Summary

  • In 1995, Microsoft tied the speed of Netscape’s access to a Windows API to a proposed “special relationship”; the API arrived after Windows 95 and Internet Explorer had launched.
  • The appeals court accepted middleware as a nascent threat to Windows while reversing the separate browser-monopoly claim for lack of a defined market and evidence of entry barriers.

Analysis

A browser did not need to replace Windows to threaten the thing that made Windows hard to leave. It needed to give developers another set of interfaces—APIs they could use to build software across operating systems. That possibility put the Microsoft case on two tracks: what might weaken an operating-system monopoly, and what evidence could prove an attempted monopoly in browsers.

The distinction first appears in a June meeting.

On 21 June 1995, Netscape executives asked Microsoft for technical information needed to make Navigator work well on Windows 95. The U.S. Department of Justice’s later court record says Microsoft proposed a closer “special relationship” and linked the timing of access to the Remote Network Access API to whether Netscape accepted it. Netscape declined. Microsoft released the API in late October, more than three months later. The district court found that the delay pushed Navigator for Windows 95 until substantially after Windows 95 and Internet Explorer launched in August, leaving Netscape outside most of the holiday selling season.

An API is not a product launch, but its delivery date can decide when a product gets one. In this episode, compatibility information sat in the same negotiation as market access. The court record dates the API delay and the release lag but does not measure how many users or developers Netscape lost because of them; that evidence would be needed to assign the delay a market effect. The negotiation put a technical gate in the platform owner’s hands.

The 1999 findings explained why that gate mattered. Windows attracted applications; applications made Windows more useful; that usefulness attracted more users and developers. The court called this reinforcing cycle the applications barrier to entry. Navigator and Java could expose their own APIs above the operating system. If developers wrote applications for those interfaces, their programs could travel across more than one operating system. Navigator had been ported to over fifteen systems, the district court found.

That was the threat on paper. It was not a ready-made replacement for Windows. The same findings said middleware then exposed too few APIs for independent software vendors to build full-featured productivity applications that relied on those interfaces alone. Reaching the point where middleware-based applications could constrain Windows pricing would take years. A browser could be a platform in the making while still depending on an operating system to manage hardware and devices.

The district court’s 2000 conclusions put several claims in Microsoft’s loss column: unlawful tying, maintaining the operating-system monopoly, and attempting to monopolize the browser market. The remedy ordered a breakup into operating-systems and applications businesses. Microsoft appealed.

The D.C. Circuit drew two lines in June 2001. It affirmed part of the finding that Microsoft maintained its operating-system monopoly through anticompetitive means. On that claim, a nascent middleware threat mattered; the conduct did not have to wait until browsers could already replace Windows. But the browser-market attempt was a different claim. It required a defined market and substantial barriers to entry that could make monopoly power possible. The appeals court found that the plaintiffs had not established either and reversed that finding without sending it back for another try.

This was not one court saying “the browser mattered” and another saying “the browser did not.” The same appellate opinion treated a future platform threat as relevant to conduct protecting Windows, while demanding proof of present market structure for a claim that Microsoft was trying to monopolize browsers. Those are different questions, asked under different parts of the antitrust law.

The breakup disappeared too, for reasons that should not be folded into the browser ruling. The appeals court found defects in the remedies process, including the lack of a hearing on remedies-specific facts and an inadequate explanation of the order. It also vacated the remedy after finding that the trial judge’s improper press contacts and public comments created an appearance of partiality; the court said it found no evidence of actual bias.

In November 2002, a consented final judgment replaced the proposed split with conduct rules: limits on retaliation against OEMs, room for OEMs to offer competing middleware and other operating systems, and disclosure of specified APIs for interoperability.

The record leaves a harder lesson than “bundling is bad” or “a browser will replace the desktop.” A platform can defend its position through the interfaces and distribution routes that determine what developers can build and where users encounter it. The court let a nascent threat count on one track and rejected a browser-market monopoly theory on another for missing proof. Windows remained the operating system underneath the browser; the contest was over who could write software above it.

Sources