distributed OLE, and the Visual Basic enterprise development tool technologies. You need a technology that turns desktops and servers into peers when it comes to storing, sharing, and finding objects.

To help make this happen. Microsoft is moving SQL Server technology downscale, onto the desktop. The plan is to use pieces of MS-SQL Server technology to rebuild desktop apps like Microsoft Access and Excel so desktops and server components can talk to each other.

"There's this huge mismatch, in terms of semantics, between the big server- based database systems and the tools that run on the desktop." says Gary Voth, group products manager, Business Systems division. "We want to build a common technology across the server-common data access semantics, common rich query technology-a unified engine architecture with the best of Access, Fox, and SQL Server itself. Then we'll move it into a single architecture, with shared technology between the server and desktop, and then componentalize it all."

Come Cairo, that common architecture will be distributed OLE and the components will be distributed OLE objects.

The first implementation of this component strategy is in SQL Server 95's new state of enterprise management tools, which are supposed to simplify the process of managing large, distributed systems across geographically dispersed servers with stuff like drag-and-drop replication, automated restore and restart. The tools, which Microsoft previously code-named Starfighter but has officially named Enterprise Administration Tools For SQL Server 95. are 'all OLE objects. Starfighter lets users build their own database management scripts using a new 32-bit version of Visual Basic. SQL Server 95 itself is. in effect. an OLE automation server for these OLE tools and scripts.

In other words. Microsoft is rebuilding SQL Server so that it can contain and manage software components. When SQL Server For Cairo is shipping, Microsoft's world of computing will become a world of OLE objects-components that a developer can link together using OLE's APIs into an application. Then SQL Server won't just store data-it will contain components.

Explains Casey Kiernan. Microsoft's program manager for SQL Server tools: "Cairo is really an OLE 2 operating system-under Cairo, you're either an OLE 2 object or you're nothing. We'll wrap OLE 2 around the legacy stuff. The OLE SQL Server object [in SQL Server 95] is the first step to Cairo."

SQL Server will be just one of Cairo's server-based components, says Voth. 'All of our server apps-SQL Server. Systems Management Server. Information Exchange Server, and SNA Server-will have this single integrated model in Cairo," he explains.

The company is serious about this, too, Vaskevitch says. No matter how long it takes, or h6w much work has to be done to make the technology compelling to commercial users, the company is committed to making its NT-based SQL Server the enterprise launching pad for its Cairo component-computing system.

'We've 'already invested three years in the planning process for SQL Server. and it doesn't bother us if it takes five. even eight years to Let to where we want to be-we don't give up." says Vaskevitch. It probably doesn't hurt to have deep pockets. either.

THE FUTURE Of TECHNOLOGY & PRICING

So there it is-the future of enterprise computing according to Microsoft. the world's richest software company. And where will 'all the UNIX database companies be, come Cairo time? Today, at least, they still act like they don't have a due.

They still think they can advance the technology by making their database systems into bigger, better, faster (and pricier) versions of what they've been selling for the past decade and a half-with, of course, the magic sobriquet 'open' pasted onto it all.

They may think that. But according to Microsoft's plan, all these big. distributed UNIX megaliths will soon seem just as rigid, overpriced, oversized, and outdated as yesterday's mainframe and mid-range systems seem today. Microsoft's low-priced component based SQL Server 95 system and the low-cost hardware it runs on, conversely, will (so Microsoft thinks) form the basis of a whole new culture of computing-kind of like DEC with its avant-garde midrange computing technologies in the heady 1970s.

Fact is, the UNIX crowd is still stuck in the fast-receding bygone era when a system's power and size determined its price-the higher the price, the better the system. "Need to query a 5 million row, one gig table fast? Here, use this. That'll be $500,000, please. Neet to run a departmental transaction record-keeping system? Okay, that's $117,000."

How do they set these prices? In what other industry could a marketing director describe product pricing strategy like this: "We look at what value our products bring to our customers' businesses and price according to that value." Tranlation: The more money your company earns, the more it should pay for its RDBMS.

They're extending that logic to conclude that Microsoft's SQL Server, at only $19,995 for an unlimited-user, multi-processor RISC-based version, must be a low-power workgroup product. If high value equals high price, then low price must equal low value, right?

It's like, they're watching Microsoft through a chink in a fence, right, and all they can se is a shiny piece of stainless steel here, a flash of glass there, maybe a lintel and a post somewhere else. So they think, "Hey, what are these guys making in there?" They figure it's a copycat version of what they've been building and selling for the past few years: a monolithic, UNIX-like, database-centric, big-business system that can pile information high to the sky without collapsing from its own weight.

"Tain't likely, Cause there really is a cultural revolution spinning out around how businesses do computing in the workplace-moving the power of computing down to the local level, onto your desktop, even down onto your lap.

Is Microsoft's 300 Million Servers, etc., vision the right one for this cultural shift? Hard to tell at this point. Maybe IBM will do a few quick flips and-presto-put together a better bunch o'buyables before Microsoft can. Or maybe NeXT's Steve Jobs and his UNIX bigots' object alliance, Sun, HP, DEC, and others, will get there first.

But one thing's for sure. Three, maybe five years from now, the UNIX database system vendors (the one or two who are still around to play) will be using a whole different set of rules-rules first sketched out up in Redmond by the likes of David Vaskevitch, Rich Tong, and Gary Voth.

You'd be crazy not to start looking seriously at Microsoft's SQL Server 95 technology. But you'd also be crazy-OK, not crazy, just adventuresome-to commit your company today to Microsoft's component enterprise-computing plan lock, stock, and betel.

Even so, rye looked at SQL Server 95 and 96 up close and undressed, sort of, and it is something really slick. Microsoft recently demonstrated an early, early version to my company in the hopes that we would port our apps to SQL Server For NT. I can't tell you my company's name, but it's one of the leading midrange manufacturing packaged software application vendors.

First off, I was surprised and impressed at the level of the technology they're showing in SQL 95 and SQL 96. Microsoft looks like it's paying attention to the issues that are important to making SQL Server an enterprise-quality database management system.

They seem to be building In a scalabilary capability for symmetric multi- processors, and they're leveraging the multithreaded capability of the NT operating system. That gives them a leg up on the other database products, which rely on different versions of UNIX. Some versions of UNIX don't support multi- threading, so everything goes through a single queue. Microsoft isn't constrained by that.

SQL Server 95's new system administration toolsets [Starfighter] are very impressive. Microsoft is plying a lot of attention to things like ease of use and the kinds of data replication issues that are necessary to manage performance and backup, and necessary for bet-back- up capabilities, parlor. mange monitoring, and job scheduling. The SQL 95 job scheduler is integrated into NT. That technology alone shows that Microsoft is trying to listen to enterprisewide needs. I don't see SQL 95 as a scaled- down version of Sybase System 10 at all. It appears to have the same robust capability that Sybase has.

That said, I still chose Sybase System 20 for the next version of my company's packaged software. Why? Because my customers can buy it today, and I know it works and works well. And if I were a CI0 or CTO at a large enterprise, I would do the same thing.

SQL Server 95 is an "NT only" solution-whatever advanced functionality Microsoft's building into it now is predicated on the success of NT. That's still an open question. Besides, Microsoft's track record on delivering both functional and technical quality products out of the box isn't what I'd like it to be for kinds of solutions I'm trying to still. I'm net soiling spreadsheets and wad processors. Plus, Microsoft has no track record selling enterprise systems or applications. It's a gamble.

My recommendation is don't bet your job or your company on Microsoft SQL Server and NT Server today. Let Microsoft's existing dedicated NT users do that instead.

But you should look at SQL Server 95 and 96 very closely for the future. If you're an IS manager, assign staff to watch SQL Server and NT technology developments very closely. Do some background prototypes using SQL Server 4.21 [Microsoft's current version]. Start building apps that include Microsoft's other enterprise server products like Hermes [Systems Management Server] and Information Exchange Server to see how the emerging Microsoft approach to enterprise computing fits your company's future needs. Then sign up as a beta site for SQL Server 95 later this year. [Microsoft is accepting beta requests as of this writing.]

That way, you'll at the latest gain valuable experience in the operating systems and client/ server technologies that will begin to dominate the industry over the next two to three years, as Microsoft moves its user base to the Cairo component-computing environment.

If you're bringing up new mission-critical apps. go with the knowns players: IBM, Oracle, Sybase, CA, or Informix.

DATAMATION AUGUST 1, 1994

TO

APPENDIX TO MEMORANDUM OF AMICI CURIAE
IN OPPOSITION TO PROPOSED FINAL JUDGMENT
IN CIVIL ACTION NO. 94-1564 (SS)
SIGNED BY GARY REBACK

Preface

The year 1993 was one of dramatic change in the PC software industry. This report will highlight the major events of 1993 in personal computing software. We analyze the positioning and directions of the top 10 vendors, dissect our data, and then analyze applications by category., operating system, and region. We conclude with our forecast of future trends in the industry.

Data included in this report is listed as needed for our discussion. For a comprehensive list of our historical data and forecasts, refer to the Personal Computing Software Worldwide Market Statistics, a series of three reports published in June 1994 (product codes: PCSW-WW-MS-9401, -9402, and -9403).

Dataquest's PC Software service tracks all major PC software business productivity applications running on the DOS, Windows, Macintosh, OS/2, and Windows NT operating systems and environments. Other services concentrate on other areas of the software market our Multi- media service tracks entertainment and education software; our Client/ Server service tracks development tools and server databases; and our Digital Documents and Operating Systems services complete the offerings.

In Appendix A, we define our market coverage boundaries. In Appendix B, we discuss the methodology used to arrive at our decisions.

We hope that you find this information useful Please contact us if you have any questions regarding the data or analysis.

The PC Software Team

• Karl Wong, Principal Analyst
  internet: kwong@dataquest.com; phone: (408) 437-8213
• Bryan Fukuda, Industry Analyst
  internet: bfukuda@dataquest.com; phone: (408) 437-8153
• Suzanne Snygg, Industry Analyst
  internet: ssnygg~dataquest.com; phone: (408) 437-8R24
• Jingsheng Huang, Research Analyst
  internet: jhuang~dataquest.com; phone: (408) 437-8g160

©1994 Dataquest Incorporated

Personal Computing Software Worldwide

Figure 4-2
1993 Unit Shipments Growth by Category

[D]

Analysis of Each Category

This section will analyze the 1993 results for each category. Future trends for each category will be discussed in Chapter 7. For a definition of each category, see Appendix C.

Accounting
Intuit, a new entry in the accounting market in 1992, jumped to the top spot in revenue in 1993 (see Table 4-1). [Note: Table 4-1 was not included in document submission]. Realworld, Peachtree, and Great Plains all ship products that have a higher ASP than Intuit; however, reve- nue for all four companies is very closely matched in the U.S.$20 million range. A slow transition to thin Windows platform contributed to the reve- nue decline in 1993.

Communication
The communication market exploded in 1993. Bundling arrangements with modem OEMs contributed significantly to the 167 percent growth in unit shipments (see Table 4-2). Revenue increased by an impressive 47 percent. Delrina's WinFax: Pro and Datastorm's Procomm Plus led the charge. The growth of the laptop market also spurred unit sales. There is still room for growth in this [market, but 1993 will be remembered as the year this market really took off.

Table 4-12
Top Vendors in the Spreadsheet Market
(Revenue in Millions of U.S. Dollars)

Table 4-13
Top Vendors in the Suites Market
(Revenue in Millions of U.S. Dollars)

Utilities/Application
WordPerfect's Grammatik for both DOS and Windows were the two lead- ing applications in 1993. WordPerfect garnered a 36 percent market share based on revenue in 1993 (see Table 4-14). This is a small market that involves small companies able to find a niche market. This is not a market where we will likely see one or two vendors dominate.

Table 4-14
Top Vendors in the Application Utilities Market
(Revenue in Millions of U.S. Dollars)

[No further explanations were included]

TO

APPENDIX TO MEMORANDUM OF AMICI CURIAE
IN OPPOSITION TO PROPOSED FINAL JUDGMENT
IN CIVIL ACTION NO. 94-1564 (SS)
SIGNED BY GARY REBACK

WASHINGTON, D.C. -- Microsoft, the world's largest and dominant computer software company, agreed to end its illegal monopolistic practices after the Department of Justice charged that the company used unfair contracts that choked off competition and preserved its monopoly position.

The company agreed to settle the charges with a consent decree that will prohibit Microsoft from engaging in these monopolistic practices in the future.

Microsoft, which makes the MS-DOS and Windows operating systems used in more than 120 million personal computers, was accused of building a barricade of exclusionary and unreasonably restrictive licensing agreements to deny others an opportunity to develop and market competing products.

Attorney General Janet Reno said, "Microsoft's unfair contracting practices have denied other U.S. companies a fair chance to compete, deprived consumers of an effective choice among competing PC operating systems, and slowed innovation. Today's settlement levels the playing field and opens the door for competition.'

'Microsoft is an American success story but there is no excuse for any company to try to cement its success through unlawful means, as Microsoft has done with its contracting practices,' said Anne K. Bingaman, Assistant Attorney General in charge of the Antitrust Division.

The settlement is the result of close coordination between the Department of Justice and the competition enforcement authorities of the European Commission, which has been investigating Microsoft since mid-1993, and which also initiated an undertaking containing essentially the same terms. This complaint and settlement marks the first coordinated effort of the two enforcement bodies in initiating and settling an antitrust enforcement action.

Bingaman, praised the Commission, noting that, 'This unprecedented, historic cooperative action sends a powerful message to firms around the world that the antitrust authorities of the United States and the European Commission are prepared to move decisively and promptly to pool resources to attack conduct by multinational firms that violate the antitrust laws of the two Jurisdictions.'

The civil complaint and consent decree were filed last night, July 15, in U.S. District Court in Washington, D.C. The consent decree, if approved by the court, would settle the suit. Until approved, Microsoft has agreed in a stipulation filed with the court to abide by the terms of the decree.

The Department alleged that Microsoft used the following unfair practices:

Exclusionary Per Processor Licenses--Microsoft makes its MS-DOS and Windows technology available on a 'per processor' basis, which requires PC manufacturers to pay a fee to Microsoft for each computer shipped, whether or not the computer contains Microsoft operating system software. The complaint alleges that this arrangement gives Microsoft an unfair advantage by causing a manufacturer selling a non-Microsoft operating system to pay at least two royalties--one to Microsoft and one to its competitor-- thereby making a non-Microsoft unit more expensive.

'Microsoft has used its monopoly power, in effect, to levy a 'tax' on PC manufacturers who would otherwise like to offer an alternative system,' said Bingaman. 'As a result, the ability of rival operating systems to compete has been impeded, innovation has been slowed and consumer choices have been limited.' She noted that Microsoft has maintained the price of its operating systems while the price of other components has fallen dramatically. Since 1988, Microsoft's share of the market has never dropped below 70 percent.

Unreasonably Lonq Licenses--The Department further alleged that Microsoft's contracts are unreasonably long. By binding manufacturers to the purchase of Microsoft products for an excessive period of time, beyond the lifetime of most operating system products, the agreements foreclose new entrants from gaining a sufficient toe-hold in the market.

Restrictive Non-Disclosure Agreements--The Department also charged that Microsoft introduced overly restrictive non- disclosure agreements to unreasonably restrict the ability of independent software companies to work with developers of non- Microsoft operating systems. Microsoft sought the agreements from companies participating in trial testing of the new version of Windows, to be released later this year. The terms of these agreements preclude applications developers from working with Microsoft's competitors for an unreasonable amount of time.

The settlement ends these practices and will help to rectify the effects of Microsoft's past unlawful conduct. In particular, the settlement prohibits Microsoft from:
--Entering into per processor licenses.
--Obligating licensees (manufacturers of personal computers) to purchase any minimum number of Microsoft's operating systems;
--Entering into any licenses with terms longer than one year (although licensees may renew for another year on the same terms).
--Requiring licensees to pay Microsoft on a "lump sum' basis.
--Requiring licensees to purchase any other Microsoft product as a condition for licensing a particular Microsoft operating system.
--Requiring developers of applications software to sign unlawfully restrictive non-disclosure agreements.

The settlement is effective immediately and will be in effect for six and a half years.

Bingaman said "this settlement resolves the competitive problems created by Microsoft's unlawful conduct quickly and effectively."

Microsoft's main corporate office is in Redmond, Washington.

94-387

ORIGINAL

PRESS CONFERENCE

With Attorney General Janet Reno and

Assistant Attorney General Anne Bingaman

Regarding the Microsoft Settlement

Saturday, July 16, 1994

(Transcribed from a provided audiotape.)

ALDERSON REPORTING COMPANY, INC.
(202)289-2260 (800) FOR DEPO
1111 FOURTEENTH STREET, N.W. SUITE 400 / WASHINGTON, D.C. 20005

P R O C E E D I N G S

ATTORNEY GENERAL RENO: Good afternoon.

The Justice Department has charged Microsoft, the world's largest software company, with using unfair marketing and contracting practices to choke off competition to preserve its monopoly position. Microsoft has agreed, yesterday, to settle the charges with a consent decree that will prohibit the company from continuing to engage in monopolistic practices in the future.

While the company fairly and lawfully climbed to the top of the industry ladder, it used unfair and illegal practices to maintain its dominant position, and kept honest competition from other U.S. companies.

The Justice Department has taken an action that is critical to the personal computer industry and the efforts to make it competitive. This settlement will save consumers money, enable them to have a choice when selecting PC operating systems, and it will stimulate innovation in this critical market.

Today's settlement is the result of close coordination between the Department of Justice and the Competition Enforcement Authorities of the European Commission, which, today, also has indicated an undertaking containing essentially the same terms.

This complaint and settlement marks the first coordinated effort of the two enforcement bodies in initiating and settling an antitrust enforcement action.

I want to thank and to recognize Anne Bingaman and the fine staff of the Antitrust Division, who have worked through long hours of negotiations to resolve quickly this significant case, and achieve the best results for the consumers of America.

And now I would like to ask Anne --

MS. BINGAMAN: Thank you.

We are proud of the achievement that the settlement filed in Federal District Court in Washington, the District of Columbia, at 9:30 last night represents. It is a significant -- in fact, historic -- breakthrough for the software industry, for innovation, for the competitiveness of the American economy.

Let me describe for you briefly what the case we filed is about and what the settlement achieves, because they are significant.

Number one, the settlement will open the playing field; it will level the playing field for Microsoft's competitors in the operating system software market, to enter this important market, to bring down prices to consumers, to innovate, to produce better products.

Microsoft, for years, and has today, monopoly Dower in the software -- operating system software market AS this chart shows, Microsoft has 79-plus percent of that market. Its competitors are other American companies who have been struggling for years to enter this market to provide better, cheaper products to American consumers, and Microsoft's contracting practices, which are challenged in this lawsuit and which are ended by the settlement we achieved, have prevented those competitors from entering the market. They have deprived consumers of choice. And they have stopped innovation -- slowed innovation in this important market.

Let me describe to you the four major things that Microsoft did and which this settlement ends.

Number one, the per-processor license, I'll describe in a moment.

Number two, contracts of extraordinarily long duration which blocks the market.

Number three, huge, 100 percent minimum commitments for years, which amounted to take-or-pay contracts, which blocked the market.

And, four, restrictive non-disclosure agreements for software writers which prevented them from writing for other software companies in some cases.

Let me turn first to the per-processor license, what that is and what this settlement does to stop it.

Number one, the settlement bans it outright. That is first. What the per-processor license has done until last night at 9:30 was to lock up 60 percent of this market in the United States in per-processor contracts which Microsoft began using in 1988. Per-processor contracts are contracts which Microsoft imposed by virtue of its dominant monopoly position on computer manufacturers, such as Dell, Compaq, Gateway, you name it, the OEM's they are called in the business, the computer makers, who have to license from Microsoft because it has had this monopoly position and the products are demanded in the marketplace.

Nothing wrong with that, but rather than simply sell those products fair and square on the merits and on price, in 1988, Microsoft invented a form of contracting called the per-processor license, under which it required the computer manufacturers -- induced them with extremely low prices to pay for every processor they shipped of a certain type not just to Microsoft, but to the competitors.

So it worked this way: Under a per-processor license, which 60 percent of the industry has had until last night, Microsoft got paid for every processor shipped by a computer maker, whether or not that processor had a Microsoft operating system loaded on it.

Now, if you are a competitor of Microsoft and you wanted to sell your competing product to a consumer, you do that through these computer manufacturers. But they had to pay Microsoft.

Now, if Microsoft -- take this hypothetical -- operating system was $15, and you came in with a better operating system or cheaper, it worked just as well, hypothetically $10 -- these numbers are lower than average, but for ease -- under the per-processor license, the computer manufacturer pays Microsoft 15 and the competitor 10 for a total of $25 on what really is a $10 item.

The result, computer manufacturer were reluctant -- extremely reluctant -- to buy from competitors. And that was the purpose and the effect of the per-processor license. It's obvious what it does. It drives prices up to consumers. It raises prices. It locks out competitors. And it slows innovation.

So, this settlement stops the per-processor license.

Two, Microsoft used contracts of three to five years in an industry that was rapidly turning over. These extraordinarily long contracts made it very difficult for competitors to get in. The settlement we achieved today reduces contract lengths to one year, with one, one-year extension on the same terms and conditions which the computer manufacturer, in its sole option, can elect.

So, we have 9one to one-year contracts, banning of per-processor.

The third important feature of this settlement is abolishing minimum commitments. Microsoft's third way to lock up this market was to say to the computer makers who had to deal with it, We will give you a lower price if you estimate a large volume.

Nothing inherently wrong with that volume discounting. The problem is Microsoft quoted these low prices in conjunction with 100 percent minimum commitments -- i.e., you get that price only if you sign on the dotted line to pay us every cent regardless of whether you actually ship our product or not -- a take-or-pay contract. You pay no matter what.

Well, what does that mean?

Over a long-term contract, what that means is if the computer manufacturer's business has not gone quite as well as it thought, it is locked into Microsoft no matter what because it owes them this minimum commitment, even if it has not sold any machines. So, minimum commitments was a third way that Microsoft locked up this market, locked out competitors, and minimum commitments are abolished. They are zero in the settlement we achieved yesterday.

Finally, NDA's, non-disclosure agreements, were restrictive agreements which Microsoft, this winter, imposed in a manner that had never been done before in the software industry on certain applications writers. It would have -- the NDA's challenged in this lawsuit and which Microsoft in the consent decree agrees to stop would have prevented applications writers from discussing Microsoft's operating systems for as long as three years after public disclosure of the operating system.

The effect could take those application writers, the software writers, forever out of business, in effect, except for Microsoft. It is another way to, in effect, lock up the market -- this time by locking up the important software applications writers.

Microsoft itself has said these NDA's were a mistake. It has agreed in this consent decree to never engage in such practices while this consent decree is in effect. And that also is a significant achievement of this settlement.

The last thing the settlement does is prohibit the use of lump-sum contracts, which would have been another way that Microsoft could have locked up this market. They had not needed to use them in the past because they had these other methods, but looking forward, our concern was that they might. And so the settlement also bans lump-sum contracts.

This settlement is everything we could have hoped for in a fully litigated case and possibly more. It is an historic achievement. I tell you, the charts we have prepared today were prepared for the lawsuit we planned to file yesterday. The lawsuit was not filed because of the settlement. We filed instead a complaint with a settlement. We are extremely proud of this result.

And the last point that the Attorney General noted, I think, deserves mention. This is the first time in history that the Competition Authorities of the European Commission and the Department of Justice have cooperated closely in investigating a major worldwide company, whose anti-competitive practices affected important markets both in Europe and the United States.

We took this under a letter -- the EC and I and the Department of Justice asked Microsoft last October to waive any confidentiality restrictions under our respective statutes so that we could work together and think about the case we were jointly -- not jointly to them -- but that we had each initiated. Microsoft agreed to that in writing.

We worked with the EC throughout the winter. We shared documents. We worked closely with them. We settled this together on terms that are substantially identical. We negotiated in Brussels the week of July 4th with Microsoft. We negotiated this week at the Department of Justice with EC officials here. And this also is a truly historic aspect of this settlement.

So, we are extremely proud of this. We are gratified that it concluded with a consent decree which achieves the really 100 percent results that any lawsuit could have achieved, and possibly more. And I want to especially note that this was the ultimate team effort. We had a group of lawyers, led by Sam Miller, who is here today, and Don Russell, who is on his way back from Brussels -- he has been in Brussels all week coordinating this hour by hour with the EC over there ~- we have had extraordinary people on this case. We had a team of lawyers I would put against anybody, and I would feel for the other side.

And I want to simply state the names on the complaint we filed last night, because I am so proud to have been part of this group. The complaint was signed by Sam Miller, Don Russell, Joyce Bartu, Bob Zastrow, Dick Irvin, Peter Gray, Justin Dempsey, Gil O'Hana, and Larry Frankel. And there were more, and we had a paralegals. And this was an effort of a remarkable, extraordinary, incredible group of lawyers that I am so proud to have been part of. And I am proud of our partnership with the EC.

So, with that, what can I tell you about any questions you have?

QUESTION: What kind of room does this give Microsoft's competitors -- (inaudible) -- civil actions?

MS. BINGAMAN: That is up to the competitors. I do not actually believe this case changes the legal status of any competitor's suit, because, by settling, Microsoft has admitted to no facts. It has consented to entry of the decree that was filed with our complaint. But facts are not established of record by a settlement, the way they are by a litigated case to conclusion, with a jury trial.

So, my own horseback impression is that the action, as such, does not change the legal status. But, as far as private suits by competitors, it has enormous impact for competitors in opening the market. This is exactly what has been needed for years and years in the software industry. And I think, in the market-opening respects and for innovation, prices to consumers, it will have tremendous impact.

QUESTION: Why has this taken so long, and why is there no monetary penalty? And I notice it says that -- you say in the press release that it bans these practices in the future, but then says it only lasts six-and-a-half years.

MS. BINGAMAN: Okay. You have got several questions there. Number one, we have had this case for a little less than one year. The FTC had it for, I think, two-and-a-half or three years before that. As everyone knows, or a lot of people, the FTC deadlocked two to two. We took the case acting as a fifth commissioner. We have looked carefully at this case because it is an important case, and we wanted to understand it fully ourselves.

So, I have no concerns whatsoever about a one-year action by the Justice Department that ends these practices.

There are no monetary penalties because they are not provided by any law and never have been. When the Justice Department settles a civil case, the Antitrust Division -- the antitrust laws do not provide for civil penalties, period.

We obtain adjunctive reliefs to open the market. Under the American legal system, private actions obtain any monetary damages, and that is just the way it is in all of our cases. They are no different.

You had a third aspect.

QUESTION: The length of time, you say --

MS. BINGAMAN: Oh, the six-and-a-half years. Our decrees normally last 10 years. We negotiated long and hard with Microsoft over the length of the decree. The EC's decrees last four-and-a-half years. We obtained immediate effect of this decree. That was a crucial aspect of the decree. And we believe we added, in effect, three to three-and-a-half years on the front end of the decree because the contract duration stops right now. The per-processor stops as of last night.

The illegal practices that had locked up the market are ended. And they do not have to wait for contracts now in effect to run out. And it was our belief that based on all of those facts, plus the EC's practice of four-and-a-half year decrees, that this was a fair balance under the circumstances.

Let me mention something. I neglected to thank -- and it was a major oversight on my part and I want to correct it -- Henry Kawati is sitting here, who worked long and hard on this case, he is an economist with our Economic Section; Rich Gilbert, who is head of that section, was in Brussels with me; and Mark Schecter, who killed himself on the case, along with Bob Lighten, but I want to thank Henry Kawati and Ken Hire and Rich Gilbert, because the economics aspect of this case, as you can imagine, was critical. We had outstanding outside economists who Henry worked with tirelessly for many, many months. And he was a critical part of it, as was Rich Gilbert and Ken. So, I wanted to say that.

QUESTION: Can you estimate how much these practices may have cost consumers over the years?

MS. BINGAMAN: We have not. Because we do not bring damage actions, we do not put efforts into trying to figure out monetary total impact. But I can, to illustrate, tell you this. If you were a consumer and wanted to buy a competing operating system, and despite Microsoft's practices, there have been, in fact, four major competitors in this market to Microsoft, who have clawed and grabbed and have managed to obtain some market share, if you bought one of those competing companies, and 20 percent of the American public does, and you were under a per-processor license, and many of these licenses, as we saw, are per-processor, you paid not just Microsoft anywhere from $15 to $50 for its operating system, you paid the competing price on top of that.

And so Microsoft, in effect, taxed every consumer who bought a competing operating system and bought it from a maker who had one of these per-processor contracts, or a similar one. And so it's not insignificant. We have not, as I said, made any effort to quantify it, but it is -- there are millions and tens of millions of PC's shipped every year, and it is a major amount of money.

We can try to come up with some numbers after the press conference. But with all the other things we have done, that has not -- our focus has been opening the market, truly, and obtaining the relief we needed.

QUESTION: To follow that up, do you have any estimate of how many computers were shipped under these agreements that would have been effected?

MS. BINGAMAN: I can come up with numbers on that. We have not tried to. It is in the tens of millions. There are 120 million total computers with Microsoft operating systems on them. Many, many were shipped with this -- under these kinds of practices. And it has been a major market problem for competitors, and has restricted choice for consumers.

Let me tell you why else this is so important to the American economy. We are about innovation and competition in this economy. That is what we are for. And Microsoft has its shot at the market. No problem. All we are saying is others should have their shot at the market, fair and square, a level playing field. That is the American way.

And they may have a better mousetrap. They may not. But what we are saying is people should get a chance to judge it fairly on quality and price and the other factors. And that is what this case is about. It levels the playing field, opens the door.

And if a competitor has a better product that can run computers faster, run them better, support better applications, build a base, cut into Microsoft's market share so that applications writers will write for it, that could have profound consequences for the American economy. What we are about is precisely that -- promoting competition, innovation, better products at cheaper prices, and letting the market take care of whatever happens.

We are not about driving the market; we are about letting the market operate freely.

QUESTION: Had this settlement not been reached, what broader or further action could Microsoft have been subjected to? And, a second question is, had there been any serious consideration about splitting Microsoft into two?

MS. BINGAMAN: I cannot discuss our internal considerations as such. I can tell you that we looked at every possible legal theory and at all the facts throughout the course of a long, tough winter, that the legal team I mentioned went through. And it was our conclusion at the end of that that the case to be filed was the case we did file. We did not bargain off any case in exchange for a settlement. This was the case that was there after thousands of hours of work. And it needed to be brought, and it was brought.

And that is really as much as I think confidentiality permits me to talk about specifics.

QUESTION: Potentially, had this gone into litigation, what could we have seen perhaps in terms of time and cost?

MS. BINGAMAN: Had this been litigated, we hoped to conclude it within a year. We planned to file it in a district in which the dockets are not crowded and we could have obtained a quick resolution, because the markets need to be open. This needed to get done. But it would have been a minimum of a year at the very best. It undoubtedly would have been appealed. And the key point is, after all that, two to two-and-a-half years at best through appeal, we could not have achieved one thing more than we got in this settlement.

And, frankly, I am not sure we would have gotten as much. I do not know, because I do not know what a judge would have done. But this settlement is 100 percent of what we would have gotten with a lawsuit.

QUESTION: Can you tell us more about the EC cooperation, how and when was that initiated? And wasn't there a British investigation as well?

MS. BINGAMAN: No, there was no British investigation. It was the European Commission. It was a result, actually -- last September, I went to Europe for consultations, which are annual consultations with the EC that we have done for years, the Antitrust Division -- it's a mutual cooperative thing -- and Klaus Ailerman, the head of the Competition Directorate said to me, What are you doing about Microsoft, because we have a Microsoft case, too, you know, and I am very interested to talk to you about it?

And I looked at him and I said, Klaus, I do not think I can say a word to you about Microsoft. Everything I know is under confidential documents. I am forbidden from talking about it. I can't speak to you.

And he said, Well, what a great pity, because we've got, as far as I can tell from press reports, the same case.

And I said, Well, it is a great pity.

And I came back to the United States -- that was the end of September -- and 10 days or two weeks later, it just hit me out of the blue one day, we should ask Microsoft to waive confidentiality so that we could cooperate and decide whether in fact there is a case and coordinate remedies.

And the coordination of remedies is really crucial for a company in Microsoft's position, which operates worldwide, literally, in -- I do not know -- tens of countries in the world. They need, for their own business reasons, to have the same contracting practices. It would be terribly disruptive -- and I called the EC. We asked Microsoft. Microsoft, for its own reasons, said that would actually -- they didn't have a problem. They waived confidentiality. And that is how it began last October, and it has continued since then.

QUESTION: How is the Justice Department going to monitor the new agreements, the new contracts that Microsoft will sign with its OEM's? And what guarantees are there that Microsoft isn't going to turn around and say, you know, if we cannot do the kinds of volume deals that we have done in the past, we are going to charge 5, 10, 15 percent for the operating system than we have in the past?

MS. BINGAMAN: If they charge more for their operating system, the competitors are there, without question, with comparable products. And the market should take care of that. That is the whole idea of this settlement. The market should take care of it.

We are allowed, in the monitoring provisions of this decree, which you should have, to request documents from Microsoft, to inspect their contracts, to talk to their people. We are further -- the decree specifically provides we can cooperate with the EC in this monitoring, so we will continue our cooperation and close work with them.

And we are watching. We are very much on the case.

QUESTION: A question about the per-processor issue. From your presentation it wasn't entirely clear to me, but it sounds as though Microsoft main pressure on computer companies was that they got -- they would offer huge, huge discounts to the companies that would accept a per-processor kind of agreement. That being the case, it seems to me that, on one level, the sin is that Microsoft is simply charging too little for the operating system.

And, to follow up on that, to follow up on that, it seems to me that the marketplace situation may not be a whole lot different, because Microsoft can continue, it seems to me, to charge that same low, low price.

MS. BINGAMAN: Ed, you been talking to Microsoft? That is their line. They are not telling you right. If that was so easy, why did they have per-processor licenses? They are the only company in the industry that did. Why did they have three- to five-year contracts? They are the only company that in the industry that did. Why did they have 100 percent minimum commitments? They are the only company in the industry that enforced that.

If this was so simple, why were they locking up the market with practices which every computer manufacturer despised and which the competitors despised and which Microsoft hung tough through four years of Government investigation to hang on to? Do you think that is because it did not matter to them? That is the story they are putting out.

You are darned right they are trying to spin it their way. That is not right. And let me tell you. Volume discounts, of course they can volume discount. No question. There is nothing wrong with volume discounting. It is done in all kinds of industries, in all kinds of situations. And the decree does not address volume discounts as such.

The problem with Microsoft's practices is that they were using volume discounts to lock up the market with per-processor contracts and 100 percent minimum commitments, which then were like iron. You could not get out of. You could not escape.

To get those low prices, you had to sell your soul and never leave Microsoft. And that is what this decree changes.

Microsoft can compete on the basis of low price. We have no problem with that. That is good. We want that. What we do not want is competing on the basis of low price and then using that to impose contract terms which exclude every other competitor.

And, Ed, the reason they were able to do that is because of their monopoly position in this market. I mean, this is an important question you are asking, because they are going to try to claim that this decree changes nothing. That is wrong. That is a lie. And people need to understand that.

Because volume discounting, in and of itself, is not a problem. There are ways volume discounting can be abused. I have discussed those ways with Microsoft and Bill Gates. We are watching. We are watching closely what they do with volume discounts. They know it. I know it. And we are going to see what happens here.

But volume discounting, in and of itself, is not a problem. There can be problems in how you structure them, whether you force -- it's a technical discussion. But, in any event, believe me, they did not hang tough on this for so long, right to the brink of a joint lawsuit by the U.S. and the EC yesterday because these practices' were so harmless and meaningless and so forth. But I can see why they say it.

(Laughter.)

QUESTION: Is there going to be an immediate effect that we will notice for consumers?

MS. BINGAMAN: I hope consumers, within a short period of time, will have more choice of operating systems, genuine choice, more innovation in computers. Certainly, the prices will lower for consumers who already buy competing operating systems. Any of these companies in the market right now can now sell just for their price, not for this double tax that Microsoft has gotten.

So, I think prices will immediately lower, and I think, over the medium to long range, this will, I hope and believe, have profound market opening impacts. It will help innovation, help the competition give us better products. You may be using a different operating system three years from now because of this -- maybe. And if you are, great. If you still want whatever, great.

But the point is you should have a choice. Everyone should have a choice. And the companies that compete with Microsoft should be able to offer you that choice fairly and evenly.

VOICE: Thank you.

QUESTION: Microsoft's competitors in applications have complained about the access that they have had to all kinds of information about the operating system code. Did the Justice Department not find that Microsoft had unfairly restricted applications developers to various aspects of the software?

MS. BINGAMAN: The nondisclosure agreement, the so-called NDA part of the case, focused on nondisclosure agreements required -- are you talking about something else?

QUESTION: I mean, certainly the NDA has been part of it, but other companies --

MS. BINGAMAN: The so-called interoperability?

QUESTION: Yes, yes, hidden calls and all of the charges that have been raised over the past few --

MS. BINGAMAN: I can tell you we have looked closely at all aspects of this case. We have examined it closely. And I think all that I can say, because of the strictures of confidentiality and the law, is that we have looked at it and this is the case we chose to bring because this is the case that is there and needed to be brought. And I think that is all I should say.

VOICE: Okay. Thank you.

VOICE: Thank you.

MS. BINGAMAN: Okay. Thank you.

VOICE: Thank you very much. (End of proceedings.)

TO

APPENDIX TO MEMORANDUM OF AMICI CURIAE
IN OPPOSITION TO PROPOSED FINAL JUDGMENT
IN CIVIL ACTION NO. 94-1564 (SS)
SIGNED BY GARY REBACK

Microsoft's Grip on Software Tightened by Antitrust Deal

Andrew Schulman

On Friday, July 15. Microsoft signed a consent decree with the Antitrust Division of the U.S. Department of Justice (DoJ), ending a four-year investigation by U.S. antimonopoly agencies-first the Federal Trade Commission (FTC) and later the DoJ--into Microsoft's trade practices. At the same time, Microsoft signed a nearly identical settlement with the Directorate-General for Competition of the European Commission. The judgment lasts for six and a half years in the U.S., four and a half in Europe.

Microsoft agreed to immediately abandon several arrangements for licensing the MS-DOS and Windows operating systems to PC hardware vendors. It also agreed to halt some "unnecessarily restrictive" clauses in its nondisclosure agreements (NDAs) for the forthcoming "Chicago" version of Windows. The consent decree explicitly excludes Windows NT.

The consent decree is still subject to a 60-day public review. The full text of the DoJ's July 15 complaint against Microsoft for violations of sections 1 and 2 of the Sherman antitrust act. the U.S. District Court final judgment in U.S. v. Microsoft, and the "Stipulation" signed by the DoJ and Microsoft consenting to the final judgment. are available via Internet Gopher from the DoJ's Gopher server.

Who Won?
The consent decree was first viewed as a victory for the DoJ and Microsoft's com- petitors. The New York Times (July 17) carried the front-page headline, "Microsoft's Grip on Software Loosened by Antitrust Deal," and crowed that "the pact could reshape the world of computing .... The accord could undermine Microsoft's near total control of tile market for operating systems." The Boston Globe's headline was equally enthusiastic: "Microsoft Accord to Greate Competition in US. Europe."

Indeed. the consent decree sounds at first as if it should cramp Microsoft's style, and lead to more competition in PC software. For years. Microsoft tins provided PC hardware manufacturers (original equipment manufacturers, or OEMs) with per-processor licenses to MS-DOS and Windows. in which the vendor pays Microsoft based on the number of machines it think it will ship, rather than the number of copies of DOS or Windows it ac- tually uses. In 1993, such per-processor agreements accounted for about 60 percent of MS-DOS OEM sales, and 43 percent of Windows OEM sales.

According to the DoJ, "Microsoft's per processor contracts penalize OEMs. during the Life of the contract, for installing a non- Microsoft operating system. OEMs that have signed per processor contracts with Microsoft are deterred from using competitive alternatives to Microsoft operating systems." The consent decree put an immediate stop to this practice, leading to the hope that non-Microsoft operating systems would now have a shot at the desktop.

But the morning after, nearly everyone realized that. in fact, U.S. v. Microsoft is a victory, for Microsoft. Directly contradicting the previous day's headline, a New York Times (July 18) news analysis by John Markoff spoke of "Microsoft's Barely Limited Future": "Rather than reining in the Microsoft Corporation, the consent de- cree... frees the company to define computer industry's ground rules through the rest of the decade." The Wall Street Journal had a similar take: "A Winning Deal: Microsoft Will Remain Dominant Despite Pact In Antitrust Dispute." According to the Journal Gates -has just won big again, this time by letting the Justice Department rake in a small pot while his company retains the power to dominate the nation's desktops.

In the Fast day of trading after the settlement. Wall Street made its statement on the consent decree: Microsoft stock rose $1.87, to $50,50. Rick Sherlund, an ana- lyst for Goldman Sachs, stated that with the settlement. Microsoft "should domi- nate the market for desktop software for the next 10 years." Another frequently quoted analyst. Richard Shaffer, announced that -The operating system wars are over--Microsoft is the winner .... Microsoft is the Standard Oil of its day."

But how could a ban on an important Microsoft trade practice be viewed as cementing Microsoft's hold on the industry?

First. to achieve the DoJ's goals, the change from per-processor to per-copy licensing probably comes about four years too late. Despite some brave words from IBM and Novell after the consent decree. it seems unlikely that the change will lead to a larger presence for OS/2 or Novell DOS. As a spokesman for Compaq (which already offers OS/2 to its customers) noted, "Windows is the standard -- not much will change."

Nor does the consent decree address the key questions about Microsoft's role in the PC software industry. Companies such as Lotus and Borland that compete with Microsoft in application areas such as word processors and spreadsheets have long asserted that Microsoft "leverages" its control of the operating system to benefit its applications -- particularly the Microsoft Office "suite," which bundles together Microsoft Word. Excel, Access, Mail, and PowerPoint -- at the expense of applications and suites from other vendors.

Grabbing the Whole Pie
More and more. Microsoft's applications seem like part of the operating system. Many PCs today come. not only with MS- DOS and Windows preinstalled on the hard disk, but also with Microsoft Office. The forthcoming "Chicago" release of Windows will include numerous features once considered the province of dried-party applications developers. Microsoft not only has a near-monopoly on the operating system. but is constantly expanding the definition of what belongs in the operating system.

Some commentators see these increasing ties, and the DoJ's apparent refusal to touch them, as a good thing. For example, Steward Alsop was quoted in the New York Times (July 18) as saying, "If you really care about improving the personal computer, you want Microsoft to take over all the pieces of the pie."

There is a certain logic in this. For example, one reason the Apple Macintosh was for so long far easier to use than a PC was that Apple had a closed architecture and completely dominated the market, guaranteeing that almost everything carne from a single vendor. Monopoly has some clear benefits. In certain situations. such as public utilities, monopoly may he the only viable industry structure, leading to a so-called "natural monopoly."

Interestingly. the superb biography Gates, by Stephen Manes and Paul Andrews Doubleday, 1993), quotes a 1981 statement by Microsoft chairman Bill Gates where he noted that volume and standards in PC software can lead to a "natural monopoly" But companies in such a favored position usually are forced to make an Important trade-off: so-called natural monopolies are generally regulated, are prevented from expanding their monopoly into new areas, and so on.

Microsoft already has MS-DOS installed on about 120 million PCs in the and Windows on about 50 million. With the DoJ consent decree, Microsoft can move even more rapidly toward its goal of becoming an unregulated, nonpublic utility providing total, one-stop shopping for all your software needs.

Exposing Microsoft's Monopoly
Microsoft continues to deny that it mo- nopolizes the PC software industry. Nor has it admitted to any guilt by consenting to the court's final judgment. The consent is explicitly "without trial or adjudication of any issue of fact or law: and without this Final Judgment constituting any evidence or admission by any party with respect to any issue of fact or law."

Nonetheless. the PC software industry has been treated to some puzzling denunciations of Microsoft trade practices from high government officials. After the signing of the consent decree. U.S. Attorney General Janet Reno said. "Microsoft's unfair contracting practices have denied other U.S. companies a fair chance to compete, deprived consumers of an effective choice among competing PC operating systems, and slowed innovation."

The Assistant Attorney General for Antitrust. Anne Bingaman, noted that `Microsoft is an American success story but there is no excuse for any company to try to cement its success through unlawful means, as Microsoft has done with its contracting practices."

`Microsoft has used its monopoly power, in effect, to levy a `tax' on PC manufacturers who would otherwise like to offer an alternative system." said Bingaman. "As a result, the ability of rival operating systems to compete has been impeded, innovation has been slowed and consumer choices have been limited." According to a DoJ press release. Bingaman noted that Microsoft has maintained the price of its operating systems even while the price of other components has fallen dramatically, and that. since 1988. Microsoft's share of the market has never dropped below 10 percent.

The Road Not Taken
No matter what else it says, the fact remains that the consent decree addresses only a narrow issue: OEM sales represent less than 25 percent of Microsoft revenue.

The complaint notes that "At least 50,000 applications now run on MS-DOS and over 5000 have been written to run on Windows. Microsoft sells a variety of its own very successful and profitable applications." But that is all it has to say about applications!

The complaint also notes that "All versions of Windows released to date require the presence of an underlying operating system, either MS-DOS or a close substitute." but says nothing about alleged tying arrangements between Windows and MS- DOS (see "Examining the Windows AARD Detection Code" DDJ. September 1993).

Similarly, the complaint mentions `critical information about the interfaces in the operating system that connect with applications -- information which the ISVs need to write applications that run on the operating system" -- yet doesn't address the issue of whether or not Microsoft unfairly withholds some critical information. trying to give its developers exclusive use of undocumented interfaces.

Likewise, the DoJ was well aware of. and quite interested in. the Issues surrounding Microsoft's ownership of the vast- ly Important DOS and Windows standard, Yet none of this is addressed in the consent decree, which ends up looking quite similar to what Microsoft probably could have got from the FTC a year ago. Even Bill Gates, who was apparently in the habit of denouncing even the mildest FTC and DoJ questions as "communistic" and "socialistic." had to admit that the final settlement was no big deal saying, after years of investigation, that "this us what they came up with" (Wall Street Journal, July 18).

Why So Little?
Why did the DoJ settle for so little? How could they seemingly ignore the entreaties of so many PC software vendors?
One theory is that the Clinton administration views Microsoft as a "national treasure." and put pressure on DoJ to leave Microsoft alone. The press made much of a May 25 meeting between Bill Gates anti Clinton's chief economic advisor. Robert Rubin. The date is significant because just one week later. Gates testified under a before the DoJ. According to one anonymous source. Gates pointed out to Rubin that Microsoft is responsible for a substantial portion of U.S. software exports (Information Week. June 27).

Frankly. I don't buy Clinton administration pressure as an explanation for the DoJ's limited settlement. Microsoft may be highly visible, but it simply isn't that important to the U.S. economy, at least when compared to companies such as IBM or GM that make tangible goods. Microsoft, remember, produces software. While software is a crucial part of the modern world economy, consider that even "giant" Microsoft has only about 15,000 employees and that its quarterly' sales are about $1.25 billion, compared to $13.3 billion for IBM. or even $2.5 billion for Apple.

What makes Microsoft different is its incredibly low costs. This is very nice for Microsoft, but it's hard to see what it does for the U.S. economy, especially when 45 of Microsoft's stock is owned by insiders. Had it wanted, the DoJ could have made a moderately plausible case to the American public that Microsoft. far from being a `national treasure,' is simply a grossly profitable monopolist, with few employees and few stockholders, that gives back little to the public.

Another explanation is that DoJ feared a repeat of U.S. v. IBM, which dragged on for 13 years, only to be dropped as "without foundation." While you could easily imagine lawyers for the DoJ not wanting to stake their careers on a losing battle, you have to wonder whether U.S. v. IBM was such a complete washout, after all. Even though the case was eventually dropped, for years it had a serious effect on IBM. You could even argue that it was this supposedly unsuccessful case that caused IBM to unbundle software from hardware, there- by opening the way to an independent software market, making room for software upstarts, including a company called Microsoft. In many cases, Microsoft was a beneficiary of U.S. v. IBM, and "the next Microsoft" could have been a beneficiary of a U.S. v. Microsoft case.

Ultimately, I think that the DoJ didn't push for more against Microsoft for the very simple reason that they felt they couldn't win anything else. Responding to widespread criticism of the settlement as a DoJ sell-out, Anne Bingaman protests, "folks, we looked at every aspect of this. We brought the case that was there to bring." According to the DoJ, the Microsoft settlement was "everything we could have hoped for in a fully litigated case. and possibly more."

This is probably true. Law like poli- tics. is an "art of the possible." While the settlement gives the Microsoft steamroller the green light, at the same time it's hard to see what the DoJ could have done differently. The DoJ's job is to enforce the antitrust laws. not to make industries more competitive -- and the two are not the same.

What all this means is that those Microsoft practices studied by the DoJ, but not covered in the settlement, are either not illegal, or would be too difficult to prove illegal.

Where To Now?
While there might be some private antitrust action from Novell, Lotus, or Borland. and while the terms of the settle- ment are subject to public review, Microsoft must be feeling emboldened by the limited scope of the consent decree. Microsoft should be able to go full- steam ahead with its plans to greatly expand the operating systems dimensions in Chicago. Microsoft Office will increasingly seem like an essential part of Windows. With policies such as its new. heavy, requirements for using the "Windows Compatible" logo (see "How to Adapt an App for Chicago: Requirements for the New Windows Logo." Microsoft Developer Network News, July 1994). Microsoft is raising the Windows develop- merit bar ever higher.

The PC-software industry is rapidly headed in the same direction as many other technology-based industries before it: rapid consolidation to a handful of vendors. There once were hundreds of U.S. car manufacturers: now there are just a few. With Novell's acquisition of WordPertect and parts or the Borland product line, with Symantec's acquisition of Central Point, and Microsoft's purchasing a minority share in Stac Electronics, we are already seeing the same (,probably inevitable) process occurring in software. As Table I shows, market shares reflect an already highly concentrated industry.

On most scales. Microsoft is about twice the size of its two nearest competitors combined. Lotus had 4450 employees and Novell also had 4450; Microsoft has 14,450. In 1993, Lotus sales were $981 million and Novell sales were $1.123 billion; Microsoft sales were $3.753 billion.

Given that the DoJ could apparently do very little about this increasing concentration in the software industry, what are software developers and vendors to do?

It is probably stating the obvious, but there is Little point in trying to compete with Microsoft over productivity apps and office suites. These are rapidly becoming a quasi part of Windows itself, and even Novell and Lotus probably have little chance in this area. Microsoft Office is everywhere and everything. Perhaps there is still some room in databases. desktop publishing, and personal-finance software.

As always, another interesting area is plugging holes in Microsoft's own offerings: add-ins to Microsoft Office, remedying the inevitable temporary problems in Chicago, and so on.

The best bet is to find areas where Microsoft doesn't have a product, and where there is a chance of a several-year window of opportunity before it does have a product. On the other hand, the only market I've ever heard of that Microsoft didn't want to get into was pornographic screen savers and related multimedia titles. As one company employee told me, "We looked carefully at adult software. and decided to leave that money on the table."

DDJ

Dr. Dobb's Journal, October 1994

TO

APPENDIX TO MEMORANDUM OF AMICI CURIAE
IN OPPOSITION TO PROPOSED FINAL JUDGMENT
IN CIVIL ACTION NO. 94-1564 (SS)
SIGNED BY GARY REBACK

The Economist
FEBRUARY 27TH-MARCH 5TH 1993

THE COMPUTER INDUSTRY

Reboot system and start again

[Note: The graphs in this document are unreadable, therefore, have not been included]

FOR three decades the computer industry seemed to epitomise the marriage of technological wizardry and business acumen. Led by IBM. the industry masterfully exploited a pace of technological change that would have left managers m most other industries gasping. It grew through boom and bust, and revolutionised the way nearly all other businesses worked. Best of all, it consistently made enormous profits.

Computer executives saw themselves as both innovators and adventurers. Some pioneered new ways to manage armies of highly educated, independently minded employees. A few left the security and prestige of a corporate career, or went straight from the university classroom, to start companies from scratch. Naturally some computer firms failed. But the industry, as a whole was largely immune to the travails that periodically beset more mundane businesses. For many people, computers were the quintessential industry of tomorrow.

Tomorrow has arrived and it is not a pretty sight. For the past two years the computer industry has been in turmoil: plummeting profits, flat sales, tens of thousands of jobs lost, vicious price wan. The industry's reversal of fortune has been so abrupt that it has left many of its leading companies floundering.

IBM, the biggest computer maker and long one of the most successful companies in the world, lost $4.9 billion in 1992, one of the biggest corporate losses in history. In January John Akers. Its boss, resigned. Last year the company shed 40,000 of its 340,000 employees in an effort to control costs (which still look hopelessly bloated). Its stockmarket value is now about the same as Microsoft, a firm which employs only 12,000 people. And the once- mighty Big Blue is not alone, DEC, the world's second-biggest computer firm, ousted its founder and chairman in 1992 and lost a whopping $2.8 billion. Olivetti, Siemens-Nixdorf. Groupe Bull, Fujitsu, Hitachi and NEC have all seen profits collapse over the past two years. Wang, a high flier until the early 1980s, ended up in a bankruptcy court.

Clearly computers are no longer recession-proof. But global recession is not the only, or even the primary, cause for the industry's recent tribulations. Recession has simply accelerated changes that have been reshaping the industry anyway. Until the mid-1980s the computer business was dominated by a handful of large firms--foremost among them IBM--whose marketing and technological prowess let them educate, reassure and control the corporate customers who bought most computers. Smaller companies often introduced the latest technology to the market; but usually their innovations were not widely accepted until they, got the imprimatur of Big Blue. These smaller firms seldom posed much of a threat.

The invention of the personal computer (PC) in the late 1970s brought in a motley collection of brash, new firms. At first the growing popularity or PCs had little effect on me fortunes of the inqustry's leading firms. Indeed. IBM itself became the world's biggest manuracturer of PCS, and its marketing clout helped their sales to soar. For 20 wars IBM had skilfully coped with technological advances that appeared to be far more radical, or disruptive, than the personal computer. Few people inside or outside the company thought that mere relatively any and rather simple machines were much of a threat to IBM's hegemony or to the stability of the industry as a whole.

Getting personal

They, were spectacularly wrong, in the past few years, personal computers and the microprocessor chips on which they, are based have upturned the economics of the business. This has happened so quickly that many computer executives are bewildered. Their industry has become one of con fusing extremes. In any large industry the fortunes of different firms will vary. But in today's computer in- dustry the differences are stark.

While many computer firms sacked thousands of workers and last huge amounts of money last year, others thrived despite price wars and recession. On the day in August 1992 that Wang filed for chapter-11 bankruptcy, Dell, a personal-computer maker, reported quarterly sales up 129% and net profits up 77%. In 1992 some companies, such as Apple and Compaq, which looked doomed because of the price wars ravaging the PC market, staged stunning comebacks (though they too had to cut)obs and other costs). Price-cutting spread from hardware to software. And yet profits at Microsoft, the world's biggest personal-computer software company, leapt 53%. The computer business still boasts many of the world's fastest-growing and mast profitable firms. But it now has some of the world's biggest loss-makers too. Today's industry offers other remarkable contrasts. Despite the fact that its overall profitability has fallen so sharply (see chart 1). hordes of new competitors continue to enter almost every part of it. And far from slowing the pace of innovation, as might be expected, hard times seem to have quickened it. An unprecedented number of new products came to market last year. This stream is about to become a flood. With chip technology improving faster than ever, a plethora of new products will reach the market over the next few years: pen-based PCs, hand-held computing and communication devices, ever more powerful versions of today's desktop and notebook computers, sophisticated network and database software. cheaper and fancier supercomputers.

Moreover, a growing part of the computer market shows many of the classic characteristics of a commodity business: there are few discernible differences between products except price, low bam- ers to entry and razor-thin profit margins. This is a novelty for such a high-tech, inventive business. The large amount of intellectual property contained in computer products, and their complexity, ought to make it easy for companies to keep out new ovals, differentiate their products and command fat matins. Instead, even in many esoteric niches of the industry, growing competition is eroding margins.

Equally remarkable is the web of collaborative deals that spans the industry. As competition has become fiercer, the number of joint ventures, alliances and mahatma agreements has multiplied rapidly, although this has done nothing to soothe the growing ferocity of competition. Nearly every firm, whether small or large, now has a vanity of ties with dozens of others. Confusingly, many alliances seem designed to compete with other alliances containing some of the same firms, as companies place multiple bets on new technologies or market wends. Although these agreements are often between companies with complementary products, many are between once-bitter rivals, such as Apple and IBM, who stress that their collaboration does not rule out tough competition between them now or in the future.

Perhaps one of the most puzzling things about the computer industry is that, for all its vitality, its glory days of high growth and gushing profits are probably over. In a recent report on the industry. McKinsey, a management-consuiting firm, predicts that the industry's sales will grow by 6% or less 3 year--scarcely more than me nominal growth rate for me world economy as a Whine. "Just surviving will be a struggle and even many of today's healthy companies could become extinct," says Michael Nevens, one of the report's authors. Others agree with this gloomy assessment. IBM predicts that software and services will grow at some 11-13% a year between now and 1997; but sales of hardware will lag well behind economic growth.

Evangelical fervour

One reason is that the cost of computing power continues to drop by 30% or mort a year, became of advances in chip technology that show no signs of slowing. This inexorable improvement has now begun to outstrip the demand for more computing power from customers. Another reason for stow growth is that, with more than $300 billion in sales, the computer industry is now so large that it probably cannot expect to capture a much bigger chunk of corporate or consumer spending. Most businesses, laboratories and classrooms already have some type of computer, Many are crammed with them. Because of the rocketing popularity of poketbook and laptop computers, so are many, briefcases.

A barrage of new products will be needed just to keep spending at current levels. In fact, the amount spent on computers per white-collar worker (the biggest users of computers) has been flat in America since 1983, and has recently levelled off in Europe and Asia as well (see chart 2). The total stock of hardware and software in developed economies is also set to level off in the next few years, according to many forecasters.

Even if the industry must learn to live with humdrum growth, there will be nothing dull about the computer business itself. This survey will spend little time on the myriad ways computers are used or how they are changing lives. Instead it will try to examine the peculiar economics of computers, and to make sense of the blizzard of news the industry generates every day. Why does it present in many conundrums? And why has no shakeout yet ended what, compared with most other large industries, looks like intolerable instability?

The growing number of competitors and the pace of technological change are raising the level of uncertainty, for both computer firms and that customers. As a result, the bosses of most computer companies are no longer the smug technologists or buttoned-down managers of a decade ago. They are preachers fervently trying to sway customers, suppliers, investors, employees--and often themselves--with their vision of the future. One of the industry's favou??te verbs is to "evangelicise." This is an odd choice for sober-suited managers care- fully investing billions of dollars. But it is all too appropriate to the opportunists or true believers now best equipped to survive in the computer business.

Personal best

TO UNDERSTAND how drastically those little personal computers have changed the industry, and why they have suddenly left it in such a segmented state, a little history is needed. Until the late 1970s, nearly all computers were large machines used to grind through mind-numbing calculations and routine book-keeping chores. Computers were especially useful in making the administration of large organisations more efficient. Outside laboratories, they were bought in the greatest numbers by large companies, which could afford to pay their hefty prices and to employ the professional programmers and technicians needed to keep the temperamental machines from breaking down.

Machines came in various sizes, the two broadest categories being mainframes and minicomputers. Both types could have several users at once, who sat at terminals to put data into the machines or to take it out. Mainframes soon stood at the heart of most of the world's biggest companies. Smaller and less expensive (though still quite pricey; minicomputers were often used by the divisions of the same firms, or by medium-sized firms which could not justify spending enough to buy a mainframe.

IBM bestrode the industry, accounting in 1980 for 38% of the industry's revenues and 60% of its profits. Even if IBM had not become such a dominant firm, a small group of large firms would prob- ably have controlled the industry in any case. There are two reasons for believing this. First, computers were the most complicated machines ever made, in Pact, they were so complicated that even individual machines were called "systems". And second, though the Industry was large in revenue terms, relatively few machines were sold each year. As recently as 1980, fewer than 10,000 mainframes and 105,000 minicomputers were sold worldwide each year. Such volumes are significant for suppliers of capital equipment, but they are minuscule compared with those of the car or other consumer industries. Customers were reluctant to buy these cranky machines from anyone but large, established suppliers. So newcomers had a hard time breaking into the lousiness.

The complexity of computers produced another crucial characteristic of the industry; it virtually ensured that computer makers would opt for vertical integration--that is, to make most of the parts of the machines themselves, with the software to run them, rather than buying parts from outside suppliers, and to do most of their own marketing, distribution, sales and service as well. A few of the smaller firms could not manage this. They, either specialised in supplying pieces of equipment, such as terminals, tape drives or printers, which were attached to computers, or they bought what they could from outside suppliers. This put them at a huge disadvantage, for the simple reason that there were so few independent suppliers for the many components needed to build a computer. So the bit computer firms built their own machines from the ground up.

The resulting structure of the industry looked something like the diagram above. Customers cared little about the various layers identified in this diagram, because they almost always bought all the layers in a stogie package from one supplier. It is useful to pause here to define these terms, because they will loom large later on.

Basic circuitry refers to the thousands of wires, transistors and other electronic bits which were m all computers. In the 1970s and 1980s most of these bits and bobs were gradually replaced with integrated circuits printed onto small pieces of silicon--ie, microchips. This allowed some specialised chip firms, such as Intel. Motorola and Texas instruments, to become parts suppliers to computer makers. But many of the biggest computer makers, most notably IBM and Japan's Fujitsu, NEC and Hitachi, made their own chips.

Computer platforms refers to the assembled machines. These were useless without operating- system software, the programs needed to make the machines do anything but hum. Once the operating system enabled the machine to respond to various commands, application software told the machine what to do: compile the payroll, store data, solve abstruse equations, perform word-processing or whatever. Several applications usually ran on the same computer. As the diagram shows, firms did most of their own distribution, although some machines were sold through computer-leasing firms or "systems integrators".

One more point must be made: all computer makers used "proprietary" standards both to build their hardware and write their software. Except for a few firms which tried to mimic IBM's standards, no firm's software worked with any other firm's software, or ran on any other firm's machine. This locked customers into a single computer supplier. As a customer's investment in computers grew, the more dependent on his supplier he became, The cost of scrapping all of a firm's existing hardware or application software (which big firms sometimes wrote themselves) to switch to another supplier became prohibitive. Occasionally a customer became restless, especially if a supplier was charging too much, fell too far behind the rest of the industry technologically or failed to service hrs machines properly. A few small, specialised firms sprang up to engineer so-called "gateways", bits of hardware and software that would allow machines from different companies to work together.

For the most part, though, customers had to commit the bulk of their spending on computers to a stogie supplier. The safest thing to do for anyone who had to make this purchasing decision--in big companies usually the data-processing manager-- was to buy from the biggest supplier, no matter what the cost And that. by a long way, was IBM.

Chips with everything

From its inception, the personal-computer market assumed a different pattern from the established industry. PCs became possible only because chip manufacturers had managed to cram a simple ver- sion of a computer's central processing unit, the circuites that did most of the actual computing, on to a single chip. Appropriately, this was called a micro- processor. Around a stogie microprocessor, a small, cheap machine could be assembled from readily available parts used to supply the consumer-electronics industry. So most personal-computer mak- ers were never vertically integrated. Separate groups of firms supplied pare, fully assembled machines (platforms), operating-system software and application software.

Personal computers were primitive compared with mainframes and minicomputers. But they, could perform simple tasks such as word processing, keeping mailing lists or playing games, and they proved surprisingly popular. To grab some of the revenue from this small but burgeoning market, IBM launched its own PC in 1981. Because it wanted to do this quickly, it assembled its machine from off-the-shelf components made by firms which were also supplying other Pc: makers. It arranged to buy the two most important parts of the machine- the microprocessor and the operating-system software--from (respectively) Intel and a small Seattle- based company called Microsoft.

With IBM's backing, personae-computer sales skvrocketed. At first this was great news for IBM. which had the lion's share of sales and considered Pc revenues simply a welcome supplement to its mainstream computer business. Thousands of small software companies ocean writing application programs for IBM's machines, boosting demand for mere still furtner.

Still, from several points of view IBM had badly miscalculated. Buying the key parts of its machines from intel and Microsoft, without demanding any kind of exclusive deal, effectively left control of the technical standards in these companies' hands. Scores of other firms, many of them new ones such as Compaq, quickly learnt to "clone" copses of IBM'S machines using Intel chips and Microsoft's Ms-Dos operating system. Their machines also ran all the software written for IBM's machines. To the user, therefore, there was no real difference between them. Users began buying the machines primarily on the basis of price. As demand for the machines took off (see chart 3), hundreds of small, lowcost producers jumped into the market. Prices began to collapse, even while the growing power of microprocessors rapidly boosted the capabilities of PCS.

Until the mid-1980s, many PCs were sold customers--individuals, schools, small businesses, professional firms--who would never have been reached by tram or other established computer makers. But when big corporate customers began tying the mast powerful PCs together into networks as alternatives to minicomputers and mainframes, IBM became alarmed. In 1987 it belatedly tried to gain control of the personal-computer market with new models containing a patented technology called Microchannel, which rivals could not copy and which made IBM's machines incompatible with everyone else's. Competitors aptly dubbed these machines "clone killers". But by then it too late. The PC market had slipped beyond IBM's grasp and the Microchannel machines flopped. Big Blue, like any other manufacturer, had to make its machines fit the industry's standards. Three were now set by millions of personal-computes users and owned by. Intel and Microsoft.

In only a few wars, before IBM or the other established computer makers had realised what was happening, an entirely new computer industry had grown up next to the old one.

Harsh new world

THE companies which made a comfortable liv- ing for so long in the old computer industry face a challenge rather like switching from making battleships to rowing-boats in just a few years. Almost every defining feature of the old industry, has been reversed in the new one. Instead of selling thousands of expensive machines to an easily identifiable set of corporate and institutional customers, the new industry, sells tens of millions of cheap machines each war to individuals, businesses of any size and sham, and every type of organisation imaginable. Unlike mainframes or minicomputers, personal computers need little maintenance. And most of their software can be bought off-the- shelf, like a can of beans, rather than custom-designed for each user by teams of expert programmers. As a result, even in large corporations the computer-purchasing decisions are now made by hundreds of people with little technical knowledge, instead of just one or two computer nerds.

Instead of the proprietary hardware and operating-system software of the old industry, "open" standards now prevail. These permit the products of a growing number of computer firms to work together, which has opened the door to thousands of new firms that now compete at every link of the "value chain", from chips to distribution. Peter Shavoir. IBM's chief business strategist, estimates that 2,500 firms took some pan in the computer industry of 1965, but that 50,000 jostle for business now. Most of the new ones entered the industry, m the 19805 along with the personal computer.

Despite this upheaval, the old computer industry will survive for some time yet. And mainframes, in particular, may never entirely disappear. Lame organisations will need to process huge mountains of data quickly and store it securely in a single, central machine for a long time to come. "Some cus- tomers will always require robust, bullet-proof, bet- your-business kind of applications. These really do belong on a mainframe," argues Nick Dono??io, head of IBM's mainframe unit.

However, sales of such big machines are shrinking. Networking, the fastest-growing segment of the new computer industry, strikes at the heart of the old. At first PCs were strung together in networks to allow the users of individual machines to send information to one another. In the industry's jargon this is known as "peer-to-peer" computing. Some peer-to-met users no longer needed to be con- nected to a larger computer to communicate.

More ominously for the makers of big machines, the honest trend in the industry is now the much more sophisticated "client-serve" network. In this type of network, a large number of personal computers ("clients") are connected to a central personal computer ("the server") which, at a fraction of the cost, does many of the things a minicomputer or mainframe once did, such as storing data, managing the flow of information between users and enabling them to work on the same documents. Many of these networks are built around a powerful type of personal computer called a workstation, based on a microprocessor called a RISC chip. Pioneered by. Sun Microsystems, work, stations were first used by, engineers. Now they are being used by, businesses for a variety, of tasks and have become one of the fastest growing, and most fiercely contested, pans of the computer market.

The rapid growth of client-server networks is eliminating the need for big computers in many organisations. Companies like IBM and DEC, which sell big machines, reply that in many cases big machines themselves will function as the server for a host of "client" PCS. Nevertheless, with spending on hardware unlikely to grow, some class of machine must suffer, and higher-cost mainframes and minicomputers seem the most likely losers.

Even if demand for bill machines holds up longer than expected, the creation of the new computer industry, has wreaked havoc with the economics of the old. Spoilt by the convenience, choice and ever-falling paces offered by personal computers, buyers have demanded the same low mature, nance, open standards and price reductions from large machines. The growing use of microprocessors in mainframes and minicomputers has enabled the old-style computer makers to provide some of what their customers want. But this has also left them adrift with armies of surplus salesmen, service staff and factory workers, which accounts for the thousands of layoffs in the past year. If microprocessor technology continues to advance as rapidly as it has clone in the past (see chart 41-- and everyone in the industry expects, it to do so-by me end of the 1990s even mainframes could be the size of PCs. They might be just as cheap too.

Well before that happens. It is likely that the new industry will have swallowed the old one. Because it represents the future, and is already where most of the action takes place, the rest of this survey will concentrate on the new computer industry and refer to the old only in passing.

Horizontal desires

Discerning a clear structure in the new industry is hard, but the diagram below is one attempt to do it. Like any such diagram. It is a simplification: technology and competition could soon change it. And yet it is a useful guide to the industry today, so will be used as the map for the rest of this survey.

Begin by comparing it with the much simpler diagram on page 9 of the old computer industry. The most stoking thing is that the new industry Is a series of horizontal layers, each containing many companies, rather than the vertical, single-company towers of the old industry. Each layer represents a distinct market. The barriers to entry, for new firms vary from laver to layer: but in no layer are they as high as they were for the old computer industry as a whole, in which any new firm hoping to challenge the established computer companies head-on had to build an entire vertical tower of its own. As a result, competition in every layer of the new industry is much fiercer than it ewer was in the old. This explains why profits for the entire industry have dropped since 1986.

The diagram is borrowed from Andy Grove, the boss of a successful American chip maker, Intel. It is easy to see why he is fond of it." For us. who deal in the fundamental technology, it's wonderful." he says. Intel's dominance of the microprocessor level (layer 1) is matched only by Microsoft's hegemony in the client/stand-alone operating-system software level (layer 3) two steps above. Barriers to entry in both these layers are relatively high, because Intel and Microsoft have established de facto industry standards with their products. Supplanting them would be hard but (more on this shortly) neither Intel nor Microsoft is unassailable.

Layer 2, computer platforms, includes assembled personal computers of every size and shape-- desktops, workstations, laptops and notebooks. Largely because Intel's dominance of microprocessors has established an industry standard in this laver, parriers to entry are minimal. Any technician who can buy intel, or intel-standard, chips land intel sells to anyone Jean ??oit together a respectable desktop PC from readily available components. So. predictably, this layer is where competition is fiercest. The continual newspaper and television ad- vertising of computers which most people see, and the brutal price war which has captured so much attention in the business press over the past two years, come from firms competing in this layer.

The next layer, operating-system software, is divided between the basic software needed to operate the central server of a network and the software needed to run diem machines in the same network or stand-alone PCs. The top half of the layer is much bigger than the bottom half--some 90m machines run these operating systems compared with just a few million functioning as servers. But the bottom layer is growing fast and is highly profitable. For the purposes of this diagram, their relative sizes are not relevant and so they are shown as equivalent.

Layer 4. applications software, is the arena in which Microsoft, Lotus, WordPerfect, and Borland battle for marker share. This layer projects into a third dimension because a few of the biggest application categories--spreadsheets, word processing, database management and graphics--are distinct markets in themselves, although all fit into the layer. Barriers to entry for any new firm hoping to grab business in one of these categories are somewhat higher than to computer platforms, became writing such complicated programs is time-con- suming and expensive. The need for a strong brand name, and the ability to market and distribute such general, purpose software packages also act as barriers to newcomers. But tens of thousands of small firms compete in the application layer outside these areas with specialised software packages.

The layer above that is probably the most competitive of all, as firms scramble to find the most efficient way to reach customers with machines and software. In recent years, some of the industry's biggest winners and losers have been here. Dell grew from nothing in 1984 to just over $2 billion in sales in 1992 because It invented a new, lower-cost way to distribute personal computers: mad order sales backed by telephone hot-lines offering technical advice. The barriers to entry, in this layer are low. Dell already has a host of imitators snapping at its heels, though this has yet to slow the firm.

Parts of the industry are left out of this map. They include memory chips, which work with microprocessors, and other components such as disk drives: peripherals such as printers and modems, and services, a fast-growing part of the industry. Services come in various forms: "outsourcing" (performing all the data-processing chores for a corporate or institutional customer); consulting (advising customers on how to reorganise their businesses to take advantage of computers): and systems integration (making a customer's computers work together).

One reason why such services are growing so quickly is that big corporate customers are confused by all the products being offered by the new computer industry. Every lame company from the old computer industry--IBM. DEC. Bull, Unisys and others--is now hoping to win much of the service business by exploiting the large marketing and service operations which they built to support the sale of their large machines.

Although memories, peripherals and services are sizable and growing segments of the computer industry, they are supplements to its core, represented in the diagram, where most of the strategic choices must be made and the technology is mov- ing fastest. This is where me industry's crucial competitive Battles are being fought, and where firms will emerge as either victors or victims.

Do it my way

THE noisiest of those competitive battles wall be bout standards. The eyes of mast sane people tend to glaze over at the very mention of technical standards. But in the computer industry, new standards can be the source of enormous wealth, or the death of corporate empires. With so much at stake. standards arouse violent passions. Much of the propaganda pumped out by individual firms is aimed at convincing customers and other firms that their product has become a "standard".

It is for the customer's sake that standards matter. All industries need them, simply because so many things made by different companies must fit together to be of any use. Standards can either be a set of specifications and practices, or they can be embodied in a single product, without some standards or other, no new industry can get off the ground. In its first two decades, the car industry Fought fiercely over standards for everything from the size of nuts and bolts to whether vehicles would have steering wheels or boat-like tillers. Eventually, car makers managed to establish standards for enough key features and components to reach a mass market. But even in today's cat industry, not everything is standardised, as anyone who has fum- bled with the controls of an unfamiliar rental car will know.

The world is full of standards that are entirely neutral, belong to nobody and simply make life easier (sliced bread fits mast toasters). But standards, and who owns them, have always been a cat- ical competitive issue in the computer industry.

In the old industry, standards were mealy set. and owned, by, vertically integrated manufacturers and used to lack in customers and lack out competitors. By contrast, the new computer industry has rejected, at least rhetorically, such proprietary standards in favour of "open" standards to which all firms have access. Customers like open standards so much that they have resisted the old computer industry adopt them as well. Mainframe and minicomputer makers now declare themselves keen advocates of openness, although mast of their products still do not connect easily to those of rivals.

Once established, open standards offer what economists call "network economies", which can entrench standards even when they are not the best available or abreast of the latest technology. In the case of personal computers, such network economies were enormous. Customers had strong reasons to buy machines built to the standard because they felt confident that large amounts of software would be available to run on them, and that most other machines would be compatible. Conversely, even tiny software firms suddenly had what promised to be a huge market at which to aim. Firms like Lotus, WordPerfect and Borland racked up hun- dreds of millions of dollars-worth of sales from a stogie hit product. A "virtuous cycle" had been created. As mote software was written for IBM-com- patible personal computers, more people wanted these machines. AS more machines were sold, demand for software increased.

And yet open standards represent a trade-off for both computer firms and that customers. If the standard is embodied in a component that contains much of the value of the finished product--as it was in Intel's microprocessor--firms which use that component can find it difficult to differentiate their products without violating the standard. The result in personal computers has been brutal price competition. And any standard, open or not, eventually becomes an obstacle to technological progress. With both microprocessor and software technology changing so rapidly, this conflict is especially acute in the computer industry. As a result, even agreed standards tend to be undermined by new technologies within a few years, compelling companies to pay the high costs of abandoning the old standard, and sparking a struggle among firms to establish a new one.

Open standards have become the religion of the new computer industry, to which everyone pays obeisance, so perhaps it is not surprising that schismatic wars have broken out over the meaning of the term. All firms now claim that their products are open, but that those of their competitors are not. "The eskimos have 21 words for snow. These guys need 21 words for `open'." says Tim Bresnahan, an economist at Stanford University. Generally there are two ways to set open standards: through negotiations by several firms or by the adoption of a standard established by a single firm.

There have been repeated efforts to establish multi-firm standards, especially for operating-system software. Most of these have been based around an operating system first developed by AT&T called Unix. different versions of which can run on every size of machine from mainframes to personal computers. AT&T pledged to license the basic programming code of Unix to any other company at minimal cost. But most multi-firm efforts have failed for the simple reason that the participating firms cannot trust each other. There are now many rival versions of Unix sponsored by various firms from IBM to Sun Microsystems, all of which are, to a significant degree, incompatible with one another, although all are promoted as open.

Standard-bearers

In fact, widespread adoption of a single firm's product is the only way truly open standards have been established in the new computer industry. "The irony of open standards is that they have to be based on a monopoly, which then earns enormous amounts of money for whatever firm owns it, observes Todd Hixon, a technoloey analyst with the Boston Cosulting Group. The most famous- some industry executives would say infamous-example is Microsoft's MS-DOS operating-system software, which now runs on 80m PCs.

Any firm in Microsoft's position has to make some difficult decisions. Owning a standard product is like possessing any monopoly; it is worthless unless a firm can derive income from it. But if a firm charges too much, other firms will rebel, and either try to copy the product or pay the cost of switching to another as a standard. Microsoft has played this delicate game with consummate skill. It has charged too little for MS-DOS to spark much rebellion, while assiduously encouraging other software firms to write application programmes which run on it. As the power of microprocessors grew, the company was also careful to develop new versions which took advantage of the new chips, but which were compatible with all earlier versions, so that users never had to scrap all their old software when they bought a new personal computer. Today nearly all PCs, except workstations and Apple's machines (which use Apple's proprietary operating system), come with MS-DOS already installed. Nevertheless, even MS-DOS's days are numbered, because of technological advances.

Every firm in the computer industry, no matter what layer it competes in, now dreams of repeating Microsoft's triumph. "Even as late as 1988 no one in the industry really understood how lucrative owning a standard could be," says David Yoffie, a professor at Harvard Business School and a board member at Intel. "Now everyone sees it. As a result no one is willing to let another company establish it. That is what makes the prospects for profitability so problematic in this industry."

A huge battle is shaping up in operating-system software. Microsoft has a big lead with a product called Windows, which runs on MS-DOS machines and mimics the easy point-and-click icons of Apple's computers. It has already sold more than 20m copies. But IBM is heavily promoting OS/2, its rival to Windows. In network operating-systems, which run on the machines at the centre of client-server networks, Novell has scored a success similar to Microsoft's. Its netware has become the standard. Netware's dominance is unlikely to last a decade, as Microsoft's MS-DOS has done. By the mid- dle of next year. Microsoft has promised to launch a product called Windows NT (for "new technology") to compete with Netware. Meanwhile Taligent, a joint venture established by Apple and IBM, is also working on an operating system that will run both on networks and on stand-alone machines. And many people in the industry believe that some version of Unix will ultimately prevail. In December Novell bought Unix Systems Laboratories from AT&T and 11 minority shareholders, with the obvious intent of making Unix an alternative standard to whatever is offered by Microsoft.

The battle over operating systems will produce the most spectacular fireworks over the next few years. Nonetheless, scores of similar struggles to establish and control "open" standards are occurring in every cornet of the computer industry.

Decisions, decisions

MANAGING any business, from a fruit stall to an oil company, is a complicated task. Demand and price go up and down, competitors disrupt the most carefully laid plans, interest rates fluctuate, laws change, employees blunder. The list of possible calamities is long, that of opportunities lamentably short. And yet for most businesses the rules of the game, and so tile types of calamities or opportunities to be faced, stay much the same for years, or even decades, at a time.

For computer companies, the rules of the game itself keep changing, which multiplies all the normal complexities and risks of running any firm. There are a number of reasons for this. First, the basic technologies of the computer business--microprocessors, memory chips, screens and software-- continue to change quickly, creating new products and altering both the capabilities and pricing of existing products in every layer of the industry, which has knock-on effects in all the other layers.

Second, these technologies are so widely dispersed that predicting which firms will succeed with a new technology, or suddenly spring up as a new competitor, is far more difficult than in most other industries. A vast corps of electronics engineers and programmers have been trained over the past two decades. Their job mobility and willingness to take risks are legendary. Firms have little difficulty recruiting talent, and lots of new companies are formed every year. Even in microprocessors--a capital-intensive, specialised business--intel now faces competition from a small, Texas-based firm called Cyrix, started by two engineers in 1988, in the late 1980s Toshiba, a distant also-ran in personal computers, shocked the industry when its laptop models, not those of the industry's established leaders, became a hit everywhere. Toshiba, in its turn, was shocked when its early lead was eroded by a wave of imitators, most of them American.

Third, far more often than most companies, computer firms are not selling their products to an established market, but trying to create demand for an entirely new product. This involves a lot of sheer guesswork. In 1991 many firms expected pen computers to take the industry by storm (these allow people to enter information by writing with a stylus on an electronic notepad rather than using a keyboard). Since then, sales have been disappointing. Pen computers are now seen as a niche product with limited potential.

Technological change is not unique to the computer industry. But its pace, and the fact that it is happening in so many areas at once, may be. So to succeed, or even to survive, computer firms now have to put an inordinate amount of effort into doing three things:

ï½· Collaborating. The multi-layered structure of the new computer industry and the large number of firms it now contains, mean that any single firm. no matter how powerful, must work closely with many others. Often this is in order to Detain access to technology or manufacturing expertise. A web of thousands of joint ventures, cross-equity holdings and marketing pacts now entangles every, firm in the industry. Even firms with a revolutionary product need to create a "community" of other firms to exploit it, argues James More, of Geo Partners, a computer-industry consultant. "A firm has to attract help from all others in the value chain and deny it to competing communities of firms."

Successful alliances are notoriously tricky to achieve in any business. An added complication in most computer-industry deals is that few alliances are exclusive. Firms usually retain the right to do business, or strike a similar alliance, with other firms. And alliances are often between firms that compete fiercely in other areas. Apple and IBM are jointly developing new chips, operating systems and multi-media products, all critical to both firms' future. But Apple, like much of the rest of the new computer industry, also remains determined to steal business from IBM's corporate customers.

Given so many uncertainties, many of the grandest computer alliances predictably fail. The most spectacular break-up has been between Microsoft and IBM. The two spent wars and hundreds of millions of dollars jointly developing OS/2, an operating system to replace MS-DOS. When an early version of OS/2 sold poorly in 1990, Microsoft threw most of its marketing efforts behind its own Windows operating system. IBM felt betrayed. It has since signed marketing pacts with Microsoft's rivals. Novell and Lotus.

ï½· Watching other firms. Firms must keep a close eye on the actions of others, even those with whom they have no formal alliance or do not compete. Most firms depend on those In other layers of the industry, to succeed. If a firm stumbles in one layer, it can deal a mortal blow to firms in other layers. In the 1980s Compaq owed much of its extraordinary success in the market for assembled PCs to Intel's willingness to provide it with carly supplies of its latest microprocessor. But when new RISC microprocessors designed by Run and others looked as if they would leave Intel's chips Par behind. Compaq had no choice but to join Microsoft, and 19 other firms whose products depended on Intel's chips, in a consortium tailed ACE assembled to search For an alternative. Alarmed. Intel accelerated plans to bring out a new generation of microprocessors and eventually persuaded ACE's members that it could keep up with RISC technology. In late 1992 ACE was dispanded. Similarly. Lotus bet that IBM would succeed with OS/2 after its split with Microsoft. When sales of Windows took off and those of OS/2 sputtered. Lotus was not prepared with a Windows version of 1- 2-3, its popular spreadsheet program. As a result, Lotus lest market share to the Windows version of Excel. Microsoft's own spreadsheet. Lotus is still scrambling to catch up.

ï½· Monitoring technology. Like any type of company, computer firms must track their direct com- petitors to avoid being caught off guard by a technological breakthrough. However, technology is changing so fast in the computer Industry that just watching competitors is not enough. Our map of the industry on page 18 will probably be completely redrawn in a few years. New technologies promise to blur the boundaries between today's layers, pitting supplier against customer and turning firms which now happily co-operate into competitors.

Chip makers are learning to put more and mote of the electronic bits in complete machines on to a single piece of silicon along with the microprocessor. This is a direct threat to assemblers of personal computers, who are already struggling to find ways to add value to machines and so earn profits.

In the next few years, microprocessors themselves will become so powerful that they will incorporate many of the functions of current operating- system software, or run "emulations" which allow them to operate with software written for other types of microprocessors. There is disagreement about whether such emulations will be efficient enough to be invisible to the user. or whether they. will slow computers down. If they prove efficient and invisible, the implications for the microprocessor and software markets are difficult to fathom.

It could prove a blow to Microsoft, Novell and others which sell operating systems. Or it could liberate them from specific chip makers. Microsoft has said that Windows NT will run on a venery of RISC microprocessors, as well as on Inters chips. If all operating systems can run on all microprocessors. then the latter could become a commodity, like memory chips, sold primarily on price. On the other hand. emulation may also allow operating systems to mimic each other. Which would mean that software written for MS-DOS or Windows could run easily on Unix. Apple s operating system. and any others mat come along. This, in turn could make operating systems indistinguishable commodities.

Microsoft's boss, Bill Gates, dismisses any such idea as really, reall wrong. It ignores the idea that there is incredibly innovative work going on in operating systems." That is just the problem, complain many application-software firms. They worry that Microsoft will incorporate so many functions into its new operating systems that there will be little opportunity for them to innovate and add value. Though Microsoft is a big application-software firm itself, it is wary of alienating other application firms because it does not want them to devote their best efforts to writing software for rival operating systems. On the other hand, the intensely competitive Mr Gates finds it difficult not to seize an opportunity, sitting right under his nose.

Meanwhile Lotus is attempting to appropriate some of the functions, and value, of the layer below the application-software layer where it normally competes with a product called Notes, which allows users on large networks of personal computers to communicate easily and share databases. Though ostensibly an applications program--it runs on various operating systems--Notes is also something of an operating system itself. Lotus is encouraging other firms to wine applications which, in rum. exploit the capabilities of Notes. Already industry pundits are calling Notes "middleware". an entirely new industry layer between operating systems and applications. Microsoft plans to incorporate many of the same features offered by Notes into its new server operating-system. Windows NT. when it appears this year, which might promptly squeeze middleware out of existence.

Given all these risks, one question companies must continually ask themselves is whether or not they should be operating in the laver above or below their main business--in other words, how vertically integrated should they be? There is no single answer to this: and. because of technological changes, whatever answer looks right today may be wrong tomorrow. Apple and Sun Microsystems claim that being in both hardware and software is an advantage for making both work together, though they are now devoting the bulk of their R&D efforts to software. Mr Gates says being in both hardware and software is too risky, though he sees an advantage in being in both operating systems and application software. IBM, which is in every layer and every, market, is floundering.

Four friends, four rivals

MOST of the computer companies mentioned in this survey art American. That is no oversight. The industry's direction has been set in America, which is also where mast of the innovation occurs. Although large, the European and Japanese computer industries are rooted in their home markets. During the past two difficult years, European companies have lost market share even at home to American rivals. Japanese companies have fared better. Their lead in memory-chip production and their skills at low-cost manufacturing have brought them modest gains in the share of world hardware sales. But at home they are facing an onsiaught from American and other companies in the personal- computer market. And they nave vet to make much of a dent in software.

Japanese firms could play a bigger role, especially if mobile, hand-held computers become as big a nit as many people predict. Until that happens, however, strategic choices made by American firms will determine the direction of the industry This article examines the strategies of four or tile most significant American firms.

As the leader of the old computer industry. IBM faces enormous challenges finding its place in the new industry. Its efforts to do so will be one of the great dramas in modern corporate history. For IBM. 1992 was a disastrous year. Even worse, it capped a precipitous slide in the company's fortunes. Since 1985 its share of the total computer market, including hardware, software and services. has slid from 30% to less than 19%. Its market capitalisation has dropped like a stone, from a peak of $106 billion to 1987 to $27 billion.

IBM has already, made wrenching changes, cutting its workforce by a quarter to 300,000 and reducing manufacturing capacity by 40% since 1986. This year it is cuing another 25.000 people. It has also reorganised its business five times over the same period. In December 1991 it announced the most drastic reorganisation of all. the division of the company into 13 autonomous businesses, each with its own balance sheet, profit and loss account and financial targets. These businesses are supposed to establish an internal marks, with prices equivalent to those offered by outsiders, which should expose hidden subsidies and obvious laggards. Whether it will make ISM as a whole more competitive is debatable. "Markets and companies are very different things," says David Teece, a professor at Haas School of Business at the University of California's Berkeley campus. "ISM may not get the full benefits of either."

Despite its troubles, IBM remains huge. Its sales are more than three times the computer sales of Fujitsu, the world's second-biggest computer company. And amid the carnage of the past few years, it has scored some remarkable successes. Its mint- computers and workstations, two markets which it entered years too late. have said well.

Today technology is coming out of IBM'S vast R&D establishment much more quickly, producing a wave of new products in 1992. It has also launched a range of low-cast personal computers and copied the direct marketing and telephone technical support pioneered by Dell. The inadvertent creator of the new computer industry, IBM has now had to adopt the new industry's ethos, pledging to make all its products connect easily to those of other companies. It has collaborative deals with thousands of firms, including many of those whose success has done so much to destroy its hegemony. Lotus, Novell. Apple and others are all too happy to let IBM's huge salesforce flog their products to large corporate customers. Whether Big Blue gets much out of this is Itself difficult to say. The company's top managers say they are now determined to give customers whatever they want, even if that means selling someone else's product, or helping a customer scrap an expensive ISM mainframe in fa- your of a cheaper network of workstations and personal computers.

If it is to remain a single entity. IBM has no choice but to adopt this strategy of being all things to all customers. But IBM is competing against thousands of specialised firms aiming at every corner of its market and every lover of its value-added chain. Even if its mainframe patrimony, still its biggest business, survives longer than sceptics suggest. IBM may not be able to remain either so vertically integrated or so ubiquitous in an industry which is fragmenting quickly. IBM executives seem at a loss about what to do next. A new boss at the company may break it up.

Cool operator

Microsoft has replaced IBM as the industry's most feared and admired company. Its financial perfor- mance has been spectacular, largely because of its near-monopoly in PC operating systems, which ac- count for 40% of its sales. Other firms in the industry. are gunning for Microsoft. Complaints by rivals of anti-competitive behaviour have sparked an investigation of the firm by America s Federal Trade Commission. which could cause Microsoft big headaches m the future. IBM's alliances with Apple. Novell and Lotus are clearly designed to deny Microsoft dominance of the next generation of operating systems, whether on stand-alone machines or the servers at the heart of client-server networks. Sun Microsystems aims to do the same thing. Mr Gates shrugs off criticisms from other firms. "Customers don't care much about whaler other companies in the industry are comfortable with us," he says. "Who gives a damn? He rubbishes rival products. IBM'S OS/2 operating system, he states flatly, will be dead in two years.

Behind the outward taunting, Mr Gates has dis. played great skill and determination in building Microsoft into a powerhouse. Ironically, in the 1980s the firm's application programs for its own operating system sold poorly. Lotus, WordPerfect and Borland became the leading suppliers of (respectively) spreadsheets, word-processors and computers. So Mr Gates directed Microsoft's efforts to writing application programs for Apple's Macintosh computer, despite the fact that Apple's machines were rivals to those that ran its own MS-DOS operating system. The strategy worked brilliantly. Microsoft's applications helped make Apple's Macintosh computer sucdeed, although MS-DOS machines continued to account for 90% of PC sales. An then, when it abandoned its joint effort with IBM on OS/2 in favour of its own Windows software, Microsoft also threw its weight behind its Windows-compatible applications, catching Lotus, WordPerfect and Borland off balance and winning a big chunk of sales in the IBM-compatible market as wll. It is waging a fierce price war to hold on to its new prize.

Through it is portrayed by rivals as the firm to beat, Microsoft will never dominate the new computer industry the way IBM did the old one. Mr Gates seems to recognise this when he admits that the company will not be able to maintain its phenomenal profitability indefinitely. But if Microsoft scores a success with Windows NT in network operating systems similar to what is has achieved with MS-DOS and Windows, it could enjoy another extraordinary decade. Its break with IBM will make that much harder to accomplish. Thousands of the corporate customers switching to client-server networks still have IBM machines. Many of them are bound to listen to IBM salesmen peddling network operating systems from Novell, IBM and others. If it fails, Microsoft will still be a formidable competitor in the application-software market. But that is business, as even Microsoft has discovered, is more hit-or-miss than operating systems.

Intel, like Microsoft, has been one of the biggest beneficiaries of the brutal price wars in the PC market, which has boosted sales of machines, most of them containing an Intel microprocessor. For more than two decades, Intel has been one of the most innovative firms in electronics. But it has never been able to rest on its laurels for long and that is truer today than never before. Despite its supremacy in microprocessors, Intel faces competition from two types of companies: those making replicas of its microprocessors and those making alternatives using a different design.

Advanced Micro Devices (AMD), which Intel licensed as a second supplier at the behest of IBM a decade ago, has since managed to "clone" versions of Intel's most popular chip, the 386. This has driven prices down rapidly in the past year. So Intel has shifted its marketing efforts to the more powerful 486 chip. It has also invested heavily in both R&D and new production capacity to accelerate the launch of its next-generation chip, called the Pentium, whichis expected in the next few months. Thought Intel has dragged AMD in and out of court claiming patent infringement to slow it down, AMD says it will have a 486 chip ready this year. Cyrix, whose designs are not clones on Intel's but are meant to run all the same software, is already selling a 486 chip and promises a Pentium-compatible chip later this year.

RISCy business

A bigger threat could come from firms making incompatible RISC microprocessor, which have traditionally been faster than Intel's. These are made by a range of firms, from Texas Instruments to Fujitsu, working to the designs of Sun Microsystems. IBM, which invented RISC technology in the 1970s is also working on a new RISC chip, called the PowerPC, with Apple and Motorola. Intel claims the Pentium will match the performance of any RISC chip, but also offer compatibility with all the software currently written for all earlier Intel chips-a big advantage. In any case, Intel has little choice but to keep running this particular race. The industry layer above, personal computers, is crowded with its own customers and an even harder market in which to make money.

Apple has always bucked the new computer industry's trends. It makes both hardware and software to its onw proprietary designs (Motorola makes its microprocessors) and, until recently, charged a premium for its products in a price-concious market. Its computers have been easier to use than those of rivals, and they have superb graphics. Apple has been the leading firm in American schools and the desk-top publishing market. Otherwise, most of its machines have been sold to individuals, not firms. Its customers are famously loyal.

The constant price-cutting by other PC makers eventually made Apple's strategy untenable. Alarmed by Apple's shrinking market share, Hoh Sculley, its chief executive, reversed the company's direction in 1990, slashing prices, cutting costs and pushing new versions of its computers out of the door faster. "We realised that we didn't have a survivable company," he explains. Apple's market share has recovered. Mr Sculley claims that it is now the world's biggest manufacturer of personal computers. The company scored a hit with its elegantly designed laptop computers, called Powerbooks, last year.

Mr Sculley also concluded that, however innovative or price competitive, Apple needed a partner. In 1991 he announced a wide-ranging alliance with IBM. During 1993 this is supposed to put Apple into the client-server market with a new Unix-based network operating-system and a RISC-based workstation powerful enought to act as a server. The alliance is also supposed to help Apple to crack the corporate market by making it easier to link Apples to IBM minicomputers and mainframes. In addition, the two firms have established joint ventures to develop multi-media software-mixing text, video, voice and graphics simultaneously on the same machine-and "object-oriented" software (chunks of standard software which users can assemble to make their own applications).

Mr Sculley is already looking far beyond any of these venues. "In five years Apple will be heavily into services, software, and systems integration. The wars today are all about hardware prices. Microsoft's monopoly of the system software. Intel's control of microprocessor technology; that's all going to be ancient history," he says. The future he predicts will be one in which people will have access to all forms of information-new, television, business data-wherever and whenever they want, using hand-held "personal digital assistnats", a kind of computerised mobile telephone. Apple announced the first tamily of these, called the Newton, last year, but has yet to bring any of them to market. By 2000, says Mr Sculley, the consumer-electronics, television, telecommmunications, entertainment and news industries will have merged into a single market, with sales of $3.5 trillion a year. Apple will thrive in this market, he claims, because it knows better than most firms how to make complex technology easy for ordinary people to use.

Lost horizons

Many fo Mr Sculley's competitors accuse him of being glib. Yet most of them also share his vision of the future. Over the next decade, most information, including television, is expected to assume the same digital form as computer data. And the spread of mobile telephones is already bringing mobile computing in its wake. Some of the latest laptop computer can laready communicate over the airwaves. More powerful chips seems certain to put enormous computing power into machines small enough to fit into the palm of the hand. More powerful software will make computers of any size and shape easier to use. The cost of storing, transmitting, manipulating andd analysing data will drop sharply. Most big computer firms believe that this means the computer industry will overlap with a variety of others.

One way to portray Mr Sculley's vision is the product map below. It was drawn by Apple on the basis of work done at Harvard University's Centre for Information Resources Policy. The horizontal axis moves from (on the extreme left) products which act as containers or creeires of information, but have none themselves, to (on the extreme right) products whose value resides almost entirely in the information they contain, not their physical form. The vertical axis moves from tangible products at the bottom to intangible services at the top. The map plots where various industries will overlap or converge, so the area convered by each industry does not reflect its relative size in terms of sales.

Mr Sculley is not alone in seeing dazzling opportunities for computer firms in these spreading blobs. Motorola is already selling pagers which instantly alert a customer to any news item on his industry, or any other subject he specifies. IBM and Sears jointly operate Prodigy, a service which provides consumer and business information to PC users in America. Mr Gates also wants Microsoft to get into the business of providing time-sensitive information, and he has established a separate venture of his own to purchase reproduction rights to thousands of images.

But the product map of the future could just as easily mean a collection of tough new competitors for the computer industry. Particularly ominous is the vast area covered by consumer electronics. Japanese firms such as Sony, Matshshita and Sharp are veterans at miniaturisation, low-cost manufacturing and selling huge volumes of low-margin products to a global market. To manufacture its Newton products, Apple has already had to turn to Sharp. Publishers and news services such as Reuters, Dow Jones and Associated Press are unlikely to let computer firms into their business without a fight. And the biggest winners of all may be the telephone companies and cellular-phone operators. They already make more money transmitting faxes than the firms which make fax machines. In November AT&T bought a 33% stake in McCaw, America's biggest cellular operator, and it already owns stakes in EO, a start-up company making mobile, hand-held computers, and General Magic, an Apple spin-off developing software for such devices.

Mr Sculley argues that the winners in this vast new market will be whoever "owns customer relationships". With consumers needing guidance through the thicket of new technology, he may be right. But consumer-electronics companies, telephone firms and news services have just as much chance of forging such relationships as computer companies and they are accustomed to talking in a language the average person understands, unlike most computer companies.

One thing seems certain: the new computer industry will never return to the stability, or high profits, of the old one. Too many companies now have access to the technology-and to the customer. The best strategy will still be to persuade other firms to use your product as a standard. As the industry collides with the telecommunications, publishing and consumer electronics industries, there will be standards battles galore. The search for alliances will become more frantic. Launching new products will require even greater leaps of faith. Nobody really knows how many people will want so much information at their fingertips, what price they will pay for it or what they will want to do with it. When firms find the right combination of features, they will strike it rich. But scoring one-off success will be easier than sustaining competitive advantage over the long term.

If the industry does not make enough profits to pay for R&D and earn a proper return on its invested capital, technological progress will eventually slow down. Even so, the future path of underlying technologies in both chips and software seems clear for some years to come. So if one firm drops the baton, another may well pick it up, make the R&D investment and try its luck. This heralds even tougher time ahead for the computer industry. For its customers, nothing could be better.

TO

APPENDIX TO MEMORANDUM OF AMICI CURIAE
IN OPPOSITION TO PROPOSED FINAL JUDGMENT
IN CIVIL ACTION NO. 94-1564 (SS)
SIGNED BY GARY REBACK

Windows stars at SD91

BY RAY WEISS

Santa Clara. Calif. -- The tenor of last week's System Development Conference was clear evidence that Microsoft Corp.'s Win. dows is well on its way to becoming the dominant operating- software platform for personal computers. SD91. here. was essentially a Windows show: More than one-third of the vendors in attendance had Windows-related products. Developers flocked to see Windows products, while software vendors launched the second wave of Windows 3.0 development software.

A typical reaction was that of Craig A. Snow. manager, software engineering, at Sophia Systems Inc. (Palo Alto. Calif.): "Everybody is going to Windows. It's inevitable. Everybody is looking for the right tool or vehicle to build Windows products."

Microsoft's dominance of Windows development tools was challenged by a number of tool vendors. Archrival Borland International debuted its next- generation Borland C + + product for Windows. which can build Windows programs without the heretofore required Microsoft System Development Kit (SDK). Jenson & Partners Inc. (JPI) announced its integrated set of Topspeed compiler/tools for Windows and DOS. JPI's tools, too. are complete Windows tool kits.

But Microsoft (Redmond. Wash.) is fighting back by preparing a new set of tools for reiease this year. To hold the fort in the meantime. Microsoft integrated its SDK and C6.0 C compiler, and dropped the combined price by 25 percent.

Breakthrough product
Borland's C + + is considered by many Windows programmers to be a breakthrough product--easier and faster to use than the older Microsoft C6.0 and SDK tools.

"This is the tool I've been looking for." said B.J. Safdie. a technology analyst with Sony Corp. (Woodcliff. N.J.).

Borland C + + has a fully integrated development environ- me, nt. including the Turbo Debugger, which can run in a DOS window in Windows' standard protect mode. The package includes the interactive WhiteWater Group Resource Toolkit. with which to build Windows applications resources (bit- maps, fonts, dialog boxes, etc.)-- a job normally handled by the Microsoft SDK resource editors. Many developers welcome Borland's offering.

The new C + + package supports Windows code. Users can build Windows programs, including DLLs (dynamic linked libraries). Additionally, Borland C + + minimizes compilation time by precompiling program header (.h) files. This saves tune. for some .h files, like Windows.h (used in all Windows programs), have more than 20.000 lines of code.

Interestingly. as Borland Challenges Microsoft. Borland itself is being challenged by Jensen & Partners. a spinoff of Borland International. Its CEO. Niels Jen- sen, was one of the cofounders of Borland.

JPI's Topspeed Professional Techkit targets Borland's traditional strength: Turbo Pascal. "Unlike our competitors." said Jensen. "our Pascal compilers are ISO compatible, as is our C compiler."

The new compilers announced by JPI at SD91 brought a new tack to PC software tools. JPI debuted four compilers for Windows development: C. C + +. Modula-2 and ISO Pascal. Unlike any other PC compilers, all of these run in a single environment (as DLLs) and share a common code generator Users can buy and add as many compilers as they want. Addition. ally. they can compile mixed code concurrently, and the libraries are shared, i.e.. C or Pascal programs can access Pascal or C li- brary procedures/functions.

What's more. the JPI compilers feature some technical break-throughs, including virtual pointers (typed pointers, which, when de-referenced, cause a function to be called) and DOS-based dynamic linking with DLLs (an overlay manager that uses the Windows DLL format). Also included with the languages is a pre-emptive. multitasking kernel that runs on top of DOS.

`We'll be there'
Microsoft is busily working on its own advanced tool sets. "You can bet that we will be there with next- generation tools." said Fred Gray. Languages general manager at Microsoft. The company is working at both better Windows development tools and a C + + compiler. Additionally, the company already has a 32-bit compiler as part of the new SDK for OS/2.

Many analysts expect Microsoft to field that 32-bit compiler for Win. dows, undercutting Borland and JPI. whose compilers are still 16-bit architectures, despite the fact that many developers are now running on 32-bit 386 and 496 machines.

Microsoft actually helped Borland in getting its Windows product out. "We have a tool-independent program," said Gray, "that treats our own languages group the same as any other ISV (independent software vendor). Microsoft is out to get Windows accepted and will help competitors like Borland. In fact. we get Windows and other operating system releases the same tune as do the ISVs."

Other vendors at SD91 presented products that support the emerging Windows development market. These include 32-bit compilers from Zortech (C + +) and Warcom (C). as well as Windows GUI (graphical user interface) builders, such as Professional WindowsMaker from Blue Sky Soft- ware Corp. (Las Vegas. Nev.) and VZ Programmer for Windows.

Additionally, two key Windows products bowed that fill critical needs for Windows developers: Pcsteam, a hardware ICE for Windows that monitors 386 systems out to 33 MHz with a fully compliant Code View debugger, and Distinct, the first TCP/IP package for Windows--it includes Berkeley Sockets. RPC/X.DR and NFS, Linking Windows applications to the Unix networking world.

Getting attention
Windows is attracting a lot of attention. "Windows provides a full graphics environment," said Isadore Sobkowski, principal. Knowledge Associates Ltd. (Riverdale. N.Y.). "It's a perfect base for our new generalized expert system. ACE."

Another company. Expert-Ease Systems Inc., is moving its process-control software to Windows. ï½· "People want Windows--it's a nstrig market." said Dave Kuhlman. senior software engineer. Expert- Ease (Belmont. Calif.). "But I will continue to develop using OS.2-- you can just do a lot more with 0S,2 than with Windows."

Many programmers accept Windows as inevitable. "Windows has the market attention." said Sony's Safdie. "But it's a lot like those kits people used to buy and put on a Volkswagen. making it look like a Maserati or some luxury car. Under the hood is still a Volkswagen."

Ronald Surratt, principal, C Carp Designs (Laytonville, Calif.), plans to use Windows as a user-interface for software tools. "Windows is here and accepted. It Lakes care of the graphical user interface as well." Surratt will combine Windows with Small for development. "You can do an awful lot with a small amount of Smalltalk code: with Windows it minimizes development time," he said.

But there are others that cannot live with Windows internals. "Windows is not deterministic." said Christopher Bajorek, president of Telephone Response Technologies Inc. "We do real-time voice systems and have built a pre-emptive multitasking operating system on top of DOS for our needs." Bruce Wallace. a development engineer at Quantum Institute, at the University of California at Santa Barbara, uses OS/2 for real-time control of a free electron laser.

TO

APPENDIX TO MEMORANDUM OF AMICI CURIAE
IN OPPOSITION TO PROPOSED FINAL JUDGMENT
IN CIVIL ACTION NO. 94-1564 (SS)
SIGNED BY GARY REBACK

Inventing-and reinventing- the proprietary architectures for open systems is critical to competitive success.

How Architecture Wins Technology Wars

by Charles R. Morris and Charles H. Ferguson

The global computer industry is undergoing radical transformation. IBM, the industry's flagship, is reeling from unaccustomed losses and is reducing staff by the tens of thousands. The very survival of DEC, the industry's number two company, is open to question. A roll call of the larger computer corn. panies - Data General, Unisys, Bull, Olivetti, Siemens, Prime-reads like a waiting list in the emergency room.

What's more, the usual explanations for the industry's turmoil are at best inadequate. It is true, for example, that centralized computing is being replaced by desktop technology. But how to explain the recent troubles at Compaq, the desktop standard setter through much of the 1980s? Or the battering suffered by IBM's PC business and most of the rest of the desktop clone makers, Asian and Western alike?

And the Japanese, for once, are unconvincing as a culprit. The fear that Japanese manufacturing prowess would sweep away the Western computer industry has not materialized. True, Japanese corn. panies dominate many commodity markets, but they have been losing share, even in products they were expected to control, like laptop computers. Earnings at their leading electronics and computer companies have been as inglorious as those of Western companies.

Explanations that look to the continuing shift in value added from hardware to software, while con- taining an important truth, are still too limited. Lotus has one of the largest installed customer bases in the industry. Nevertheless, the company has been suffering through some very rough times. Meanwhile, Borland continues to pile up losses.

Nor are innovation and design skills a surefire recipe for success. LSI Logic and Cypress Semiconductor are among the most innovative and well- managed companies in the industry, yet they still lose money. Design-based "fabless," "computer. less" companies such as MIPS have fared very hadly too. MIPS was saved from bankruptcy only by a friendly takeover. And Chips and Technologies is in dire straits.

Government protection and subsidies are no panacea either. The European computer industry is the most heavily subsidized in the world but still has no serious players m global computer markets.

Scale, friendly government policies, world-class manufacturing prowess, a strong position in desktop markets, excellent software, top design and innovative skills-none of these, it seems, is sufficient, either by itself or in combination with each other, to ensure competitive success in this field.

A new paradigm is required to explain patterns of competitive success and failure in information technology. Simply stated, competitive success flows to the company that manages to establish proprietary architectural control over a broad, fast- moving, competitive space.

Architectural strategies have become of paramount importance in information technology because of the astonishing rate of improvement in microprocessors and other semiconductor components. The performance/price ratio of cheap processors is roughly doubling every eighteen months or so, sweeping greater and greater expanses of the information industry within the reach of ever-smaller and less expensive machines. Since no single vendor can keep pace with the deluge of cheap, powerful, mass-produced components, customers insist on stitching together their own local system solutions. Architectures impose order on the system and make the interconnections possible.

An architectural controller is a company that controls one or more of the standards by which the entire information package is assembled. Much current conventional wisdom argues that, in an "open-systems" era, proprietary architectural control is no longer possible, or even desirable. In fact, the exact opposite is true. In an open-systems era, architectural coherence becomes even more necessary. While any single product is apt to become quickly outdated, a well-designed and open-ended architecture can evolve along with critical tech. nologies, providing a fixed point of stability for customers and serving as the platform for a radiating and long-lived product family.

Proprietary architectures in open systems are not only possible but also indispensable to competitive success - and are also in the best interest of the consumer. They will become increasingly critical as the worlds of computers, telecommunication, and consumer electronics continue to converge.

Architectures in Open Systems

In order to understand architecture as a tool for competitive success in information technology, consider first the many components that make up a typical information system and the types of companies that supply those components.

Take the computer configuration in a typical Wall Street trading or brokerage operation. Powerful workstations with 50 MIPS millions of instructions per second) - comparable to the power of standard mainframes-sit on every desk. The workstations are connected in a network so they can communicate with each other or with several others at a time. Teams of workstations can be harnessed together to crunch away on a truly big problem. Powerful computers called servers support the network and manage the huge databases-bond pricing histories, for instance - from which the workstations draw.

Such a modern network will be almost entirely open, or externally accessible by other vendors; critical elements, from perhaps as many as a hundred vendors, plug interchangeably into the network. The workstations themselves are from companies like Sun Microsystems, Hewlett-Packard, and IBM, or they may be powerful personal computers from Apple or any of a number of IBM-compatible PC manufacturers. IBM and Hewlett-Packard make their own workstation microprocessors; most workstation or personal computer makers buy microprocessors from companies like Intel, Motorola, Texas Instruments. LSI Logic, AMD, and Cyrix. Almost all the display screens are made in Japan by Sony, NEC, and many other companies; the disk drives come from American companies like Seagate or Conner Peripherals. The memory chips are made in Japan or Korea. The network printers will typically have laser printing engines from Japan or, if they are high-performance printers, from Xerox or IBM; the powerful processors needed to control modern printers will come from AMD, Motorola, or Intel. The rest of the standardized hardware components on the network, like modems, accelerator boards, coprocessors, network interface boards, and the like, will be made by a wide variety of Asian and American companies.

The network will have many layers of software, most of it "shrink-wrapped" from American companies. The operating system-the software that controls the basic interaction of a computer's corn. ponents-may be a version of AT&T's UNIX, specially tailored by the workstation vendor, as with Sun and IBM, or it may come from a third party, like Microsoft. Many vendors, like Lotus and Borland will supply applications software. The complex software required to manage the interaction of the servers and workstations on the network will, in most cases, be supplied by Novell The software that converts digital data into instructions for printer engines is sold by Hewlett-Packard. Adobe. or one of their many clones. Each smaller element in the system, like a modem or video accelerator, will have its own specialized software, often supplied by a vendor other than the manufacturer.

It is possible to construct open systems of this kind because for each layer of the network there are published standards and interface protocols that allow hardware and software products from many vendors to blend seamlessly into the network. The standards define how programs and commands will work and how data will move around the system- the communication protocols and formats that hardware components must adhere to, the rules for exchanging signals between applications software and the operating system, the processor's command structure, the allowable font descriptions for a printer, and so forth. We call this complex of standards and rules an "architecture."

A small handful of the companies supplying components to the network will define and control the system's critical architectures, each for a specific layer of the system. The architectural standard setters typically include the microprocessor designer (such as Sun or Intel); operating system vendors {possibly Sun or Microsoft); the network system {usually Novell); the printer page-description system (Adobe or Hewlett-Packard); and a small num- ber of others, depending on the nature of the network. Each of these is a proprietary architecture; although the rules for transmitting signals to an Intel processor, for example, are published openly for all vendors, the underlying design of the processor is owned by Intel, just as the design of Sun's operating system is owned by Sun, and so on for Microsoft's Windows/DOS, Novell's Netware, or Adobe's PostScript.

Companies that control proprietary architectural standards have an advantage over other vendors. Since they control the architecture, they are usually better positioned to develop products that maximize its capabilities; by modifying the architecture. they can discipline competing product vendors. In an open-systems era, the most consistently successful information technology companies wall be the ones who manage to establish a proprietary architectural standard over a substantial competitive space and defend it against the assaults of both clones and rival architectural sponsors.

It has been conventional wisdom to argue that users, and the cause of technological progress, are better served by nonproprietary systems architectures. This is emphatically untrue. There are many examples of nonproprietary architectures, like the CCITT fax standard or the NTSC television standard, most of them established by government bodies or industry groups. Because they are set by committees, they usually settle on lowest-common-denominator, compromise solutions. And they are hard to change. The NTSC has been upgraded only once (for color) in a half-century; committees have been squabbling over an improved fax standard for years. Proprietary architectures, by contrast, because they are such extremely valuable franchises, are under constant competitive attack and must be vigorously defended. It is this dynamic that compels a very rapid pace of technological improvement.

Architectural Competitions

The computer industry has been competing on architecture for years. Take the example of the product that established IBM's dominance in the mainframe computer business-the IBM System/360. The 360 was arguably the first pervasive, partially open, information technology architecture. In the late 1960s, once the System/360 be- came the dominant mainframe solution, IBM began to unbundle component pricing and selectively open the system, in part because of government pressure. Published standards permitted competitors and component suppliers to produce a wide range of IBM-compatible products and programs that were interchangeable with, and sometimes superior to, IBM's own. By licensing its MVS operating system to Amdahl, for example, IBM made it possible for Fujitsu, Amdahl's partner, to produce clones of the IBM mainframe. Much of what was not licensed away voluntarily was acquired anyway by the Japanese through massive intellectual property theft.

Hundreds of new companies selling IBM-compatible mainframe products and software placed intense competitive pressure on IBM. But they also assured that the IBM standard would always be pervasive throughout the mainframe computing world. As a result, even today IBM controls some two-thirds of the IBM-compatible mainframe market and an even higher share of its profits, not only for central processing units but also for disk drives, systems software, and aftermarket products like expanded memory. Because they have no choice but to maintain compatibility with the IBM standard, competitors must wait to reverse-engineer IBM products after they are introduced. Typically, by the time competitive products are on the market, IBM is well down the learning curve or already moving on to the next generation. And as the owner of the dominant architecture, IBM can subtly and precisely raise the hurdles whenever a particular competitor begins to pose a threat. For over 20 years, in generation after generation, IBM has played this game brilliantly and won every time.

Ironically, IBM badly fumbled an equivalent opportunity in desktop computing, handing over the two most critical PC architectural control points-the systems software and the microprocessor-to Microsoft and Intel. Since any clone maker could acquire the operating system software from Microsoft and the microprocessor from Intel, making PCs became a brutal commodity business. As a high-cost manufacturer, IBM now holds only about l5 % of the market it created.

In a related error, Compaq made the mistake of assuming that IBM would always control the PC architectural standard. On that premise, the company geared its cost structure and pricing policy to IBM's, only to find itself almost fatally vulnerable when the savage PC price wars of the early 1990s exposed the commoditized character of PC manufacturing. Tellingly, while IBM and Compaq struggle to eke out profits from their PC businesses, Microsoft and Intel are enjoying after-tax margins of about 20%, on sales of more than $4 billion and $6 billion respectively, and together they have more cash than IBM.

For a similar example, consider the case of Lotus. Lotus got its start in a market-spreadsheet software - where products are complex and feature- rich, hardly commodities. And over the years, the company acquired or developed a broad array of other products - Jazz Manuscript, Improv, AmiPro, Notes and Freelance-some of which are technicallyexcellent. Lotus's competitive problem, however, is that these products lack any deep architectural commonality. Indeed, even the embedded spreadsheet software in the company's various offerings ais incompatible from one to another.

Point product vendors like Lotus can be very profitable for a time. However, they are always at risk when an architectural leader changes the rules or the game. For example, while Lotus was accumulating a grab bag of point products, Microsoft was creating an architectural lock on the graphical user interface (GUI) for DOS-based computers. (See the insert "Scenarios for Architectural Competition: Graphical User Interfaces.") And Windows now defines the environment in which Lotus's software must compete. The great power of Windows is that it creates a relatively simple, intuitive, and reasonably uniform interface between a user and a very wide range of applications software. As users become accustomed to the greater ease of Windows, they insist on it, and point product vendors like Lotus are forced to adapt their software to run under the Windows architecture. But Microsoft also offers its own line or point products, like Excel and Word, and since they arguably better exploit the Windows architecture, they are steadily encroaching on Lotus's market share.

The irony is that for a time in the 1980s Lotus had suc a powerful market position that it almost certainly could have established a GUI standard itself. But the company neglected to do so. Such strategic errors spell the difference between an architectural winner and loser.

Principles and Phases of Architectural Competition

There are five basis imperatives that drive most architectural contests:

1. Good products are not enough. Producs distribute architectures and can contribute to the success of an architectural strategy. However, as the case of Lotus suggests, good products alone are not enough. But if the sponsor invests heavily in continuous product improvement, products of only modest capabilities can become the basis for architectural leadership. For example, both Zilog and AMD at various stages in the PC microprocessor contest made Intel-compatible chips that were superior to Intel's own; but neither company matched Intel's commitment to R&D, and both were left behind as Intel rolled out one generation of improved processor after another. Once an architecture is established, it in turn becomes a distribution channel for additional products, with the architectural controller's products holding the favored position.

2. Implementations matter. Manufacturing decisions are playing an increasingly important role in product strategy. But since successful architectures have a high design content and usually a high software content, manufacturing skills by themselves are not sufficient to prevail in architectural competition. Japanese and other Asian companies, for example, despite their great manufacturing prowess, have only rarely established architectural franchises. Generally, they have settled for positions as clone makers or commodity implementors. Perhaps the only area where Japanese companies have established proprietary control over an important architectural space is in video games. But even the leaders in this arena, Nintendo and Sega, are at risk. (See the insert "Scenarios for Architectural Competition: Video Games.")

While insufficient on their own, however, manufacturing skills may well be an essential competence for success in the architectural contests of the 1990s. The reason: implementation is increasingly becoming the key to winning architectural control. In microprocessors, for example, a good implementation can improve performance by a factor of two. That's why architectural leaders like Intel typically make their own chips. By contrast, Sun Microsystems has chosen to focus solely on the design of its SPARC microporcessor, a decision that has been a source of recent difficulty for the company because subpar supplier implementations have compromised SPARC performance. High-quality implementations are equally important in the new generations of hand-held computers. Indeed, the more advanced information technology makes inroads into consumer markets, the more manufacturing skills will prove invaluable.

3. Successful architectures are proprietary, but open. Closed architectures do not win broad franchises. Choosing the right degree or openness is one or the most subtle and difficult decisions in architectural contests. IBM opened its PC architecture too broadly - it should have, and could have, retamed control of either or both the operating system and microprocessor standard. Apple made the opposite mistake of bundling the Mac operating system too closely to its own hardware. Sun, in contrast to Apple, opened its SPARC RISC architecture very early, both to software developers and pro- cessor cloners; it has the lead position in workstations, and its broad base of third-party software support has helped maintain customer loyalty though a series of technical stumbles. Autodesk's computer-aided design (CAD) software for builders is open to add-on third-party packages, like kitchen design tools, and its broad base of supporting software has given it control of a small but very profitable franchise.

4. General-purpose architectures absorb special-purpose solutions. Architectures that cannot evolve to occupy an ever-broader competitive space are dead ends. Wang's lucrative word processor franchise was absorbed by general-purpose PCs. Special-purpose CAD workstations from Daisy, Applicon, and others were absorbed by more general-purpose desktop machines. Special-purpose game machines will, in all likelihood, be absorbed by more general-purpose consumer systems.

5. Low-end systems swallow high-end systems. Minicomputers poached away huge chunks of mainframe territory and were assaulted in turn by workstations and networks. Workstations are under pressure by increasingly high performance PCs. Traditional supercomputers and very high-end mainframes are vulnerable to parallel arrays of inexpensive microprocessors. High-end data-storage systems are similarly under attack from arrays of inexpensive, redundant disks. Although IBM helped create the personal computer revolution, it steadfastly refused to recognize its implications. Until relatively recently, it even called its desktop products division "Entry Systems," ignoring the fact that today's microprocessor-based machines are a replacement for traditional computers, not an entry point or way station to them.

However, managers must keep in mind that even those companies that best follow these principles are not necessarily guaranteed continued success in the marketplace Architectural con- tests typically move through a number of different phases and only those companies that successfully navigate them all, maintaining their pace and direction in the fluid environment of rapidly evolving technologies, emerge as winners over the long term. It's a delicate balancing act, and one that requires ever-increasing flexibility as the technologies mature.

There are five principal phases to architectural competition:

Commitment. Architectural challenges usually emerge from the early-stage chaos of competing point products. Before the IBM PC, personal computers were rigid, closed systems that tended to bundle their own operating systems and applications software. Compaq had the insight that by purchasing a Microsoft operating system identical to that of the PC, it could ride the wave of the PC's success. Microsoft then insisted that all subsequent clone makers buy the same operating system and so seized the critical PC software architectural standard. Microsoft's insight was to realize that it was in an architectural contest and to take the appropriate steps, including steadily expanding the generality and scope of its systems to come out the winner.

Diffusion. Large proms come from broad franchises. Open architectures are successful because they can be broadly diffused. Xerox's Interpress page-description software, which converts digital data into printer instructions, is excellent but can be purchased only with Xerox high-end printers. Adobe, by contrast, has widely licensed its PostScript language and has become the industry standard setter. Intel widely licensed the early versions of its xx86 processors, then sharply restricted licensing of its 386 chip after the Intel standard had become firmly entrenched. IBM, on the other hand. has long resisted diffusing its mainframe and minicom- puter software. Of course, diffusion decisions are not without risk. Once again, balance and timing are essential. For example, Philips licensed its compact disc technology to Sony to increase market penetration. But Sony outperformed Philips and Took, half the market. Philips's standard was a static one that it never developed further.

Lock-to. A company has a "lock" on an architecture when competi- tors are trained to wait until the architectural leader introduces each new product generation. Intel and Microsoft, at least temporarily, seem to have achieved this position in PC markets. Sun was on the verge of a locked- in franchise in workstations but may have fallen short; the performance of its SPARC RISC processor design has been lagging behind the competition, and the company neglected to solidify its franchise by moving rapidly down to lower end platforms.

But lock-in is sustainable only when a company aggressively and continuously cannibalizes its own product line and continually and compatibly extends the architecture itself. This is a strategic choice that many companies find difficult to make. Often, managers become overprotective of the products that brought them their original success. IBM, for example, has frittered away a powerful lock on back-office transaction processing and operating systems. In a misguided effort to protect hardware sales, It has refused to release products, long since developed internally, that would adapt Its best-selling AS400 minicomputer software to the RS6000 workstation. Such reflexive self-protection simply hands over a valuable franchise to the Microsofts and other vendors storming up from the low end.

Harvest. Of course, the ultimate objective of architectural competition is to win a market leader's share of the profits, lust to give one dramatic example, profit margins on Intel's xx86 family of chips are m the 40% to 50% range and account for well over 100% of the company's earnings. But no locked-in position is ever completely safe, and companies must be careful when they harvest not to rest on their previous successes. Indeed, Intel may have harvested too aggressively, drawing out spirited recent attacks by clone makers such as AMD and Cyrix.

Obsolescence and Regeneration. lust as products must be cannibalized, so must architectures themselves. The better the architecture. the longer its lifespan: but sooner or later every architecture, no matter how well designed becomes ob- solete. And before it does, the market leader must be prepared to move ahead, to do away with the old and introduce the new. Industry, leaders often fall to cannibalize their old architectures, but although nothing IS more painful, to do so is absolutely necessary. Otherwise, competitors quickly move to create and introduce rival franchises, and these eventually dominate the industry. IBM's failure to cannibalize Its mainframe and minicomputer franchises provides a stark example of the catastrophic effects of waiting too long.

DEC provides another example. The company developed outstanding RISC products very early. But DEC declined to cannibalize its profitable VAX- VMS architecture because its VMS operating system, the source of its franchise, was tightly integrated with its aging VAX hardware. Predictably, DEC was beaten out by vendors such as Sun Microsystems and Microsoft, which didn't hesitate to move in with their newer, more powerful alternatives. (The main developer of DEC's advanced systems, Dave Cutler, IS now In charge of developing NT for Microsoft.)

There are three lessons here. First, with better architecture DEC could have kept VMS alive longer. If VMS had been "portable," that is, not restricted to VAX hardware, DEC could have ported VMS to other vendors' hardware, making VMS an industry standard. Indeed, the company could have used RISC technology itself without losing its VMS franchise. Second, DEC would have been better off cannibalizing itself, rather than waiting to be cannibalized by others.

The third lesson, though, is the most important. As DEC's experiences with VMS and IBM's mistakes with the mainframe and minicomputer franchises show, the cultural and organizational structures useful for managing traditional, closed, integrated businesses will not work for companies that intend to compete with architectural strategy. In fact, we believe that architectural competition is stimulating the development of a new form of business organization.

This new structure, which we call the Silicon Valley Model, has major implications both for information technology and for many other industries. The model still young and rapidly changing and although Microsoft probably comes closest to company ?? ??.

Managing Architectural Competition: The Silicon Valley Model

The Silicon Valley Model arose a decade ago when early architectural competitors noticed that they faced the same problems in managing organizations that they faces with technologies and architectural strategies.

In retrospect, this is not surprising. Architecture responds to the same imperatives in both systems and organizations. It reduces complexity. It permits clean separation between centralized or specialized functions. It enables management of unpredictability and change; individual technologies, components, or procedures can be switched without the need to redo everything. For similar reasons, good architecture facilitates experimentation and competition; once the framework is specified, multiple approaches can compete without jeopardizing compatibility. An finally, a standard architecture permits many systems and organizations to be developed independently and still work together gracefully.

As an organizational paradigm, the Silicon Valley Model therefore has several characteristic features and advantages. Following are the most important:

1. Organizational architecture and decision making that mirror technical architecture. Any organization should develop and use good technical architectures. But Silicon Valley Model firms take an additional step: the structure of the firm itself mirrors the technical architectures it uses.

Thus, for example, Microsoft is structured so that its existing systems software and applications software are managed separately, as are new architectural efforts such as NT. In this manner Microsoft can diffuse its applications acrosss multiple operating systems (both its own and others, like the Apple Machintosh), while also marketing its operating systems by courting other vendors' applications. The two businesses can work largely independently, yet only Microsoft gains the benefits of their synergism. Most decisions can be made directly within the organization responsible for the relevant architectural domain; this minimizes complex vertical and horizontal debates.

2. Meritocracy and direct feedback. Silicon Valley Model firms enable and force direct performance feedback, at levels ranging from individuals to business units. At Microsoft, team members rate each other periodically In peer reviews. Outstanding performers are rewarded; laggards are warned, then fired. Technical expertise is required for a large fraction of senior management, and communication occurs directly between the relevant parties, unbuffered by hierarchy.

By contrast, performance ratings in traditional bureaucracies are determined by managers at higher levels, and compensation is rarely based on longterm corporate performance. The process is often heavily politicized; dissent is suppressed, and incompetence goes unpunished.

Architectural competition also exposes Silicon Valley Model firms to another form of peer review - product competition. To succeed as industry standard setters, firms must license their architectures to competitors, while also developing critical products themselves. As a result, each layer of the firm land of the architecturel is exposed to direct competition and market feedback. Hence although Microsoft controls Windows, application groups still compete individually: Excel against Lotus and QuattroPro, Word against WordPerfect and AmiPro, and so forth. Architectural leadership provides an advantage, but prevents a cover-up. Silicon Valley Model firms are structured so that excellence is the only defense.

3. Clean boundaries, both internal and external. In architected corporate structures, organizations can create and dissolve alliances rapidly, both internally and externally. Organizations are very flat, and development groups have simple, clean interfaces to each other determined by architectural boundaries. Architecture and point products can be kept apart. Moreover, products can invisibly incorporate architected "engines" developed by other organizations, including competitors. For example, a start-up called InfoNow has organized alliances involving itself, Microsoft, publishers, computer vendors, and other software companies. InfoNow packages software products, together with reviews and samples of them, which are preloaded for free on computers, the software products, however, are encrypted. Users can sample them. read reviews, and then purchase them by telephone, which triggers electronic decryption. Adding new software packages is trivial.

4. Internal proprietary control of architecture and critical implementations, externalized commodities and niches. Silicon Valley Model firms seek to externalize the maximum possible fraction of their total system, while carefully controlling those at. eas required to establish and hold an architectural franchise. Thus core development of the general purpose architecture is always internally controlled. So usually are critical product implementations, which cover the broadest markets and are required either for early diffusion or later harvesting. Silicon Valley firms also carefully manage their dependencies, so as not to become unilaterally dependent on architectural competitors.

On balance, however, Silicon Valley Model firms are much less autarkic than traditional large firms. Niche products, commodity components, and at. chitectures controlled by others are outsourced, and/or relegated to licensees. In fact, Silicon Valley firms actively seek to commoitize regions not under their control.

This yields several benefits. For one, companies can focus on what they do best and on the efforts critical to architectural success. For another, broad outsourcing and licensing create competition among suppliers and licensees, which broadens the market and benefits the architectural leader. PC pricee wars delight Intel, Microsoft, and Novell; IBM and Compaq take the heat.

Interestingly, this contradicts the 1980s conventional wisdom that firms should avoid broad, cost- sensitive markets in favor of high-price niches. In fact, the broad market is the strategic high ground, if it is covered by a proprietary architecture. Niche product vendors can make profits, but they will remain minor players.

5. Migration and evolution over time. Just as architectures evolve and eventually become obsolete, so too with organizations. Thus the firm's internal structure and external alliances evolve along with its architecture and market position. As new layers are added to an existing architectural position (Windows on top of DOS, then NT underneath Win- dows), new organizations are created: a similar situation occurs when an architecture must be cannibalized. Some Silicon Valley Model firms will soon face cannibalization; it will be interesting to see how they do.

Broader Implications of the Silicon Valley Model

The Silicon Valley Model is very much a product of a few companies in the computer sector, just as mass production was invented by Ford and just-in- time production by Toyota. And as in those cases, we believe that the Silicon Valley Model will diffuse throughout the broader information technology sector as the computer, telecommunications, information services, and consumer electronics industries merge.

In addition, however, as industrial competition in all industries becomes more complex and technological change accelerates, the model may have important effects upon many other fields. We think that it provides a framework that allows proprietary leaders in general to have the greatest span of control and profitability with the least complexity and smallest size. In fact, we think that the model is appropriate for small and large companies alike; it does, however, penalize unnecessary size. (Microsoft, with fewer than 15,000 employees, has a market capitalization equal to IBM's.) We will therefore close with an example of how architec- tural strategy and the Silicon Valley Model could have been used more than a decade ago, by Xerox.

Xerox became a large, global company through a single proprietary technology-xerography. Xerographic "marking engines" are the core of photocopiers, printers, and facsimile machines, all or which Xerox invented. But Xerox chose to exploit its control of xerography using the traditional strategy of integrated companies.

Where Xerox felt it could not develop products profitably itself, it simply left the market vacant. As a result, when the company's patent position eroded, Japanese competitors took the bulk of the blossoming low-end markets for personal copiers, laser printers, and fax machines. Xerox's market share declined from nearly 100% to about 30%.

Instead, Xerox could have developed an architecture for a broad family of machines and control systems, including interfaces for scanners, document handlers, and "finishers" for collating, stapling, and binding. It could have licensed its technology to other firms, and/or sold them xerographic engines. It could have developed products for core markets, leaving others to niche companies.

Every few years, the company could have changed or enhanced its architectures to improve its products and competitive position. The result could have been a Microsoft-like position, with Xerox holding the lion's share of the profits in a highly competitive, dynamic market-yet one under its own effective control. We think that similar strategies are available to companies in other complex in- dustries-aerospace and machine tools, among others. If so, the information sector's strategic and organizational innovations might prove as interesting as its technology.

Graphical User Interface (GUIs) are the software that permits users to maneuver around applications visually - for example, issuing commends by pointing to icons - providing a simple, consistent method of working with many different programs. The evolution of the GUI market provides a dramatic example of the dynamics of architectural competition.

The original GUI was developed at Xerox's famed Palo Alto Research Center (PARC) and unveiled with the Xerox Star in the early 1980s. The Star was a brilliant achievement for its time - a high-performance, if very expensive, easty-to-use networked workstation. But it was a completely closed system; there was no published applications-program interface, so no one but Xerox could supply software to run on the Star. It's appea; was therefore far too limited ever to become a pervasive desktop standard.

Steve Jobs adapted the Star technology to Apple, but it took several tries before Apple began to make inroads in the GUI arena. Apple's first try was Lisa, a substantially closed system that failed to attract any market share. The company got it more nearly right with the Machintosh. At least in later incarnations, the Mac has been hospitable to third-party software developers. It is considerable less expnsive than the Lisa and has a superb operating system/GUI architecture. But Apple has still sharply limited its distribution potential by insisting on bundling its architecture with only its own, second-rate hardware. The Mac is hardly a failure, but had Apple and its microprocessor partner, Motorola, could have exercised the same architectural control over personal computing that Microsoft and Intel do now.

The operating system/GUI architectural struggle is far from over and will be one of the most heated competitive arenas of the 1990s. IBM OS/22/0 is technically excellent but suffers from a very late start; Microsoft's brand new NT system, which will run Windows applications, will raise the hurdles yet again. A variety of UNIX-based standards are alternatives to systems derived from the original DOS. And IBM and Apple have joined forces on a next-generation operating system/GUI in their Taligent partnership.

As the ongoing GUI contest suggests, architectural battles are fast-moving, hotly challenged, and rarely completely settled. The rewards to a winner, however, can be great.

The home video game industry, dominated by Nintendo and Sega is a serious industry. Some 30 million American homes, or about 70% of all home with a child between the ages of eight and twelve, own a video game. Both Nintendo and Sega sell video game consoles (basic, 16-bit, 286-level computers) with tightly bundled operating systems. Game software is developed by independent vendors under tightly controlled licenses but distributed only through the two companies' networks at hefty markups. Profits flow from game sales, not consoles.

Bundled architectures are ripe for attack my more open systems, just as the Apple II was overwhelmed by the IBM PC. In fact, a number of American companies have targeted the game market. Electronic Arts, for one, has won a copyright suit allowing it to reverse-engineer Sega's operating systems. The availabilityof a Sega system clone would break theat company's hold over game software and open up console manufacturing to clones. Another company, 3DO (formerly the San Mateo Software Group), has plans to release a powerful consumer-oriented poerating syustems that will be ideally suited for games. Specifications have been provided to a number of Asian manufacturers including Matsushita, a 3DO investor, for a CD-ROM-based console. Existing best-selling games, presumably, could readily be adapted to the new system. 3DO's ojective is to own a Windows-like architectural franchise in the consumer world.