Breaking Windows

With Open Source software projects, understanding why certain features are developed while others aren't, or even why entire projects split apart into contending factions, is often as simple as reading mailing list archives and web sites where the involved parties hash out (or at least air) their differences. Within a large corporation, it's a lot harder. Slashdot reader (and "former Microsoftie") Adam Barr contributes this review of Breaking Windows, which he describes as an imperfect but revealing look into the internal politics of Microsoft, and how clashing groups within the company have struggled to get their vision of Windows on the desktop -- sometimes a messy process.

Breaking Windows author David Bank pages 288 publisher Free Press rating 8 reviewer Adam Barr ISBN 0743203151 summary Tells the story of the battle that raged within Microsoft from 1997 to 2000, between those advocating sticking with the Windows strategy and those wanting a full-fledged shift to the Internet. The Scoop This is one of the best-written books about Microsoft that I have read, and as a former employee I have read most of them. Focusing on the internal battles gives a new perspective on the company. It hopefully shatters, once and for all, the myth that Microsoft is a hive community marching in line behind Bill Gates. Executives and regular employees are shown battling over issues large and small, with a consistent public story emerging only at the end, if at all. Bank also shows how Microsoft's legal strategy in the Justice Department case was affected by the political and technical battles that were simultaneously going on within the company. What's To Like The book does a great job of telling its story efficiently and clearly. Bank quotes from internal emails, but doesn't overuse them, preserving the value of these rare glimpses into the Microsoft decision-making process. He gives just the right amount of history, and avoids ill-fitting analogies to describe the various pieces of software (in most cases he simply gives a minimal explanation, which might confuse a computing novice but is perfect for a typical Slashdot reader). He also describes the right reasons for Microsoft's success: not marketing as many people say, but its strategy of defining a small number of software platforms and evangelizing them to other developers.

The battle being fought here is between the "Windows hawks," led by Microsoft Vice President Jim Allchin, and the "Internet doves," led by another Vice President, Brad Silverberg. Allchin was in charge of Windows NT; Silverberg shipped Windows 95 and early versions of Internet Explorer. The book has some great insight into how this battle proceeded and why the participants acted as they did.

For example, the book discusses Jim Allchin's famous email in early 1997, in which he discussed competing with Netscape and wrote, "I do not feel we are going to win on our current path -- I am convinced we have to use Windows, this is the one thing they don't have -- We need something with more Windows integration." This email was brought up in the Justice trial to show that Microsoft used browser integration to unfairly attack Netscape, but the book shows that Allchin at the time was trying to counteract feelings within Microsoft that the browser was all that mattered, and was therefore concerned not so much that non-integration would hurt the browser as he was concerned that non-integration would hurt Windows.

Or consider the following sentence from the book: "In the same way that Gates began to view Microsoft's Internet team as the internal representation of Netscape, he came to see Microsoft's Java team as the internal agents of Sun Microsystems." This is an extraordinarily perceptive statement, and the fact that a reader can appreciate its meaning 74 pages into the book is a tribute to the explanatory powers of Bank's writing.

What's To Consider If the terms "Internet doves" and "Windows hawks" didn't tip you off, Bank is trying to show that the "fumble" of the subtitle occurred in 1997, when Bill Gates decided against supporting a Microsoft project known as Megaserver. This would have been a platform for Internet development: a set of back-end services, tied in to the browser.

Bank also discusses another, more well-known "fumble," the mismanagement of the Justice Department lawsuit. His writing here is still excellent, but this topic has been covered elsewhere so the information is not as surprising.

In the Justice lawsuit, he does a good job of showing how Gates was the main force behind two of Microsoft's poorest showings in the case: Gates'evasive videotaped deposition, and the response to the judge's order to allow computer manufacturer to ship Windows 95 without Internet Explorer (which involved allowing them to either ship a two-year-old version of Windows 95, or one that did not work at all).

In fact Bank spends much more time talking about the legal foibles than talking about his first argument, that Gates blew his role as technical leader of Microsoft by not endorsing Megaserver in 1997. But this really needs to be the core of his argument: saying that Gates' main mistake was made in the legal arena, in which he was a novice, is not nearly as compelling as claiming that Gates, the ultimate geek, botched the kind of technical decision that should have been his strength.

Megaserver was a Brad Silverberg project, and Jim Allchin was the main opposition. In Bank's mythology, Silverberg is the hero, pushing for the Internet. Allchin is the villain, sticking with Windows. But what really went on here?

Consider a story Bank relates from a Microsoft developer named Ben Slivka, one of the most strident of the Internet doves:

Slivka recounted the experience of one Windows developer who presented Allchin with his ideas for a simply, reliable operating system suitable for home users. Instead of saying "Great idea, go do it," Allchin had insisted that the new operating system be based on Windows NT. The developer objected that the huge NT operating system wasn't suitable for the drop-dead simple appliance he had envisioned. Allchin challenged him to list the parts of Windows NT he would strip out.

To me this looks like Allchin is doing his job. What would happen if he authorized everyone who so desired to go off and write their own operating system? I/ll tell you what would happen: Windows CE. Enough said.

Allchin also had little patience for Microsoft employees who were advocating a move towards Java and free software:

I don't want to be remembered as the guy who destroyed one of the most amazing business in history. We could have done it [meaning we could have destroyed the business] with engineers who didn't understand and didn't have any responsibility for the financial aspects of the company at all. Who live in this paradise where the stock goes up, revenues keep going up, earnings keep doing up. And all they have to do is crank software. Somehow it gets into packages and makes money. Well, it doesn't work that way.

Sounds reasonable to me. The notions of first-mover advantage and trading profits for users have been discredited in the dot-com meltdown. But the quote doesn't fit into Bank's view of Allchin as the bad guy, so he simply throws it out there, with no discussion.

History is often written by the winners, but in some ways the middle of this book is history written by the losers. The path not taken is discussed, but since it exists only as a perfect creation in the minds of its inventors (who obviously had Bank's ear when he was doing his research), it is depicted as flawless. Statements claiming that the new goal "was not to get thousands of developers to adopt your arcane PC programming interfaces but rather to get tens of millions of users to use your services every day=94 are accepted as holy truth.

Bank is convinced that Megaserver would have somehow "expanded the commons" of software development, that any Internet platform would have been an open platform. But consider what the Megaserver would have been as proposed back in 1997: A set of Microsoft servers with Microsoft data, talking to a browser that was customized to talk to those servers.

In short, it would have been a clone of AOL. Furthermore, this would have been architected by the team that brought you Windows 95. Would this have been a good thing? Does integrating your browser with your Web servers produce a more open environment than integrating it with your operating system?

Thus, it is hard to fault Gates for not supporting Megaserver in 1997. In fact, Microsoft is now pushing heavily towards .Net, which is the 2001 version of Megaserver. Why support it now? As Bank himself writes, about Microsoft executive Paul Maritz, "He had long known the problem was bigger than Win32, Maritz said. But now he could articulate the message. The difference, he later said, was XML." It was not so much that Microsoft did not recognize the need to move beyond the Win32 API; it was that in 1997 it didn't have the technology to do so.

Bank makes the claim that Gates was forced out as CEO because of his "fumbles." This is arguably the big revelation in the book, but it is hard to prove this conclusively: The trial missteps certainly did happen, Microsoft was drifting from a technology perspective, and Gates stepped down. Did he fall or was he pushed? The timing of events supports either conclusion. In any event, I found the behind-the-scenes descriptions much more interesting than this particular allegation.

Furthermore, Bank points out that Gates allowed an employee to set up a hands-off incubator within Microsoft that eventually led to the company-wide adoption of XML and .NET, and was the only top executive who really understood the .NET protocols. Thus it is hard to fault him for not supporting an Internet platform in 1997, when he planted the seeds for an Internet platform in 2001.

If the middle of the book is imperfect but still fascinating, the last chapter gets really strange. After playing Brad Silverberg up as the hero, Bank suddenly cuts him down. Earlier in the book, the decision to adopt Active Desktop in Windows 98 is mentioned, but with mysterious silence on who made the final call; it merely states that after seeing Netscape demonstrate a similar product called Constellation, "the browser team was given the additional job of creating a shell for all of Windows." That shell was Active Desktop, and this particular decision got Microsoft in antitrust trouble both because it increased the amount of browser integration that Microsoft had to defend in court, and because Microsoft started leaning on computer manufacturers in an effort to freeze out Netscape's product. Furthermore, the battle was basically for naught since Channels, the big Active Desktop feature, went nowhere. Gates himself said later, "That's a case where the browser guys, they had the Internet religion, but they pushed it too far in terms of what was a practical user experience."

So who decided to go with Active Desktop? You figure it had to be Silverberg, but Bank doesn't say that. In the final chapter, however, he slips a bit, pointing out that Silverberg's team was responsible for the tying of the browser, the semi-exclusive contracts with content and access providers, and the war against Java -- the main issues that the Justice department sued over. Furthermore, if the Megaserver strategy had been pursued, Microsoft might have been in even more legal trouble.

Gates, meanwhile, gets rehabilitated in the last chapter. His tactics in 1998 and 1999 are now described as a strategic stall, waiting for the right technology to appear for Microsoft's Internet platform: "The power to control the pace of innovation is a competitive advantage at least as crucial as the ability to innovate itself." Gates is portrayed as a leader once again, planning strategy ten years out, and the book ends with a prediction (for no reason other than the author's gut feeling) that Gates will do the right thing and usher in a new age of innovation, whatever that consists of.

I'm not sure what to make of this flip-flop. I assume this book was originally proposed to a publisher in 1999, written in 2000, and polished up in early 2001. In 1999 a book about the demise of Microsoft seemed a plausible undertaking, but two years later it turned out that the story wasn't over, and Microsoft appeared to be bouncing back. So Bank had time to equivocate, modifying his original thesis and explaining how perhaps Microsoft had a future after all.

Describing this latest turn of events, however, Bank doesn't have reams of email released during a trial, or sympathetic former Microsofties to interpret it for him. As a result, he can fire off sentences like, "The infrastructure for the digital age will be based on competition on the merits and a common code of open interfaces," with apparent complete sincerity. He believes that Microsoft asking AOL to open its Instant Messaging protocol is a harbinger of this golden future, and that Microsoft's Shared Source program shows it is moving towards open source. In short, he is buying the current Microsoft PR story, hook line and sinker.

Well, let this former Microsoftie (and former Windows hawk who worked in Allchin's group) explain a few things. Statements like "Interoperability, not lock-in, has become the winning strategy" are patently false. Right now there are two Internets: The AOL one, with its own client, servers, content, email, messaging, authentication, billing, security, and all the rest; and the plain old Internet. Microsoft wants to create a third Internet, the .NET Internet, with all the stuff that the AOL Internet has. Then it will pursue a lock-in like the world has never seen before.

Summary and Table of Contents But hey, enough quibbling. Bank may be wrong about the future of Microsoft, but he does a fantastic job covering the past. I spent some time discussing what I disagreed with, but there is so much more that I agree with. I knew about a lot of the events that are described in the book, but I still learned an incredible amount. If you want to know what things are like inside Microsoft, buy this book. Table of Contents

• Prologue: The E-mail Trail

1. Track the Inevitable
2. Hawks and Doves.
3. The Path Not Taken
4. Citizen Gates
5. Vicious Cycle
6. Monopolist's Dilemma
7. Loosely Coupled

• Key Dates
• Notes
• Acknowledgements
• Index

You can purchase this book at Fatbrain.

Comments

I wonder how the author would explain this

[WeaselGod]

http://www.ntk.net/ballmer/dancemonkeyboy.mpg

Well..

[reflective recursion]

I don't know about being able to understand open source. I look at sourceforge.net and sometimes wonder why the hell certain programs are even there. Then I look for their source code and wonder where their source code is. Then I look at freshmeat.net and wonder why there is the 100000th mp3 web-frontend written in perl. Maybe it's just me.