Eyes on the Prize

Microsoft launched Windows 95 amidst much fanfare. And the hoopla was well deserved. Though the company’s perpetual critics derided the product for its lack of technical excellence, they were, quite simply, wrong.

Alas, the public afforded Microsoft not a moment to rest on its laurels. For no sooner had it launched Windows 95 than the public began to clamor for easy Internet access and Web browsers. Now this particular consumer demand hardly caught Microsoft by surprise. In fact, Microsoft released a set of Internet access tools, including Internet Explorer 1.0 and MSN (Microsoft’s first attempt to compete with AOL) concurrent with the Windows 95 launch. But no one really took these products seriously— particularly the Internet Explorer browser. Microsoft knew that its future lay with Windows. The Internet properties could develop into valuable divisions at some point, but they were hardly central to the company’s growth strategy. More to the point, though, the public wasn’t terribly interested. They were happy using software circulated by their Internet service providers (ISPs) for access to gateways other than MSN. And they were even happier eschewing Internet Explorer for competing browsers, notably but not solely Netscape’s Navigator.

At some point between August 24, 1995, when Microsoft first shipped Windows 95, and December 7, 1995, when Bill Gates gave his now- famous Pearl Harbor Day speech, Microsoft experienced an epiphany.

As luck would have it, a fair amount has been written about these browser wars. This material gave me the opportunity to test some aca­demic lessons in the real world. It allowed me to see how real software companies put the various principles of software engineering into prac­tice. It also provided insights into what actually drove Microsoft’s behav- ior—the combination of marketing, product development, and business strategy that guided monopolist behavior.

Various authors offered various takes on the matter. Competing on Internet Time took the perspective of management science. Cusumano and Yoffie spent a good deal of time getting to know key players at both Microsoft and Netscape, followed the battle as it unfolded, and tried to distill the strategic approaches that worked from those that failed.¹⁹ And while they did explain a good deal about strategy in fast-moving soft­ware markets, they didn’t dwell much on one of the differences between the two companies that I found most fascinating. Netscape was about twenty years behind Microsoft on the corporate personality development scale. Microsoft started in the mid-1970s as an ambitious company led by talented technologists intent on propelling discovery and innovation forward. That description also fit Netscape in the mid-1990s. But Microsoft matured in some dangerous directions. Within ten years, it had become a creative, consumer-focused software company interested both in developing new products and in knocking out its competitors.

Ten years after that, it was destroying both products and markets to ensure that all software innovation was shoehorned through Windows. That was the Microsoft that Netscape faced, and I wanted to understand how that deadly transition had occurred. Cusumano and Yoffie, while interesting, were of little help in that investigation. I had to turn elsewhere.

David Bank, who spent many years covering Microsoft for the Wall Street Journal gave me the answers in his insightful description of the internal struggles for Microsoft’s soul. Microsoft is a big place, full of people with definite ideas and nonnegligible egos. At various points in the company’s history, it made strategic choices that favored some prod­ucts over others. As you might guess, every product had its internal backers, and every strategic juncture led to a vociferous debate—fre­quently resolved way up at the top. Bank’s Breaking Windows revealed the nature of those debates.²⁰ In particular, it answered one nagging tech­nical question that continued to bother me through everything else that I had heard or read about the browser wars.

Now in all fairness, the trial did raise a related question, namely whether Windows and Internet Explorer were one product or two. While this distinction may be meaningless to computer scientists, it can have a fair amount of legal significance. Tying, a form of leveraging illegal under the antitrust laws, occurs when a monopolist refuses to sell consumers its popular “must have” product unless they also buy some other junk that they may or may not want. If the junk that I don’t want is part of the product I need, I’m stuck. But if the two are distinct products, then we have a tying claim—and a violation of the antitrust laws.

No one questioned, for example, Microsoft’s right to insist that I buy the parts of Windows that display icons if I wanted to buy the rest of the program. Had I called Microsoft and asked them to ship me a reduced-cost version of Windows that simply displayed the words “recycle bin” on the desktop where the picture of the bin normally sits, Microsoft would have been well within its rights to refuse—and the gov­ernment would have backed Microsoft. On the other hand, if Microsoft refused to sell me Windows unless I also agreed to buy its keyboard when, quite frankly, I wanted to use a competing Logitech keyboard, that would have been an obvious case of tying—a not-too-subtle attempt to drive Logitech out of the keyboard business. But where was Internet Explorer? It seemed to be somewhere in between. If it was an inde­pendent product distinct from Windows, Microsoft’s insistence on pack­aging them together was illegal tying.

But because of my techie roots, I was also unsatisfied with that debate. I still wanted to know why and how Microsoft decided to integrate this product when it did. Evolutionary integration is usually slow. Any good software designer knows enough to keep new functions as application modules as long as possible, and to integrate them downward into the platform slowly and deliberately. Microsoft followed this principle with Windows. It took ten years to migrate Windows from a middleware application down to a part of the platform—and another six to finish the job, because until Windows XP’s 2001 release the integration was not seamless. Internet Explorer made it into the platform from its initial launch—and most objective software engineers probably would tell you that it was incorporated prematurely. In other words, Windows and Internet Explorer should have been two products at the time of the trial, whether they were or not.

But techie question or not, the issue of premature integration has just as much legal significance as the direct question about tying. One of Microsoft’s most devastating defenses to all of the charges levied against it was always: “Hey, we’re just a bunch of dumb software engineers making engineering decisions. Don’t you want us to develop good prod­ucts to sell to consumers?” Judges hate to second-guess business decisions. Questions about the timing of product integration are engineering ques­tions. No judge in the country would try to examine the internal work­ings of Microsoft’s engineering and design teams to decide whether or not they made the right decisions about modular design, product integration, or the evolution of the platform sitting at the translation frontier. And since they knew Judge Jackson wouldn’t do that, the attorneys spent less time than they might have looking into it, leaving me with this annoying, nagging question: Why did the integration happen prematurely?

My guess was that somewhere inside Microsoft, a battle had emerged between two factions.

I surmised that strategic marketing, not engineering excellence, had driven Microsoft’s decision to integrate Internet Explorer into Windows prematurely. And any competent judge would know how to interpret that preference. Intent is often important in the law, and it’s certainly important in complex antitrust trials. The government exerted a good deal of time and effort showing that Microsoft intended to control the market by strong-arming distributors and competitors. A strategic deci­sion to integrate a product prematurely in order to leverage a monopoly in the desktop platform market into the browser is not an engineering decision. It’s an economic decision to engage in behavior that runs explic­itly counter to the antitrust laws.²¹

So it seems that my little techie question could have had some intense legal consequences as well. But I couldn’t find the answer anywhere— until I found Breaking Windows. Bank described a deeply entrenched culture of strategic market manipulation at Microsoft. He didn’t describe a single instance of engineering concerns overriding market strategy. In fact, he related a widespread view among Microsoft’s own developers that “the company sacrificed innovation for ‘strategy,’ the complex set of hooks and lock-in techniques that Gates invariably insisted on to steer customers toward Microsoft’s end-to-end product line and keep them from being able to competitive products—and which customers hated for the very same reason.... The ‘strategy tax’ could be deeply demor­alizing.”²² Monopolistic business strategy, not software engineering, guided Microsoft’s product-design decisions. That’s hardly surprising. A corporation owes its first duty to the considerations of the market. Its entire raison d’etre is to extract maximum profits from the market and to shower them upon its shareholders. Bank demonstrated that Microsoft never let product quality stand in the way of strategic con­cerns; Microsoft simply deferred quality considerations until it had resolved the more important strategic issues.

Given the choice between spending a few months improving a product selling in a competitive environment and dedicating those months to securing the market, Microsoft will always choose to secure the market. With the market secure, Microsoft will return to improve the product. Eventually, consumers will have exactly one technically competent product on the market—Microsoft’s. The product may arrive later than it should, and there will only be one choice, but if it works reasonably well, who would complain? Certainly not most consumers; they continue to believe that things should be as they are because, after all, when were they ever better? No, no one other than the government and its market cops are likely to complain.

So Microsoft set out to do what it did best, and it left the government to worry about problems and complaints. Microsoft acted like a monop­olist. Microsoft came up with a number of clever techniques designed to make Netscape either play ball or die—where playing ball, of course, meant playing by Gates’s rules. Microsoft tried honey. It tried vinegar. It tried making Netscape some offers it couldn’t refuse. But Netscape did refuse, and Microsoft seethed. But Netscape was neither cowed nor amused. Its key technical leader, Marc Andreessen, made a number of public comments displaying the sort of brashness that the software world had not seen since, well, since a certain young Mr. Gates had made his presence known some twenty years earlier; for example, Andreessen reportedly boasted of his plans to turn Navigator into a platform and thereby to reduce Microsoft’s newly released crown jewel to a “slightly buggy set of device drivers.” Netscape’s chairman, Silicon Valley veteran Jim Clark, took a slightly subtler approach. He asked his attorney, Gary Reback, to let the government know what was going on and to see if they were interested in reopening their file on Microsoft.

Clark was hardly alone in believing that government action was needed. Many of the biggest names in Silicon Valley agreed with him. Sun’s CEO Scott McNealy quickly emerged as one of Microsoft’s harsh­est critics. Though Reback wrote his letter two-and-a-half years before McNealy spoke at the 1999 World Economic Forum in Davos, Switzer­land, his concerns were known early on. In Davos, McNealy told the audience (which included Bill Gates) that “Microsoft is a planned economy. Left unfettered, unscrutinized, [and] unchecked, monopoly power can be leveraged into other businesses.”²³ But as far back as May 1996, when a reporter asked McNealy whether or not he was concerned that Microsoft might abuse its license of Sun’s platform-independent Java language to develop a proprietary standard, he replied “We’re always worried people will try and hijack the standards on the network and make something that says, ‘Looks best under such-and-such a browser,’ or, ‘Only runs under Explorer,’ or ‘Only runs and gets access to the fol­lowing database from our browser.’”²⁴ At the time, though, he believed that the openness of the Internet architecture would protect Java. He was wrong. About a year later, Sun sued Microsoft for corrupting the integrity of Java by developing precisely such a version.

Industry support is one thing, though. While the Antitrust Division isn’t a particularly political agency, it is part of the executive branch. Joel Klein, by then promoted to Assistant Attorney General for Antitrust, knew that if he moved against a company as big, as prominent, as rich, and as important as Microsoft, there would be fallout somewhere. His bosses, Janet Reno and Bill Clinton, would end up taking at least some of the heat. He needed political cover. Senator Orrin Hatch of Utah (proud home of Novell), the staunchly conservative Republican Chairman of the Senate Judiciary Committee whose primary relation­ship with the Clinton administration had been to block its judicial nominees, gave Klein the cover he needed when Hatch came out in favor of investigating Microsoft. The attorneys general of twenty states gave Klein even more cover. Klein had broad, bipartisan political backing to go with that of much of the high-tech industry. Thus insulated, he braced himself for the final battle.