Bonus: Split Up Microsoft and Managing the Verdict

"Not later than four months after entry of this Final Judgment, Microsoft shall submit to the Court and the Plaintiffs a proposed plan of divestiture." —Final Judgement June 7, 2000

Bonus Post

On June 7, 2000 the Judge in the US v. Microsoft case announced his verdict in the trial—Microsoft should be split into two companies, one for operating systems and one for applications (with a number of details to work out). Today in 2021 is the anniversary of that decision. Hardcore Software is emerging chronologically, but given the anniversary and a chance to show some of the less nostalgic and more operational events transpired I wanted to share these two sections as a bonus post.

Quite a few whole books have been written about the trial and Microsoft through the era. None of them had much help or support from the Microsoft side. I wanted to share some of the context from my perspective, not about the legal issues, but about living through such an ordeal. One way or another, I found myself at least peripherally involved in the litigation going back to when I first started working as BillG’s Technical Assistant when the first FTC investigations took place. Even before that litigation, as I’ll write about below, was part of Microsoft’s employee narrative.

By way of context, in mid-2000 my title was Senior Vice President of Office and I managed the product development and marketing team for the suite of applications and servers, about 2000 people total. I reported to Bob Muglia who reported to Paul Martiz. Steve Ballmer had been CEO for all of a few months. PCs sold about 100 million units in 1999. Microsoft had successfully pivoted to an internet-centric company and more importantly and enterprise company. Hardcore Software will have a good deal on the transition to enterprise.

Thoughts or comments are welcome.

Leave a comment

Split Up Microsoft

On June 7, 2000, the verdict was delivered. I read the verdict on my Blackberry flying back from a Windows conference. In-flight connectivity didn’t exist, but the Blackberry magically worked over the pager network, downloading a few sentences at a time while I avoided looking like I was using electronics in the air.

Judge Thomas Penfield Jackson ordered the breakup of Microsoft into two companies. Originally filed May 18, 1998, the antitrust trial of the Department of Justice and twenty State Attorneys General decision was a low point for Microsoft and, for the first time, the Office team. In his initial findings of facts in November 1999, the judge found that the company held a monopoly.

In his order, one company was to be the Windows company, and the other was to be made up of the rest of Microsoft including Office. The case had finally hit close to home.

Looking back this was a long road. The investigation started 7 years earlier, after the Federal Trade Commission (FTC) dropped its case in a deadlocked vote and passed the authority to DOJ. I remember the early meetings from when I was working for BillG and how “crazy” all this felt at the time. While perhaps at the highest level the complaints did not change, the details and reasoning changed as the company saw more success. It is amazing to think how much the industry changed over this time—Windows 95 and the internet came to be—and still many said the industry shifts were just starting, yet the case was still there.

As a practical matter, once a trial started things didn’t change unless, well, we lost. Losing took a much longer time (for both sides). Plus, there was always an appeal.

Still, that day in June, it obviously felt like we lost . . . badly.

Microsoft had always been comfortable in the context of litigation, perhaps owing to BillG’s upbringing as the son of a prominent Seattle attorney. The earliest days of the company were characterized by a lawyerly Open Letter to Hobbyists, penned by BillG in 1976.1 In the letter, he argued that software should be a royalty-based product like music. The letter was controversial in a world where all the money was in hardware with freely bundled and shared software but ushered in the pure-play software company we now know.

The early software industry wrestled with how law applied to this new product, a product required for hardware, dreamed up like art, but manifested in proprietary digital encoding.

In 1988, Microsoft found itself in what it would describe as a straightforward contract dispute, and what Apple would characterize more broadly as an intellectual property dispute in Apple versus Microsoft.2 Apple agreed to license elements of the Macintosh software for use in Windows 1.0, partially in exchange for an effort to secure Microsoft applications for Macintosh. The case was front and center of the industry as Apple claimed a right to the “look and feel” of the Macintosh, which seemed to many rather unbounded, though obviously their product was unique on many levels. In a key ruling for all of software the court stated that, "Apple cannot get patent-like protection for the idea of a graphical user interface, or the idea of a desktop metaphor." While ultimately resolved in Microsoft’s favor in 1996, based on contractual terms, the litigation served to condition employees to the hurry up and wait, and the ups and downs, of the winding nature of the US legal system. For years at the annual company meeting someone inevitably submitted a question for BillG about the case and every year he would say there is nothing new but that we felt good on the merits. In between those times, the various motions and courtroom events were rather baffling to non-lawyers, somewhat like trying to watch a cricket match for the first time and not being sure if something good was happening and for which team it was happening.

Litigation was a significant part of the industry in the early days of software as the rules of the road were established for software patents, copyright, and contracts. Another closely watched case was Lotus Corporation, a giant, suing Borland International, an upstart, for copyright violation in 1990. Borland had essentially cloned the interface of Lotus 1-2-3 and expanded upon it in its Quattro Pro product, in an effort to smooth the transition from 1-2-3 to Quattro by providing a compatibility mode. This case had profound impact on the ability for upstarts to enter an existing market because whether it was user-interface or API, providing compatibility by reverse-engineering (without having access to source code or trade secrets) was key to expanding the industry. This case was decided in Borland’s favor, allowing for the copyright of the Lotus implementation but not the expression of user interface in Borland’s product.3

While these suits were winding their way through the system, Microsoft’s rise to the largest software company and its new power position as an unabated leader continued. From the outside, Microsoft had all the appearances of a growing software empire. From the inside, Microsoft was paranoid and felt everything was fragile and could evaporate at any moment—just as we had seen happen to the fortunes of nearly every technology company before us. Somewhere between fragile upstart and unstoppable force was the truth. It would take more than a decade from the first regulatory inquiries until resolution reaching some sort of détente with regulators around the world.

In hindsight, it shouldn’t have been a surprise that a company could rise to be one of the most well-capitalized and profitable companies in the world and also be subject to regulation. Microsoft’s views that we were just selling software at very low prices that customers and partners put to good use seemed rather quaint and naïve. The government was struggling to wrap itself around how such a huge success could come to exist without any involvement of regulators. The rise of the internet, originally funded by government research, only served as a reminder that something huge was shaping our economy and was essentially free of any government oversight.

This mismatch of perspectives—Microsoft as a paranoid upstart just trying to keep up with the popularity of its products and a government blindsided by unregulated corporate growth and power—created a difficult situation, which required the arduous legal system to resolve. Analysts, pundits, former regulators, and competitors can propose “remedies” (as if the success of Microsoft was an affliction) faster than the system can understand the problem (few in government had any expertise in software).

Competitors complained about one set of problems. Consumers complained about another. Partners had their own issues. Economists and academics had views too. Two things were notable about this early time in Microsoft’s massive success and “power.”

First, parties were seeking a remedy for a problem that was not yet defined. Was it simply the scale of Microsoft? Was it that Windows had come to dominate the operating system market for PCs? Was Windows a monopoly? Were PCs to be treated like common utilities? Was Microsoft’s business model of low-price, high-volume problematic? Was it unacceptable for one company to sell operating systems and also to sell applications? Or was this about some other type of product integration such as browsers and media players? These questions did not have obvious or consistent answers back then, even among third parties.

Second, assuming agreement was reached on the problems being solved, what would the right regulatory framework be? How do you solve the problems identified? The experts in regulation (and antitrust) were themselves products of the incredibly long-running cases of IBM, AT&T, and others. In the technology industry we looked at the IBM case and saw litigation solving the problem long after it mattered—the whole industry moved on to mini-computers, workstations, and then PCs from mainframes and it seemed this case was still going on, a condition that created a view that regulating the fast-moving technology industry did not make sense the way it might for the industrial economy. The AT&T case seemed remote as it was established as a monopoly and primarily involved physical cables, and much of the unleashing of competition that took place came about not because of the new regulatory framework as much as what AT&T fought for (for example, they sold off the cellphone operation for a small amount to focus their win on long distance lines). But the breakup of AT&T was on everyone’s mind and that led to calls to breakup Microsoft—it seemed clear, if there’s a monopoly it needs to be broken up into pieces. The debate over whether regulation simply stifled one of the most inventive and successful companies in US history continued.

This set up for a confrontation as the process wound through, with each side articulating extremes, and neither side particularly good at stating problems or matching problems and remedies. Microsoft, especially from its paranoid mindset as an upstart, insisted that it had done nothing wrong but make products people bought and so any interference was tantamount to killing off innovation just as had happened to IBM. The punditry would opine about the need for choice and alternatives in products and suggest that Microsoft was itself already stifling innovation.

All of this activity changed the company’s narrative. A few years earlier, BillG was a boy wonder, the under 30 founder who had grown a new industry for the world through the magic of software. By the mid-1990s, Bill and the company were ruthless competitors who rolled over every other entity, dictated terms for the industry, and above all could enter any industry and dominate. It was this fear of what Microsoft might choose to do next that drove the most extreme views of regulatory remedies—the government needed to do something to prevent Microsoft from becoming a real-world RAMJAC, from Kurt Vonnegut novels.

Regulatory norms over a new industry were unavoidable. Governments are empowered to provide oversight and there was simply no way the newest and seemingly largest and most important industry would escape regulation. It did not matter how we thought of the fragility of our industry or even how much evidence the IBM case offered as to the futility of regulating technology.

We wished someone could have made a good argument as to why technology was different than say banks, telephones, farms, oil, autos, theaters, or shoes. But technology is not different.

On the other hand, it would have been equally fair to have asked for those calling for remedies to do a better job articulating the problem being solved. And therein lies the problem. Jumping to remedies that so clearly did not address the problem only made one question motives and create the appearance that parties were further apart. Championing remedies that seem to be designed to simply kneecap a single company don’t serve an industry, or an economy.

Given the inevitable, but also the wide gap between parties, the process took a long time. It wasn’t debilitating as some might have suggested. There was learning, discovery, and socialization. The process was more like having a chronic condition with relapsing-remitting pathology. Long periods of time went by without symptoms, then suddenly and unexpectedly there was a flare up, like the opening of new complaint by a regulator, a country getting involved for the first time, a legal filing, or even a dramatic courtroom moment.

The conclusion felt anti-climactic.

With these kinds of cases, the results are rarely as dramatic as early predictions and tend to be far more specific and, well, rational solutions to identified problems. Regulation does work when it is eventually created through the system. It might not be ideal for a newly formed company or for one hoping for more extreme remedies, but it ends up designed to solve the problems that regulators can solve. The problems Microsoft ultimately showed to have pertained to how business was conducted, in a sense these were understood to be problems of monopoly maintenance. Being declared a monopoly was certainly not fun, but in many ways, it was reality—Windows had, in fact, won. It was time for Microsoft to admit that. Time would show that Microsoft’s argument that technology winning in one era will have a hard time winning in the next was decidedly true. Continuing to debate the end-state wasn’t only futile, but not done. Once a company loses in these cases, it becomes necessary to make way for the winners to own the new narrative. This might be the most difficult part to live through in the near term, and over the long term these same patterns will again play out because of that.

Managing a Verdict

We received little guidance about how to talk about legal matters. I was never under orders to avoid speaking about the trial, though that seemed like common sense. Once the verdict came down, teammates were starting to ask questions, wondering what the case meant for Office. I knew enough to know that absent anything official, people made up their own reality. I was worried that this could become a local press issue, with people talking to friends and friends talking to friends, ending up in the Seattle Times.

Time Magazine cover from November 1999 "Busting Bill"
Magazine covers like this were far more difficult for most employees than the actual legal issues. So many questions came up over the holidays that year as families gathered. (Source: Time Magazine, November 15, 1999, personal collection).

I organized an impromptu all-hands in the atrium of building 17. Anyone who wanted could attend. This was the largest space we had without going off campus (also where we presented the Office10 vision). Using a single speaker audio system, I spoke into a handheld corded microphone like a lounge singer. I walked the team through the trial and what had happened, not adding anything that was not already available to the press and public, but simply tried to explain what I could in a matter-of-fact manner. What was Microsoft accused of? What was a monopoly? What does a breakup order mean? The trial team was so focused on the external press that we did not have an internal process, so I did the best I could.

I had little to offer by way of details. I took a lesson from a former test leader on the Windows team—a management lesson that permeated Microsoft, perhaps to the point of becoming apocryphal. David Maritz (DavidMa) was formerly an Israeli tank commander during his army service. His unit of tanks out in the desert would sit there in a defensive posture in the dark of night. If the radio was silent for too long, each of the tanks started to worry something was wrong with the other. Panic might sweep across the unit. David said the way they avoided this was for him to check in with the other tanks and periodically let them know that everything was okay—even though he didn’t know himself. He taught us with that anecdote that even when leaders have no information, communicating something was better than nothing.

In between describing the intricacies of the legal process that would play out over years, people were worried that we were being immediately broken up, as in over the course of the coming weeks a spouse, partner, or roommate might work at “the other Microsoft.” I reiterated that there were still many things that could happen before this order could become a reality, and that much was still unclear.

At least there was humor in the situation. No one, in the atrium, was clear on the legal goal of splitting up Microsoft between Windows and Office. As engineers and employees on the ground, it seemed kind of nuts. Presumably, the issue was that Windows and Office were working too closely, even illegally, together and needed to stop.

In reality, Office and Windows could barely get anything done together. In fact, that was literally the topic of every meeting across the executive team. Different schedules, different customers, different system requirements, and more reinforced how far-fetched this idea was. More than crazy, by some measures this could have the potential to be a huge relief. Office might finally be treated as a vendor, like Lotus, which we always believed received better placement at Windows developer conferences!

For a decade there were rumors that the Office team accessed secret Windows source code that no one outside of Microsoft could see and that, somehow, that was an advantage. There were rumors of APIs in Windows that were secretly used by Microsoft to make Office better than competitors. There was no proof of any of this, though it made for a conspiracy theory. Back in the earliest days of a tiny Microsoft, with just tens of developers on big projects, we didn’t even have the technology to secure code from each other even if we wanted to. Ironically, many on the Office team remember diving in and trying to make Windows products work, not the other way around; whether it was Windows graphics for charts in Excel or printing in OS/2, it seemed that the advantage flowed to Windows. In the atrium, people were asking about this topic and it brought a sense of levity to an otherwise unique situation because most were not around for the early days of Windows 2 and 3, or even Windows 95.

After a brutal series of motions, briefs, and other legal warfare, a year later on June 28, 2001, a federal appeals court reversed the breakup order, reprimanding and removing Judge Jackson and appointing a new judge. As often happens in these complex cases, the judge, Colleen Kollar-Kotelly, pushed to have the parties resolve their differences outside the court. By September 2001, the plaintiffs withdrew their effort to seek the breakup of Microsoft. By November, the case worked out a settlement, which Judge Kollar-Kotelly ruled served the public interest. There were no issues in the settlement regarding Office directly, though later when I moved to Windows in early 2006 some of my immediate responsibilities included complying with the terms of the settlement, which was scheduled to end in November 2007. We voluntarily extended that by two years, which meant the first release of Windows that I worked on included making sure it followed the compliance order.

While much speculation has gone into how the legal issues impacted Microsoft execution and product strategy, my view, even on the front lines back then, was that by far the biggest issue was not in the workplace specifically, but outside of it. Even though they had nothing to do with them, everyone on the team endured the negative comments about the company and its business practices. That’s where the litigation and scrutiny truly caused difficulty. Consider those holiday dinners and family gatherings where an engineer on the team was called to the carpet to explain or defend Microsoft. Similarly, when recruiting college students, I frequently found myself on the phone with parents of candidates walking them through the case and the culture of Microsoft while also defending us. 

Those side effects of litigation were far more difficult than the specific structural and regulatory remedies.



Apple Computer, Inc. v. Microsoft Corp., 35 F.3d 1435 (9th Cir.1994).


Lotus Development Corporation v Borland International, Inc 516 U.S. 233 (1996)