Hard Drive: Bill Gates and the Making of the Microsoft Empire (33 page)

Nikora didn’t agree either, but he dutifully went along.

A former Microsoft manager described the atmosphere under Gates in 1984 as management by crisis. “Everything was a crisis.”

In August, Microsoft and Gates faced up to the reality that at least part of that haphazard style, which had helped power a remarkable if sometimes inefficient software incubator for nearly ten years, was creating too many management problems. The crash program to develop Windows had stretched into nearly a year, and still there was no marketable product in sight. Gates, under the firm prodding of Jon Shirley, agreed to take himself out of the day-to-day decision making. He was trying to oversee five product lines, and key planning decisions were often delayed or were never made at all.

“It was a strange structure,” Gates admitted in an interview with
Business Week
after the company’s reorganization. “We could never focus enough on the different parts of the business.”

After Microsoft was reorganized into separate systems software and business applications divisions, each with a corporate vice-president, Gates’ responsibilities became somewhat less defined. He no longer had anyone reporting directly to him. He was supposed to think about the future, something he had been brilliant at all along. He also had more time to become a more visible spokesperson for Microsoft.

But there was another reason for the split into separate divisions not disclosed to the press. There were complaints from software developers that Microsoft had a competitive advantage because its application group got inside information from the operating systems group, including the group developing Windows. Programmers working on applications at Microsoft to run under Windows, for example, knew a lot more about Windows than outside competitors who were developing applications to work with Windows. With this inside knowledge, Microsoft’s products would naturally be better. Also, each small change in the software forced these companies to meet with Microsoft to make sure the change would also work with MS-DOS. This gave Microsoft, which has its own application programs, an inside look at a competitor’s best products. It was an advantage enjoyed by no other company in the software industry.

Gates insisted that Microsoft kept the playing field level by erecting an imaginary barrier between the company’s Operating Systems Group and its application division. This barrier became known in the industry as the “Chinese Wall.”

MacGregor said that at the time of the reorganization, information flowed freely between applications developers and Windows developers. Dan Lipkie, the former Xerox programmer, worked on Microsoft Word before transferring to Windows. And Neil Konzen, a top Microsoft software engineer, switched from the applications group to Windows in August 1984.

“All of Konzen’s friends were in the applications group, and they’d talk at lunch all the time, so it really was true,” MacGregor said. “The application developers at Microsoft had better knowledge of Windows than programmers outside the company.”

Two months after the August reorganization, Leo Nikora found himself on the phone to the press, explaining yet another delay. Windows was supposed to be finished in October. But after a year of development, Microsoft officials had reached the conclusion that the program took up too much memory and was too slow to run on the 8088 microchip effectively. Major parts of the program would have to be redesigned. A flashy product rollout slated for the fall Comdex was canceled, and the release was rescheduled for spring, 1985.

The situation was frustrating. Each time Gates told his Windows marketing crew the schedule was being moved back, he would also tell them that this time the release date would be met. And each time, Nikora found himself going back out to the independent software vendors to tell them that the target had been missed again. “I just felt like an idiot after two or three of those,” Nikora said.

Things weren’t easier for Gates, who in early October 1984 wrote a letter to one impatient software publisher explaining the latest glitch: “Windows is the most strategic product that Microsoft is working on,” Gates wrote. “We want it to be the environment of choice for the next generation of graphical applications. To achieve this goal, it must have the functionality and performance required by this new generation of graphical applications. This will not be easy; it will be a significant advancement of the state of the art.”

He went on to explain how he felt there had not been enough progress on improving the product’s performance, and he told the developer that Windows would be delayed until April 15. The press, he said, would be told that Windows would be available in June. This was an attempt by Gates to make it appear the already embarrassed company had actually beaten its self-imposed deadline with the April release. Gates also wrote that the company intended to pull out all the stops promoting Windows at the spring Comdex. Microsoft, he said, would make a “big splash.”

“We hope that you too can beat this date with your Windows application before June,” Gates wrote. “We will be talking to the press between now and October 15th. Should you receive any inquiries from the press before then, please refer them to Microsoft; after then, please support us in keeping expectations to the June time frame.”

By this time, however, no amount of spin control could undo the damage to Microsoft’s reputation. Word of the latest delay prompted numerous questions in computer publications of the need and utility of graphical user interfaces.
Forbes
magazine noted that VisiCorp and Quarterdeck were on their deathbeds because of expensive and futile gambles that PC users would flock to these windowing products. One trade publication predicted all the excitement over GUI was a flash in the pan and would burn off.
InfoWorld
used the term “vaporware” to describe Windows and all other software that was hyped ad nau- seum but never made it to dealers’ shelves.

In the past, Gates’ strategy was to always get out first with a product and grab a share of the market and fix the problems later. With Windows, he inexplicably took just the opposite approach. He couldn’t bring himself to turn it lose until everything was perfect.

“Most of the reason Windows didn’t get out was because Bill kept adding functionality [features] and changing the rules,” Nikora said. “All these other products were on the market. I was saying, ‘Look, Bill, you’ve got to be out there, take what you’ve got and get it out there. You can always make it better, you can always improve it later, but get out there.’ Bill’s feeling was, ‘No, we’ve got to go out there with something that’s just head and shoulders above everybody else. It’s gotta be the right product.’ ”

Gates was increasingly under fire from the industry. For the first time, his judgment and business acumen were being questioned.

The numerous delays of Windows provided little encouragement for software publishers to develop applications for the graphical PC environment. VisiOn had been on the market for more than a year and had won few converts, despite VisiCorp’s three-year, $10 million investment. VisiCorp was going broke and would soon be gone. Quarterdeck introduced its DESQ windowing product, but it too suffered disappointing sales'. Digital Research haid a graphical user interface product called GEM due for release soon. But the battle, as the industry press saw it, was going to be between Windows and TopView, which IBM planned to market in early 1985.

In the opening days of 1985, Microsoft was forced into a key strategic marketing shift. After months of testing and tinkering, Gates had to admit Windows just didn’t work fast enough or well enough on existing hardware. That ended plans to bundle Windows as part of a package with new PCs made by Compaq, Tandy, and the rest of the OEMs who had originally supported the program.

With IBM pushing TopView instead of Windows, Microsoft had only one move left if it wanted the program to become a standard. It would have to sell Windows separately on the retail market and hope consumers bought it in sufficient numbers.

In-house, meanwhile, there was more trouble on the project. Ballmer was not getting along with Nikora. He would storm into Nikora’s office almost daily, pound his fist on the table, call Nikora an “idiot,” and dismiss his ideas as “complete jokes.”

“They didn’t like what I was doing in marketing,” Nikora said. “They didn’t think I was a very good marketer. And you know what, they were right.” Nikora had, after all, arrived at the company as an accomplished programmer and manager. “I’d never marketed a day in my life—and they never did anything to teach me.”

Nikora soon quit. “I was burned out,” he said. “I also left because Microsoft wasn’t quite ready for middle management. They thought they were ready, but they weren’t.”

MacGregor was also having his problems with Ballmer. The former equipment manager of the Harvard football team had become something of a wild man. He was a Gates’ clone, but without Gates’ charm or technical know-how.

Said one member of the Windows team: “None of the developers admired Ballmer, because Bill had established this ethos that development managers coded. Then he appoints Steve Ballmer, who had never written a line of code in his life.” Gates and Ballmer had very different styles. Ballmer was more of a cheerleader. Instead of working with the programming team through MacGregor, Ballmer was always intervening personally, exhorting the engineers to work faster. He would get the software engineers psyched to really go all out and push the outside of the envelope. But this caused monumental problems for MacGregor. By their nature, software engineers, when excited about a project, would always underestimate how long something would take. Then Ballmer would get in meetings with the engineers, look at their development schedules, and give ’em hell. “You can do better than that!” he would shout while beating his fist on the table.

“These guys were pretty young, just out of school,” said MacGregor, “and they’d try to take some time off their schedules which were already unmakeable. This was incredibly frustrating for me, because you’d get a quote to do a particular function in four weeks, and I knew they didn’t have a prayer of doing it in less than six, and here Ballmer is pushing to do it in two or three. It’s set up for failure. He knew people were going to bust their butts, but there was a lot of stress in the group, marriages were strained. My philosophy is, you want to build a team for the long term. Its okay to sprint, you need to sprint to get products out the door. But you can’t sprint all the time.”

Ballmer believed that scheduling was something to be “negotiated” rather than an attempt to accurately project how much time something would take. “It meant that we’d quote dates [to software publishers writing Windows applications] that were totally unrealistic,” said MacGregor.

But there were fewer and fewer customers to worry about. According to the president of a software publishing firm that supported Windows, the repeated delays had caused a crisis of confidence in the industry: “What was interesting was everybody from Lotus to Ashton-Tate, everybody that was anybody, was at that Windows developers conference in early 1984 and they were all up on the stage saying ‘Yeah, we’re going to do Windows and Windows is great, blah, blah, blah. But by mid-

One of the biggest disappointments for Microsoft was that Lotus decided not to follow through on a pledge to support Windows. Mitch Kapor knew he couldn’t afford to have Microsoft set the graphical user interface standard for the PC. Allowing Windows to run 1-2-3, the best-selling spreadsheet on the market, would have given Microsoft too much leverage. “If Mitch had put the seal of approval on Windows, it would have been a big deal, and he didn’t want to do that,” said one of Microsoft’s managers. Both Gates and Ballmer did a lot of arm twisting with Kapor, but he refused to come around.

By the spring of 1985 MacGregor, too, had decided to leave. “There was some frustration, a sense of wanting to do some

other things,” he said. “The team was pretty stressed out. It had been a long haul. I didn’t agree with some of the strategy. ... I wasn’t having fun. I think its very important, if you have the luxury of choosing jobs, to choose one that’s fun. But I have positive feelings toward the company. I’m glad to have worked there, glad to have known the people. But sometimes you just want to do other things.”

Differences with Ballmer helped MacGregor make up his mind that it was time to move on.

“The fact that Steve and I had our differences was probably one of the reasons I didn’t stay. I think his style of directly interacting with the programmers and beating them up is just not a good management technique. He probably felt I was dragging the project out, and that could have been his rationale.” Gates did his best to persuade MacGregor to stay on. In the weeks before MacGregor left, they often went out to dinner and talked about the company. Gates would drive over to his apartment in his brown Mercedes, the back seat filled with empty milk cartons. Gates drank a lot of milk and he woulcl toss the empty carton over his shoulder and continue driving. The car smelled like sour milk.

Gates took it personally when MacGregor finally left. He couldn’t understand why the relationship couldn’t be repaired, according to MacGregor. The day after MacGregor left, Ballmer moved into his vacated office.

Some time later that spring, with another deadline approaching, Gates called Ballmer himself into his office on a rain- sodden weekday morning. The night before, Gates had uncovered a bug in the latest version of the Windows program. He had had enough. Fixing Ballmer with a cold blue stare, Gates began shouting at him. If Windows was not on the shelves by the end of the year, Gates threatened, Ballmer would be looking for another job.

The ultimatum galvanized Ballmer. He gathered his engineering staff together to pass along the message. “Kids,” Ballmer said, “we must ship this product before the snow falls.”

Dedication quickly turned into fanaticism. Gabe Newell, one of the Windows testers who went on to enjoy a long and successful career at Microsoft, showed up at the office with a sleeping bag. For a solid month, he camped in his office, working around the clock and catching a few catnaps when he could no longer stay awake. He became known as “Madman” Newell from that point on.

The unrelenting push, however, required an occasional diversion. While waiting for code to be compiled in the early morning hours, several programmers took to creating bombs and rockets out of sugar and saltpeter. The makeshift explosives were carted outside and set off in the night air of the quiet suburb. The games ended when the police arrived with bomb- sniffing dogs, though no one was caught.

In an attempt to amp up the power of the homemade explosives, one young programmer came up with a plan to melt the sugar in the cafeteria’s microwave oven to give the mixture more kick. The fuel ignited while it was brewing, blowing up the oven and sending a plume of smoke throughout the building.

The police were alerted on one other occasion. This time, the more musically inclined, after growing bored with full-vol- ume jam sessions in the hallways, lifted their amplifiers to the roof and serenaded the surrounding community with electric guitars. The cops weren’t fast enough; by the time they arrived, the programmers were snickering over their computer terminals.

One ex-Microsoft employee summed up the programmers feelings under such terrible pressure: “You felt you were at the center of the universe. That was the motivation, that and just trying to get clean code out there. It was an invigorating feeling to be working for Microsoft. And all this pounding by Steve Ballmer, and yanking by Bill, was the price you paid to be there.”

When Comdex rolled around that May, Microsoft demonstrated an advanced version of Windows. But there was no “big splash” as Gates had promised in his letter to software publishers seven months before. The company’s Comdex profile was uncharacteristically subdued. Windows wasn’t ready. The new target had been set for June. June came and went.

In the end, Windows wasn’t ready until November, at the fall Comdex. Fortunately for Ballmer, it rarely snowed in Seattle before Christmas (when it snowed at all). Ballmer kept his job.

“We will never have another delay like Windows,” vowed Ballmer to a writer for
Business Week.

There was an anticlimactic atmosphere to the final product announcement Microsoft made the evening of Nov. 21, 1985 at the fall Comdex. The major PC magazine writers and editors were in attendance, as were representatives of the software and computer distribution chains. The event was marked by a roast for Microsoft, a chance for Gates and Ballmer to be served a generous helping of crow.

After more than two years of delays, disappointments, public embarrassment, and overblown promises made by Gates and other top company officials, Windows was “officially” finished. During that time, Gates and Ballmer had ridden and prodded the team working on Windows to the limits of their endurance. They had yelled, screamed and verbally whipped the team mercilessly as one deadline after another slipped by. Now it was time for the industry to indulge in a little payback.

Stewart Alsop from
InfoWorld
fired the opening salvo when he presented Gates with the Golden Vaporware Award, the derisive term coined by the magazine to describe software products that existed mainly in the words and vapor of grandiose and overreaching marketing plans. Another guest speaker, John Dvorak of PC
Magazine,
wisecracked that when Windows was announced in late 1983, the balding Ballmer still had some of his hair. What he didn’t realize was that Ballmer was lucky to have a job.

At the end of the roasting, Ballmer led the audience in a song to which Gates joined in as a giant shopping cart filled with 500 packaged Windows programs was rolled out on stage. On this night, as laughter and singing filled the room, all seemed forgiven. But a long trail of human carnage left behind at Microsoft marked the frenzied forced march to develop Windows.

The shipment of the first Windows products was a watershed in the evolution of the PC, as well as the fruition of a vision Gates had been nursing for years. But the first version of Windows was not the success Gates had hoped for and promised. In fact, it was a flop. It would take two major revisions before Windows was made right. Not until the release of Windows 3.0 in 1990 would it deliver as promised.

“Windows was a pig,” acknowledged one of the programmers who had labored almost two years to develop it. The program was too ambitious to run efficiently on most of the personal computers then in use. The machines lacked the memory and speed to take advantage of the best features the software had to offer. In addition, because of all the delays at Microsoft, there were few application programs available to run under Windows. The graphical environment, for all of Gates’ efforts to boost it, looked dead. Fortunately for Microsoft, IBM’s TopView had also been unsuccessful, despite the marketing clout of Big Blue. Windows was seemingly a product whose time had not yet come.

In some ways Bill Gates, the CEO
Time
magazine described as looking like an undernourished graduate student, had became something of a Silicon bully by 1985, kicking sand in the face of competitors who got in his way. Apple Computer, for one, found itself picking sand from its eyes for years after tangling with Gates over the development of Windows.

On November 22, one day after the Comdex roast, Microsoft and Apple signed a confidential agreement that three years later would become the crux of the most significant lawsuit in the history of the personal computer industry. Gates had dragged Apple to the negotiating table kicking and screaming. Apple claimed that Windows was a rip-off of Macintosh technology. As the
Wall Street Journal
would write later, Gates used “high pressure tactics ... to extract a virtual blank check to borrow many Macintosh ideas for Microsoft s own products. These ideas included mouse-activated pull-down command menus and overlapping windows of on-screen text.”

Throughout 1985, Apple officials had become increasingly concerned that Microsoft was borrowing ideas from the Macintosh to use in Windows and other products in violation of a 1982 agreement signed by Gates and Jobs. Under terms of that agreement, Microsoft was to develop application programs for the Macintosh. By providing Microsoft with prototypes of the Mac, as well as software tools to write those applications, Apple believed Microsoft was developing graphical user interface programs only for the Mac. Instead, Microsoft turned around and developed Windows for competing IBM clone machines.

When Apple threatened to sue Microsoft in 1985 over Windows for copyright violations, Gates said he would stop development of Excel and Word for the Mac, which at the time were desperately needed software applications which Apple hoped would spur sagging sales of the Macintosh. Apple had no choice but to back down on its threat to take legal action. Instead, it signed a licensing agreement giving Microsoft royalty-free rights to use the graphical display technology developed for the Macintosh.

“The parties have a long history of cooperation and trust and wish to maintain that mutual beneficial relationship,” the agreement stated. “However, a dispute has arisen concerning the ownership of and possible copyright infringement as to certain visual displays generated by several Microsoft software products.” Those products included Excel, Windows, Multiplan, and Word.

“For purposes of resolving this dispute and in consideration of the license grant from Apple . . . ,” the agreement went on, “Microsoft acknowledges that the visual displays in [Excel, Windows, Word, and Multiplan] are derivative works of the visual displays generated by Apple s Lisa and Macintosh graphic user interface programs.”

The agreement was signed by Gates and John Sculley, Apple’s new chairman. In a reorganization of Apple, Sculley had recently forced out his friend Steve Jobs in a much publicized showdown. Sculley and others on the Apple board felt Jobs was not an effective enough corporate manager to take the company through the tough times Sculley foresaw.

Apple received little in return for granting Microsoft an exclusive license to use Macintosh graphics technology. But Microsoft did promise to fix problems in Word, which had been released for the Macintosh in late 1984 with numerous bugs. (Gates later admitted Microsoft took too many shortcuts in trying to get the product out quickly.)

At the same time Gates was brow-beating Sculley to the negotiating table in 1985, he was also using his muscle to force Apple to stop work on a Macintosh program of its own. Apple was planning to bring out its version of MacBASIC, even though Microsoft had a similar product that was already being sold with the computer. Apple engineers believed their BASIC was better. When Gates heard of Apple’s plans, he was beside himself. He felt betrayed, since Microsoft had thrown so much of its resources into developing applications for the Macintosh. Arguably, Microsoft saved the Mac with the introduction of Excel. Gates demanded that Sculley cancel the project and sign over to Microsoft rights to the MacBASIC name. As a lever, Gates told Sculley he would not renegotiate the license for Apple to use Microsoft’s BASIC on the best-selling Apple II. At the time, sales for the Macintosh were down to nothing and the Apple II was the company’s bread and butter product.

“Essentially, since Microsoft started their company with BASIC, they felt proprietary towards it,” said Andy Hertzfeld, the Apple software engineer who had worked on the Macintosh