Continuing my series of posts on the tenth anniversary of the Microsoft antitrust case, I want to look today at the government’s theory of the case, and how it looks with ten years of hindsight.
The source of Microsoft’s power in Windows was what the government dubbed the “applications barrier to entry”. Users chose their operating system in order to get the application software they wanted. Windows had by far the biggest and best selection of applications, due to its high market share (over 95% on the PC platform). To enter the PC OS market, a company would not only have to develop a competitive operating system but would also have to entice application developers to port their applications to the new system, which would be very slow and expensive if not impossible. This barrier to entry, coupled with its high market share, gave Microsoft monopoly power.
The rise of the browser, specifically Netscape Navigator and its built-in Java engine, threatened to reduce the applications barrier to entry, the government claimed. Software would be written to run in the browser rather than using the operating system’s services directly, and such software would run immediately on any new operating system as soon as the browser was ported to the new system. Cross-platform browsers would reduce the applications barrier to entry and thereby weaken Microsoft’s Windows monopoly. The government accused Microsoft of acting anticompetitively to sabotage the development of cross-platform browser technology.
The imminent flowering of browser-based applications was widely predicted at the time, and the evidence showed that top executives at Netscape, Microsoft, and Sun seemed to believe it. Yet we know in hindsight that things didn’t unfold that way: browser-based applications were not a big trend in 1998-2003. Why not? There are two possible explanations. Either the government was right and Microsoft did succeed in squashing the trend toward browser-based applications, or the government and the conventional wisdom were both wrong and there was really no trend to squash.
This highlights one of the main difficulties in antitrust analysis: hypothetical worlds. To evaluate the key issue of whether consumers and competition were harmed, one always needs to compare the actual world against a hypothetical world in which the defendant did not commit the accused acts. What would have happened if Microsoft had simply competed to produce the best Internet Explorer browser? It’s a fascinating question which we can never answer with certainty.
What actually happened, after Microsoft’s accused acts, the lawsuit, and the settlement, in the years since the case was filed? Netscape crumbled. The browser market became quiet; Microsoft tweaked Internet Explorer here and there but the pace of innovation was much slower than it had been during the browser war. Then the open source browser Mozilla Firefox arose from the ashes of Netscape. Firefox was slow to start but gained momentum as its developer community grew. When Firefox passed 10% market share and (arguably) exceeded IE technically, Microsoft stepped up the pace of its browser work, leading to what might be another browser war.
We also saw, finally, the rise of browser-based applications that had been predicted a decade ago. Today browser-based applications are all the rage. The applications barrier to entry is starting to shrink, though the barrier will still be significant until browser-based office suites reach parity with Microsoft Office. In short, the scenario the government predicted (absent Microsoft’s accused acts) is developing now, ten years later.
Why now? One reason is the state of technology. Today’s browser-based applications simply couldn’t have run on the computers of 1998, but today’s computers have the horsepower to handle browser-based apps and more is known about how to make them work. Another reason, perhaps, is that Microsoft is not acting against Firefox in the way it acted against Netscape a decade ago. A new browser war – in which Microsoft and Firefox compete to make the most attractive product – is the best outcome for consumers.
Life doesn’t always offer do-overs, but we may get a do-over on the browser war, and this time it looks like Microsoft will take the high road.
What’s with the cryptograms?
> Another reason, perhaps, is that Microsoft is not acting against Firefox in the way it acted against Netscape a decade ago.
Oi. You had me going until this statement. The reason Microsoft is not giving Firefox the treatment it gave Netscape is that it *can’t*. Firefox and Netscape are two separate entities that are different on a fundamental level. “Cutting off Firefox’s air supply” would have precisely no effect — it’s an open-source application that can exist independently of any company, and a good portion of the revenues used to pay people to work on it come from voluntary user donations. This really can’t be used as evidence of a changed Microsoft; it’s like comparing visible light to neutrinos.
i agree with JEF, the outlook of future income streams always promotes new technologies and breakthroughs
In 1998 you also didn’t generally have the high bandwidth, low latency and low jitter that you have today (relatively speaking). How many K get burned, for example, to send and deliver the data required for one character of live preview at a web site that provides such?
One huge barrier to rich web applications was the horribly fragmented support for Javascript/DOM across web browsers. Not only were there huge differences between Netscape and IE, there were significant differences between browser versions. Until the pre-IE6 and Netscape4 browsers fell out of use, programmers were essentially stuck with writing the browser side of a web application at least twice (and again when a new browser came out). Took until roughly 2001-2003 for programming to become a more reasonable task.
You still have to do some things twice, as some aspects of IE (events in particular) differ from the W3C standards. This choice by Microsoft makes web applications more difficult. Whether this was an intentionally anti-competitive choice is not clear, but that is the result.
Other factor is that the published examples for programming using client-side Javascript/DOM were (and many still are) fairly horrible. Could be this is due in part to more experienced developers avoiding dynamic client side behaviors until programming for the browser was less of a mess (certainly that was my choice).
The increase in broadband penetration and introduction of viable advertising networks (Google AdSense) were also major factors in the move to web apps.
Arguably, circa-1998 computers had enough horsepower for modern web apps (those machines were comparable in some ways to modern iPhones). What’s really changed since then is the integration of JavaScript with the browser document object model (DOM). With that, JavaScript can be used to dynamically update a web page on the client side, allowing applications like Gmail to flourish. Of course, the other big change is the rise of massive, scalable server infrastructure: a crazy new idea in 1998 and standard operating procedure today.
Minor irony: Microsoft was on the leading edge of enabling modern web apps in the browser, notably introducing the predecessor of the modern XmlHttpRequest into IE5 to support Outlook Web Access. Modern AJAX apps are all built on this.