During the first six months of last year I worked at Microsoft's headquarters, in Redmond, Washington, on a team designing the next release of Microsoft Word. 
For me this started out as the midlife fulfillment of a fantasy. I'd been intrigued by software design since 1979, when I started using my first word-processing program, The Electric Pencil, on a prehistoric computer called the Processor Technology SOL-20. As programs became bigger, fatter, stronger, and more fully controlled by Microsoft, I found myself drawing up increasingly elaborate wish lists of the features that a really great piece of software for writers would include. So I made a proposal to try to persuade Microsoft designers to see things my way. The terms were that I'd work at the company as a consultant for a fixed period of time. Any of "my" features they decided were worthwhile could be included as part of Word or Office, but anything they decided not to use would remain my intellectual property, so that in some theoretical other life I could start a company and sell my own little word processor in competition with theirs.
The terms also covered what I could say about the experience when it was over. Naturally, I had to promise to protect Microsoft's trade secrets. I also volunteered not to publish a memoir or an "inside Microsoft" confessional—or not to do so without allowing the company to read and approve it ahead of time, which is the same thing as agreeing not to do it. In exchange Microsoft allowed me to "draw on my experience" at the company in future writing about technology. It is a hazy distinction, but I think we understood each other. I didn't want to make it less likely that another interested outsider would be taken seriously by the company, but I also couldn't accept a blanket prohibition against ever saying anything about the interaction.
My contacts at Microsoft knew that through the 1990s I'd written warnings about the company's growing monopoly power. So why did I want to work there? Because Microsoft had eliminated the competition. If you want to affect the program people use for writing, you have to deal with Word. We agreed to disagree about the antitrust trial, which was under way while I worked there; I kept my views about it to myself unless asked.
Under these terms I spent an almost entirely enjoyable, and more or less productive, six months in Redmond, working mainly on Word but also on features that might be used in some other parts of the Microsoft Office "suite" of programs (which includes Word, PowerPoint, the spreadsheet program Excel, the e-mail and scheduling program Outlook, and others). Some of the features I was most excited about may well survive to appear in the next release of Word, which is supposed to go on sale sometime next year. Others were killed off early in the process. Sorry, no details! This is life in the world of trade secrets. You'll know that my features prevailed if you find me grabbing strangers next year, saying, "Hey, look how cool this program is! Wonder where they got the idea for it?" Until then what I can try to convey, in a form that matches the "generalities only" spirit of my agreement with Microsoft, is the instructive surprise of the experience. Having read about software design and Microsoft culture for many years, I found a few things that exactly matched my expectations. But many more things did not. Just before leaving Microsoft, I gave a talk "on campus" about what I'd learned. These are the things that I told them had surprised me:
The people are nice. Okay, this sounds like a high school bromide. The reason it's worth mentioning is that it is a surprise, given both the public's and the software industry's impressions of the company. At Redmond the prevailing view is that Microsoft has prospered strictly because it builds great software, and that Americans can't help being proud that their dynamic national culture has spawned such entrepreneurs. That strikes me as being years out of sync with the real national attitude. Polls indicate that people still admire Bill Gates—who will, after all, probably rank as the greatest American business strategist of the twentieth century. But the devastating "findings of fact" in the antitrust case last November said that the company had gone out of its way to squash competitors, which in effect reduced the rate of innovation in mainstream software products to whatever Microsoft chose to do. This is the view the software industry has held for a long time—with the added twist that Microsoft employees are thought to be haughty, sharp-tongued, and prickly to deal with.
Perhaps my standards were skewed. After all, before working at Microsoft I'd been hanging out with journalists and political types in Washington, D.C. And I never had to face Microsoft's intensity as a competitor. Viewed from within, though, this was about as collegial and nonbackbiting an environment as I've ever been part of. What people considered a sharp exchange about features or strategy (they call such a disagreement "pushback") was nothing compared with the way lawyers, journalists, or politicians snap at one another.
A small but noticeable group within the Redmond work force would have to be considered geeks: grossly over- or under-weight, weirdo hair and clothes, various hygienic oddities. One guy appeared to have a boa constrictor living in his office (it was in a cage). Another office contained several thousand empty soft-drink cans. A man who befriended me when I arrived had rigged up a way to see what his cats were doing at home in Seattle while he was at work in Redmond.
But even the oddest people seemed generally to have a sense of humor about themselves, and at least as large a group seemed happy, well balanced, normally proportioned, and so on. I got into a little psychological cold war with one manager, who considered me a spy and wouldn't talk to me. I took every opportunity to glower at him in the halls. But I was in good spirits during the forty-minute drive from Seattle to Redmond each morning, because I looked forward to spending time with everyone else.
There was a lot of time to spend, because the next surprise was that the pace is slow. I figured out, not quite soon enough, that it drove my colleagues crazy to hear this observation, so eventually I stopped mentioning it. Nonetheless, it was true: the work week seemed less grueling at Microsoft than at other organizations I've been part of. There was a noticeable thinning in the parking lots by the middle of Friday afternoons. It seemed as if every week or so work shut down for some campus-wide or division-wide party.
I had assumed that Microsoft's pace would be at the workaholic extreme of society, but compared with people in other very hard-driving organizations—high-powered law firms, investment banks, presidential election campaigns, Internet startups, even newspapers and weekly magazines—the people at Microsoft seemed to average fewer hours a week and a lower percentage of all-nighters. Compared with other companies of more than 30,000 employees, of course, Microsoft keeps up quite a brisk pace.
Naturally, some people at Microsoft, including senior managers, do work all the time, and at certain points in the software production cycle lots of people have to work nonstop. This is especially so during "bug-bashing" sessions, when defects in program code have to be cleaned up to meet deadlines. Colleagues pointed out to me that if I stayed for a whole development cycle, over two or three years, I'd see the pace pick up. I was used to the much shorter cycles of the magazine world, where the "slow times" are a day or two, not six-month stretches.
Somehow I had also assumed that the higher one moves up the technology pyramid, the more "virtual" the working experience becomes. Even journalism and publishing, no one's idea of technology pioneers, have been transformed in the past half-dozen years by e-mail (as they were earlier by the telegraph, the telephone, and the fax). Over the phone and by e-mail editors and writers often have very close working relationships for months on end without actually seeing each other. I could only imagine how much more disembodied the working style would be at a company whose very products are designed to let people work together with no more than electronic connections.
I was intrigued, therefore, to find that Microsoft relies as heavily on face-to-face contact as any organization I've ever seen. The culture is meeting-centric. To make a difference one must be at the headquarters, must come to the office, must sit through meetings. Microsoft has employees scattered around the globe, but nearly everyone whose job is to plan or develop software works within a mile of Bill Gates's office. (The exceptions are people developing foreign-language versions or working at certain subsidiary companies that Microsoft has bought but not moved to Redmond.) The basic units of the working day are collaborative meetings, scheduled in blocks of one or more hours, where people argue over what features to include in a program or how to get rid of certain bugs. Every employee has an office; every office has a whiteboard, on which people can write ideas or priority lists for others to see during a meeting. Much of the company's work occurs in planned or unplanned gatherings around a whiteboard. Between the meetings are hallway conversations or talks in the cafeteria, where nearly everyone eats nearly every day.People do work on their own. The developers, who do the actual software coding, may hole up for days to work on certain features. The program managers, who are supposed to itemize the tricks the developers will make a program do, may also vanish, to finish program specifications. But it seemed as if the company was as dependent as a football team on having its members at the same place at the same time. Maybe this is because developing good software requires an unusual mixture of talents—that, at least, is the explanation I often heard at Microsoft for its emphasis on face-to-face collaboration. Or maybe it suggests some skepticism about the extent to which workplaces will ever become virtual.
The planning process is radically "bottom-up" and surprisingly nonpolitical. I am sure that "Now hear this!" directives about the evolution of new software come periodically from Bill Gates, Steve Ballmer (the president of Microsoft), and the company's other strategists. In 1995 Gates famously ordered the company to reverse course and take the Internet seriously. Office 2000, which had been completed just as I got there, reflected a from-the-top emphasis on making all programs compatible with the Internet. But strategic documents never filtered down to my level. What I did see was a quite amazing process in which programs were decided on, shaped, revised, and implemented almost entirely by people at the working level, without the need for big shots to resolve arguments.
Part of my role was to argue that it would be good for customers, and therefore ultimately good for Microsoft, to build in certain features that would make Word or Outlook easier to use. In some organizations the most effective scheme of action might be: 1) persuade the boss that this feature is a good idea, and 2) have the boss tell someone to produce the feature. At Microsoft the process seems to be: 1) persuade your colleagues that a certain feature will be popular, and that it can be created, and 2) create it. If there is something you love or hate about Microsoft programs, don't thank or blame Bill Gates; some specific member of the Microsoft team decided to "own" that feature and include it in a program. There is even a person who created the "It looks like you're writing a letter" auto-annoyance feature in Word. I had to sign a separate confidentiality clause promising not to name him.
I didn't always, or even usually, agree with the list of features that survived this process. But I was struck by the dispassion and apparent lack of scheming that went into deliberations about what to include. This may have been in part because supervisors stayed away from the details of product planning—there is no need to play the courtier if final decisions rest with your peers. But I think it also reflected a shared understanding that everyone was on the same team: the Microsoft Stock Option Team. If someone on the other side of the table had a better idea than yours for making the program attractive, you might as well give in gracefully, in view of the potentially stupendous rewards.
Which brings us to money. For the first week or two I was hit surprisingly hard by the realization that I was the poorest nonjanitorial worker in my building. (I was there as a temporary consultant; only long-term employees are eligible for stock options.) At some point I stopped thinking about it and tried to observe the way everyone else thought and talked about money. They talked about its consequences quite a lot. Standard lunch-table chat would concern the new boat or sports car someone had bought. And when Microsoft stock was moving up or down, there was a fair amount of hallway chatter about it. But on the fundamental questions—How much do you have? How much do you need?—people were taciturn. Everyone recognized the concept of having "enough" money to quit and do something else, as a large number of Microsoft veterans have done. But almost no one would specify how much would be enough.
Certain aspects of Microsoft culture reminded me of Japan. The company is self-contained and thinks of itself as separate from the rest of the industry. It cares about market share above all else. Its talent pool is remarkable at the low end for how competent the weakest person is. But more and more often I found myself thinking that Microsoft is like the military. I don't mean by this what the company's detractors presumably would: that it is a rampaging force bent on world domination. I mean that in many small aspects of daily life and in the large questions of long-term strategy the organization resembles a big, successful military establishment.
Once this idea got into my head, I couldn't stop seeing evidence for it. When on the job, people in the Army wear clothes saying U.S. Army. At least a third of the people in the Microsoft cafeteria wore shirts or jackets saying Microsoft. The military has a PX; Microsoft has the Company Store, where employees can buy Microsoft products at 70 to 80 percent off list price. For the mess hall there is the cafeteria. For the rec center there is a huge fitness club—officially run by a different company, but nearly all the members are from Microsoft. A military base is full of vaguely insulting posters urging service people not to pass bad checks, not to get VD, not to have fatal accidents. Microsoft buildings are full of posters with messages like WRITE GREAT CODE! and YOU CAN WORK AS ONE.
Most Microsoft employees thought that my comparison with the military fed the industry view of their company as a heartless machine, but I meant it as a compliment. The image I had in mind was the great, dominant U.S. military of the Second World War and the decade after, combining varied talents and getting the job done. Especially in combat, the military tolerates all sorts of idiosyncrasies, as long as everyone pulls together when it counts. The superficially oddball developers do pull together: indeed, there seems to be very little grousing about the rightness of the largest corporate goals. The military famously prizes effectiveness rather than efficiency: it matters more that you beat the enemy than that you get high productivity per manhour out of your troops. So, too, in software. Half the time developers are shooting Nerf cannons at one another, waiting for instructions from the program managers, but then they all do what needs to be done.There is one crucial difference between the tech economy and the military. This point is very touchy, and I am careful not to associate it with Microsoft in particular—not just because I signed an agreement. The software world, like the U.S. military, contains great racial diversity. People of every skin color work in Redmond—and in Sunnyvale and in San Jose. The difference is that the dark-skinned people who work in high tech are nearly all non-American. They are from India, perhaps from Malaysia, sometimes from Kenya or Ethiopia.
American blacks have about the same representation in software development that American Jews have in pro basketball. Eventually this will become a problem. To his credit, Gates took a step toward heading it off with his Gates Millennium Scholars Program, which will support advanced science and technology education for 20,000 students from under-represented minorities.
One other discovery helped me understand why the company has remained so profitable and dominant. Microsoft understands exactly who its most important customers are. Unfortunately, that group does not include people like me.
Financial analysts have long recognized that Microsoft's profit really comes from two sources. One is operating systems (Windows, in all its varieties), and the other is the Office suite of programs. Everything else—Flight Simulator, Slate, MSNBC, mice and keyboards—is financially meaningless. What these two big categories have in common is that individuals are not the significant customers. Operating systems are sold mainly to computer companies such as Dell and Compaq, which pass them pre-loaded to individual consumers. And the main paying customers for Office are big corporations (or what the high-tech world calls LORGs, for "large-size organizations"), which may buy thousands of "seats" for their employees at hundreds of dollars apiece. Product planning, therefore, is focused with admirable clarity on those whose decisions really matter to Microsoft—the information-technology manager at Chevron or the U.S. Department of Agriculture, for example—rather than some writer with an idea about how to make his colleagues happier with a program.
This survival focus did not please me in all its consequences, but it made me think that this is an organization that knows how to succeed. And Microsoft's very success in satisfying the LORG market makes clear another step in the evolution of the tech economy. The software business no longer seems to have room for small companies that address the tastes of a specialized audience—for instance, professional writers. But the Internet, with its near-zero distribution costs and its ability to group widely dispersed people with common tastes, may be a hospitable environment for boutique enterprises. Maybe that's where I'll launch my little word-processor company—unless Microsoft, now alert to my plans, decides that a great program for magazine writers really would be a "killer app."