Startup

by S. Jerrold Kaplan

  Bob Evans’s background was atypical for a Silicon Valley entrepreneur. He was sixty-six years old, and his primary relevant experience was managing large projects at IBM, most notably with the company’s mainframe development organization. After thirty-two years with IBM, he joined the investment-banking firm of Hambrecht and Quist. But his distinguished history as an IBM vice president was not the best preparation for the job of managing EO.

  The EO executive staff rapidly discovered that Evans thought of engineers more as interchangeable parts in a machine than as specialized cells in an organism. He refused to approve monetary incentives to retain key people, preferring instead to replace them with new hires. As a result, the company lost talented engineers daily. Adding together the two big layoffs—one as part of the merger, the other in February—and the individuals who left on their own initiative, more than 200 years of accumulated experience with the Penpoint code base had walked out the door, leaving only about 150 years at the company to pick up the slack. At GO, the average software engineer had been responsible for 10,000 to 20,000 lines of code, but at EO the load rose to about 80,000. Since the training programs had been dismantled, all newly hired engineers had to learn while on the job, substantially reducing the productivity of the remaining staff.

  Evans and his executive staff spent the month of March revising the company’s business plan once again before presenting it to the EO board for approval at their meeting on April 7. But just as the staff was finishing up its work, a group of managers from AT&T showed up for a preliminary review. Among them was the strategic advisor to Kavner, whom I had chased from Basking Ridge to a Tokyo sushi bar. He recommended a new strategy for the company—that among other things EO should get out of the hardware business altogether after the current products were completed and concentrate on licensing its software to other manufacturers instead, which was GO’s original aim before the merger.

  With the company bleeding engineers, the Hobbit project canceled, all of GO’s ISVs and hardware licensees gone, and Evans considering reinstating the original GO business plan, Mike Homer realized that the next EO product would never ship. He finally resigned in frustration.

  Less than three months later, AT&T decided to withdraw funding for EO completely. On July 29, 1994, AT&T put out a press release, fired nearly all of the remaining staff, and closed the doors for good. A public auction to liquidate the remaining assets was held six weeks later.

  GO stockholders never received their EO stock certificates—which hardly mattered, because they were worthless. The only communication of any kind, formal or informal, from AT&T or EO to the original GO stockholders was a terse letter from Bob Evans announcing that the company would cease operations. He added a brief personal postscript: “Those who had the initial vision were correct. It is unfortunate that EO did not succeed to validate that vision.”

  In looking back over the entire GO-EO experience, it is tempting to blame the failure on management errors, aggressive actions by competitors, and indifference on the part of large corporate partners. While all these played important roles, the project might have withstood them if we had succeeded in building a useful product at a reasonable price that met a clear market need. Just as a coroner may have difficulty assigning a specific cause of death when a person dies of old age, it is hard to point to a single factor that alone was responsible for a company’s demise. Like a terminally ill patient, the absence of strong vital signs allowed problems that might otherwise have been correctable to grow unchecked, until the organism fell under their combined weight.

  The real question is not why the project died, but rather why it survived as long as it did with no meaningful sales. Without the unflagging efforts of a broad group of supporters, GO might have quietly closed its doors years earlier. The project’s longevity is a testament to the force of will and the compelling power of belief in an idea. It is comforting to know that the human impulse to make the world a better place is not today confined to the young or the foolish. It is alive and well among the people that live and work in Silicon Valley.

  After leaving Apple in October 1993, John Sculley took over as CEO of Spectrum Information Technologies, a small East Coast company. Shortly thereafter he resigned, after some preexisting irregularities were uncovered.

  In June 1994, Bob Kavner resigned from AT&T to take a position at Creative Artists Agency, the largest Hollywood talent agency, where he was to head its efforts in multimedia entertainment.

  After taking up John Doerr’s offer of temporary office space at Kleiner Perkins, Bill Campbell was asked to become CEO of Intuit, the maker of the popular Quicken financial software. Intuit soon agreed to be acquired by Microsoft.

  Randy Komisar became CEO of LucasArts Entertainment Company, a manufacturer of computer games, owned by George Lucas, the director best known for his Star Wars films.

  Robert Carr accepted a job as vice president of Core Technology, developing new products for Autodesk, the leading provider of computer-aided design software.

  After IBM acquired his company, Metaphor, David Liddle founded a Silicon Valley laboratory called Interval Research with the backing of Paul Allen, the other founder of Microsoft. Kevin Doren later joined the laboratory’s staff of respected scientists.

  After the FTC deadlocked, 2–2, on whether to take action against Microsoft and officially ended its investigation, the Justice Department announced, in August 1993, that it would initiate a similar antitrust probe of Microsoft. In July 1994 the Justice Department and Microsoft agreed to a consent decree, and the department filed an antitrust complaint alleging that Microsoft used exclusionary and anticompetitive licensing agreements to market certain of its software products and thus had unlawfully maintained its monopoly of personal computer operating systems. The consent decree specifically prohibited Microsoft from requiring its licensees to pay a “per processor” royalty on every product using an Intel chip, whether or not it uses a Microsoft operating system. The decree also prohibited Microsoft from entering into contracts that restrict PC manufacturers from licensing, selling, or distributing products using competing operating systems.

  As of December 1994, Apple’s Newton product line had produced only lackluster sales and was widely regarded as having fallen short of expectations. General Magic had just launched its first version of Magic Cap, in conjunction with AT&T. Microsoft announced that it had canceled Winpad, its unreleased successor to Pen Windows, after Intel dropped its plans to produce Polar, its response to AT&T’s Hobbit chip. The Winpad team was folded into the Windows group, to start over yet again.

  State Farm had completed a field test of its first Penpoint-based claims-estimating application. Its preliminary studies indicated that this one application could save State Farm as much as $85 million a year, and so Norm Vincent was preparing to order $50 million worth of pen computers. Seeing no future for Penpoint, he was finally persuaded to switch to Pen OS/2.

  After a few days of relaxing at home, I began to think about what to do next. Alan Fisher, an old friend of mine from my time at Teknowledge, called to see if I had any thoughts about how he could leverage some interesting technology he had recently developed. His team of software engineers had just completed a contract with Charles Schwab, the discount brokerage house, to build a Windows front end for its portfolio-tracking and order-management system, called Street Smart. “Basically, customers can dial up the Schwab data center from their home computers, update their portfolios, get stock quotes, place buy and sell orders, and browse through the latest news about companies and products,” he explained.

  I called him back the next day. “You know, it seems to me that buying and selling stocks is a lot like buying and selling anything else—it’s just much more efficient. When you think about it, why should there be a fixed price for anything, whether it’s stereos or cereal? I wonder what would happen if we put together the ultimate electronic market for consumer goods.”

  He was intrigued, but had his doubts. “El
ectronic retailing has had limited success at best. It’s horrible, just boring lists of items and prices.”

  “But suppose you did it differently,” I said. “You could build an interface that was engaging and attractive, with lots of pictures and product information. You make it free to the customers and just take a percentage of each sale. But most important, you let the prices float in response to supply and demand just like the stock market.”

  “You know what else,” he said. “Since we could see which items a person was looking at, and for how long, we would get information on their interests before they bought anything. Then we could shoot them an electronic coupon—say five bucks off on a pair of skis they were examining—but it’s only good for the next thirty minutes.”

  We both paused to imagine the possibilities. Alan was the first to speak. “Do you realize what this means? This could be a whole new form of retail, a whole new distribution channel for goods, with optimal pricing and infinite shelf space. A billion-dollar industry.”

  I felt that wonderful sense of epiphany again—but I didn’t know if I had the energy to blindly follow my heart this time. “Christ, Alan, I’d like to take a break before jumping into something like this right after GO.”

  “Yeah, I heard it was a pretty wild ride.”

  “I’m telling you, no one would believe it.”

  “Maybe you should write a book about it. Then we can get started.” Which I did, and it ends with this sentence.

  Author’s Note

  From the founding of GO in 1987 through its acquisition by AT&T in 1994, I kept a thorough diary with the idea of one day writing a book about the experience of starting a new company. For the first two years—until the time we moved the office from San Francisco to Foster City—I was able to find a few hours each weekend to sit down and write up the week’s important events. After the move, when GO started growing quickly, this proved impossible, so I bought a pocket tape recorder and kept an oral commentary instead, making recordings once or twice a week while commuting to work. Periodically, I sent off a completed tape to a typing service. The transcribed notes began arriving in large manila envelopes, which I stuffed into a drawer at home.

  When I transferred the contents into binders, I discovered that the approximately twenty hours of tape comprised almost a thousand pages of detailed notes. This running chronicle was supplemented by more than fifty megabytes of incoming and outgoing e-mail, memos, budgets, reports, minutes, proposals, and performance reviews. In addition to this electronic record, I collected appointment calendars, correspondence, brochures, newspaper clippings, drawings, prototypes, pictures, and videotapes. It was no secret to GO’s board of directors and employees that I was gathering information for a book project, and many of them helped by donating important documents.

  The careful reader will notice that I was not present for several scenes in the latter part of the book. To reconstruct these episodes, I relied on the taped recollections of as many of the participants as possible. I am deeply indebted to several people—especially Robert Carr, Bill Campbell, Randy Komisar, and John Doerr—who gave freely of their time to describe these scenes.

  From these sources and my own recollections I selected the small fraction of details and incidents that best represented the courage and imagination, frustration and humor inherent in a business venture. I made every effort to ensure accuracy, though in many places I had to reconstruct dialogue that I felt was true to the meaning, intent, and style of the speakers.

  I am grateful to John Sterling, my editor, who kept the story and pace on track, and particularly to Larry Cooper, my manuscript editor, who tuned up the prose and kept it running smoothly. My tireless agent, Kris Dahl, and her dedicated assistant, Dorothea Herrey, explained the inner workings of the publishing industry and provided invaluable feedback on interim manuscript drafts. Thanks also to Betsy Peterson for the legal review and to Joseph DeRuvo, Jr., for permission to reprint his letter to me. I am indebted to my wife, Layne, who provided her usual sage advice, and to our new baby girl, Lily, who was kind enough to wait until the first draft was complete before joining us.

  From this experience I learned three lessons: real life is infinitely more subtle and complex than what you can fit in a book; writing is more fun than reading, and easy work if you can get it; and no matter how much you warn people in advance, some of them still get upset when they see in print what they actually said and did. I apologize to all those who worked so hard for so many years and weren’t mentioned in the book. I also apologize to all those who worked so hard for so many years and were mentioned in the book.

  If you’d like to contact Jerry Kaplan regarding Startup, please send your e-mail messages to kaplanj@hmco.com.

  Chronology

  FEBRUARY 1987 Mitchell Kapor and Jerry Kaplan discuss the idea of a pen computer during a cross-country flight.

  AUGUST 1987 GO is formally incorporated.

  JULY 1988 Bill Gates of Microsoft is shown a deskbound prototype of GO’s pen computer.

  MARCH 1989 GO presents its plans to the Research Board, a group of the top information systems executives in the country.

  JUNE 1989 GO demonstrates its pen computer to State Farm, attempting to win the insurer’s business over IBM, Hewlett Packard, and Wang.

  MARCH 1990 After learning of State Farm’s decision to work with GO, IBM agrees to license GO’s Penpoint operating system.

  JULY 1990 GO announces a partnership with IBM.

  JANUARY 1991 GO’s Penpoint “developer’s release” is announced to broad computer industry acclaim.

  MARCH 1991 GO and Microsoft face off publicly for the first time at the PC Forum in Tucson.

  APRIL 1991 Jerry Kaplan briefs Federal Trade Commission staffers who are investigating possible antitrust violations by Microsoft.

  JUNE 1991 GO and Microsoft both present their operating systems at the NCR3125 pen computer announcement in New York City.

  JULY 1991 GO agrees to sell its hardware design group to AT&T and other investors. EO, an offshoot of GO, is formed to build pen computers based on Penpoint.

  OCTOBER 1991 IBM begins promoting Pen OS/2 as an alternative to Penpoint.

  APRIL 1992 Penpoint version 1.0 is released. IBM announces its Thinkpad pen computer.

  MAY 1992 John Sculley discusses Newton, Apple’s “personal digital assistant,” at the Consumer Electronics Show in Chicago.

  JULY 1992 GO announces a partnership with AT&T. Apple demonstrates Newton at the Mobile92 conference.

  NOVEMBER 1992 EO and AT&T demonstrate the EO440 personal communicator at COMDEX.

  FEBRUARY 1993 General Magic announces a partnership with AT&T, Sony, Motorola, Matsushita, and Phillips to promote Magic Cap.

  JUNE 1993 AT&T buys a majority position in EO.

  AUGUST 1993 After AT&T considers dropping GO’s Penpoint for Apple’s Newton, GO agrees in principle to sell the company to AT&T by merging with EO.

  JANUARY 1994 The EO-GO merger is completed.

  JULY 1994 AT&T decides to close down EO.

  Appendix

  The following is the full text of an e-mail message sent by Bill Gates to his senior staff soon after the meeting at GO described on [>]. (E-mail addresses have been deleted by the author.)

  From: billg Thu Jul 14 16:42:49 1988

  To: [6 names deleted]

  Subject: GO corporation

  Cc: [3 names deleted]

  Date: Thu Jul 14 16:42:45 1988

  Jeff Harbers and I met with Jerry Kaplan and Bob Carr of GO corporation Monday afternoon.

  Basically they are building a machine that Kay and I talked about building a long time ago—a machine with no keyboard and no disk using static memory. It’s like an 80286 version of the model 200 with 2meg-8meg using a writing stylus with handwriting recognition for input. According to Marquardt there are a few other people building things like this—in fact there was one discussed in the WSJ this week. It’s notebook size. The LCD is 640×400 so about 55
DPI (which I don’t think is enough).

  They are doing ALL their own system software—a protect mode OS for 286 using visual objects (like everyone!). It’s multitasking. The interface metaphor is a set of named folders with tabs on the right hand side each containing any number of numbered pages and each page has on it just ink (writing) or rectangles that contain application sessions (which can be zoomed/unzoomed). All the old ideas like using gestures for various commands they have “rediscovered.” They will announce in 1990 at $3k. Modem is optional. They will bundle some drawing/wp/filing/notetaking/mail software but they want to get third party ISVs including Microsoft.

  We tried out their handwriting stuff and it was terrible. It’s very possible to do this stuff correctly and maybe they will but they haven’t yet.

  ANALYSIS: This machine should be built as an open standard by a bunch of Japanese makers. The software layers should be more compatible with desktop stuff. Kaplan isn’t the best CEO. They have some OK ideas but I don’t think this thing will be big. We do need to think about note taking and the fact that small machines can be used everywhere especially with this input approach but I don’t think we should be an ISV for them.