Idea Man

by Paul Allen

  In years to come, I hope to find new ways to supply electric power and clean water in Africa, and to conserve threatened animal populations in the wild. If we do these things right, we’ll create a better future while still guarding and respecting the past.

  IN SUM

  When Saturday Night Live celebrated its twenty-fifth anniversary in 1999 in New York, Dan Aykroyd took me underneath the stage and flashed back to how it all started for him. “I was mainly a writer,” he said, “but then they asked me to do one of the first skits,” playing a home security technician who breaks into a home to show a terrified couple—John Belushi and Gilda Radner—that they need his service. Dan admitted he was nervous, “but the skit went great. When I came off the stage, I knew that I’d found what I wanted to do.”

  I know that feeling. I found my own path when I helped create Altair BASIC in that two-month rush of creativity back in 1975. Later, when the IBM PC shipped with our operating system at its core, it struck me that the code I had helped to write would fundamentally change the way people worked, played, and communicated. Having that kind of impact forever changes your sense of purpose in life. It’s a feeling you’ll always want to find again.

  When I became gravely ill in my twenties, I found myself regretting that my life was so narrowly focused. But after I recovered and traveled the world, I soon became restless. I discovered that what I missed most was creating things. And so I went back to work.

  If there’s any irony to my life, it’s that my time with Microsoft was atypically one-dimensional. When I was younger, I immersed myself in rockets, robots, music, and chemistry. An omnivorous reader, I thrilled to the exploits of pilots and explorers. I was inspired by Thomas Edison and Alexander Graham Bell. My curiosity was boundless.

  I went on to spend eight driven years with the single purpose of making Microsoft the leader of the personal computer revolution. And it happened, far beyond what I could have hoped or expected. But my old passions still tugged at me, deferred but not forgotten. They were squeezed into playing along with Hendrix at three in the morning or stealing a weekend from coding to watch a momentous spaceflight.

  After I left Microsoft, the wealth that I’d helped create there—and then the company’s explosive growth—freed me to pick up where I had left off. At times I cast my net too widely. But my choice of ventures wasn’t arbitrary. Most of them were seeded long ago, in my youth. Over the last twenty-seven years, I’ve been able to do things I once only imagined.

  I have now lived half my life post-Microsoft. What we achieved there will always be a source of pride. But my second act, in all its range and variety, is truer to my nature.

  SOME PEOPLE ARE motivated by a need for recognition, some by money, and some by a broad social goal. I start from a different place, from the love of ideas and the urge to put them into motion and see where they might lead. The creative path is rocky, with the risk of failure ever present and no guarantees. But even with its detours and blind alleys, it’s the only road that I find fulfilling.

  From early on in the Microsoft era, I was looked at as the source of seminal ideas. These days, my role is often to listen to smart people and recognize when something special has emerged. Then I try to place the thought into a new context or extend it into something more powerful, as we did in our neuroscience charrette. The idea of a genetic brain map had been batted about in many private meetings, but it crystallized when a dozen top scientists came together and engaged in a free flow of ideas. The Allen Brain Atlas, a product of their consensus, emerged as the most persuasive way to move the field forward.

  Few things worth doing can be done alone. To get past the conceptual stage, ideas need to become crusades; you’ve got to convince people to join you. I was lucky right off the bat to find Bill Gates, whose passion for business matched mine for tracking technology. Later I’d be fortunate to meet Bert Rutan en route to SpaceShipOne and to find Allan Jones to lead our brain work.

  I’ve also seen what can happen when the right team isn’t in place, how the best ideas can founder. I made more mistakes in pursuing the Wired World than I can count, but the first and worst was this: I often failed to find the right people to help me execute my vision. My own history probably swayed me to take a flier on some with slim track records and to entrust them with too much too soon. Since then I have learned to be more careful. Talent is indeed essential, but seasoning and maturity are not to be underestimated.

  Above all, I’ve learned the pitfalls of getting so locked in to looking ahead that you miss the pothole that makes you stumble, or the iceberg that sinks you. Still, any crusade requires optimism and the ambition to aim high. For as long as I can remember, I’ve wanted to find my own challenges, see them through to fruition, and—if everything breaks right—change the world for the better.

  PEOPLE ASK HOW wealth has changed me. It’s a question I find difficult to answer. There are times when I feel unaffected, and then I wonder if I’m kidding myself. The manifestations of wealth—homes, boats, planes—have clearly altered how I live and get around. More important, though, are the doors of possibility thrown open to me, the opportunities I’ve enjoyed.

  Yet for all these evident changes, the people I’ve known longest tell me that I’m much the same in the ways that matter. I still try to take people as I meet them. And I’m still a dreamer more intrigued by what might be doable than by what has already been done.

  My recent illness made me more impatient and patient, simultaneously. It was a harsh reminder that there is no time to waste, and it’s made me more urgent and demanding of myself and those who work with me. Still, it’s humbling to await the results of a PET scan and know that you can’t make the clock wind faster. I’ve come to realize that many things happen at their own pace, beyond your control, from the development of a young point guard to the trial of a potential Alzheimer’s therapy. I’m learning to be less harried in anticipation and more accepting of each necessary, incremental step.

  I do my best to keep up with science, technology, and current affairs, and most of my reading now takes place online. I want to keep stretching the boundaries of the possible; I want my thinking to stay forward-looking and unconstrained. What is next? That’s a question that will never get old for me. I’ll always be on the hunt for the next Big Idea.

  And there’s one more thing that hasn’t changed. I’m still fascinated by the inner workings of machines of all sorts; I still love to delve into their intricacies. At a minute level of detail, I’m doing it with the Allen Institute’s journey to understand the human brain, the most complex mechanism in the history of the planet. At the other end of the spectrum, I’m just now considering a new initiative with that magical contraption I never wearied of sketching as a boy: the rocket ship.

  Someone, after all, is going to have to get behind SpaceShipThree.

  ACKNOWLEDGMENTS

  The creation of this memoir has followed much the same path as the ideas that compose its subject matter. Inspiration may begin with an individual, but I learned long ago that it does not reach its full fruition without collaborative development. The more ambitious and challenging the project, the more intensive and inclusive that collaboration needs to be.

  Idea Man was highly challenging and, at least for me, extremely ambitious; it was one of the hardest things I’ve ever done. And so I have many people to thank. If anyone is omitted here, it reflects the limits of my memory and not of my gratitude.

  From start to end, I received vital guidance from what became an in-house editorial board: Richard Hutton, Bert Kolde, Jonathan Lazarus, and David Postman. They put in untold hours reviewing countless drafts and were invaluable as honest critics. I am genuinely in their debt.

  For their detailed review and keen-eyed comments on all or parts of the manuscript, I’d like to thank Rich Alderson, Dan Aykroyd, Lea Carpenter, Paul Ghaffari, Allen Israel, Rob Glaser, Mark Greaves, Allan Jones, Tod Leiweke, David Marquardt, Bill McGrath, Larry Miller, Dave Moore, Rosalyn N
guyen, Christina Orr-Cahall, Nancy Peretsman, Pat Peyser, Albert Rich, Alessandra Rubelli, Doug Shane, Mike Slade, Dave Stewart, William Turner, Jann Wenner, and Mark Zbikowski.

  I am grateful to all of those who contributed their recollections of the eventful times we shared—and who helped me to retrieve many memories that enrich this book. From my pre-Microsoft years: Bob Barnett, John Black, Craig Buhl, Monte Davidoff, Mike Flood, Bruce Flory, Doug Fullmer, Paul Gilbert, Tom Grubbs, Guela Johnson, Chris Larson, Marc McDonald, Rita Schenck, Dee Simpson-Snyder, and Jeff Wedgwood.

  From the Microsoft period: Richard Brodie, David Bunnell, Don Burtis, Eddie Currie, Pamela Duran, Bob Greenberg, Dottie Hall, Mike Hunter, Gordon Letwin, Kazuhiko Nishi, Bob O’Rear, Tim Patterson, Chris Peters, Vern Raburn, Gary Runyan, Charles Simonyi, Tandy Trower, and Steve Wood.

  And from my life after Microsoft: David Anderson, Jim Billmaier, Jim Boyden, Bucky Buckwalter, Sue Coliton, Lance Conn, Terry Davison, Ralph Derrickson, Glenn Edens, Marwan Fawaz, Rob Glaser, Harry Glickman, Mike Holmgren, David Liddle, Mike Melvill, Geoff Petrie, Tom Phillips, Kevin Pritchard, Geoff Reiss, Robbie Robertson, Burt Rutan, Bill Savoy, Neil Smit, Jill Tarter, Jennifer Todd, Nathan Troutman, Larry Wangberg, and Nick Wechsler.

  I am also thankful to those who provided more general or logistical assistance for this demanding project: Marilyn Valentine, Jane Repass, Dave Dysart, Anson Fatland, Bill Gates, Steve Hall, Miles Harris, Adrian Hunt, Elaine Jones, Ferina Keshavjee, Ian King, Betty Mayfield, Keith Perez, Allen Range, Nick Saggese, Will Stewart, Andrea Weatherhead, and Nathan Mumm, and the executive support group. And I’d like particularly to thank Erik Davidson, who helped with the jacket design and other graphics.

  Jill Jackson, my archivist, helped me locate contemporaneous documents that made this memoir more immediate. I also appreciate the help of Amy Stevenson, her counterpart at Microsoft.

  I didn’t fully realize it at the time, but the genesis of this book was an oral history project that I initiated in 2000 to preserve firsthand accounts from my school days through the early Microsoft years. My gratitude to Faye Gardner Allen, Chuck Bower, David Dekker, Roger Fisher, Stu Goldberg, Dick Hamlet, Andrea Lewis, Bob McCaw, Rudy Miller, Forrest Mims, Harvey Motulsky, Bud Pembroke, Steve Russell, Paul van Baalen, Nelson Winkless, Bill Weiher, Fred Wright, Marla Wood, Carl Young, and Robert Zaller.

  And to those who participated in the oral history project but are no longer with us: Miriam Lubow, Aaron Reynolds, Ed Roberts, Bob Wallace, and Ric Weiland.

  I’d like to thank my editors at Portfolio, Adrian Zackheim and David Moldawer, and their talented team: Emily Angell, Katherine Griggs, Jaime Putorti, and Gary Stimeling. And my literary agent, Esther Newberg at ICM, who brought me to them.

  I must acknowledge Dr. Brad Harris and Dr. Hank Kaplan for getting me through not one but two lifesaving recoveries, spaced nearly thirty years apart. They made it physically possible for me to finish this book.

  I am especially grateful to Valentina Turri, for her extraordinary support throughout this process. And to Jeff Coplon, who helped me to realize the book I had envisioned.

  My sister, Jody Allen, has played an indispensable role in my life as my business partner, sounding board, and the adviser who knows me best. As with so many of my initiatives, I couldn’t possibly have taken on this project without her encouragement and support.

  APPENDIX : ARTIFICIAL INTELLIGENCE, THE DIGITAL ARISTOTLE, AND PROJECT HALO

  Over the past thirty years, researchers have made real progress engineering artificial intelligence (AI) into commercial systems. Automatic translation, speech understanding, reasoning with constraints, logic, game playing, image recognition, and industrial robotics are all well on their way to being mastered. But one benchmark problem still exposes some of AI’s deepest remaining challenges: reproducing the simple act of reading a textbook, understanding the material inside, and answering questions about it.

  Why is this so difficult for computers? After all, learning new things, working though their implications, and answering questions are all second nature to us—we do them so easily that we rarely stop to consider the mechanisms involved. And computers certainly have enough raw power to do the job; modern search engines can sift through the Web in less than a second and deliver pages that match our search terms, ranked in order of usefulness. Nevertheless, getting a computer to answer ordinary questions of the sort commonly found on high school exams and answered by millions of students is extremely challenging to replicate.

  The problem has to do with the nature of human knowledge itself. Knowledge is often thought of as a large collection of facts, like multiplication tables or lists of chemical properties. Indeed, existing artificial intelligence technologies can answer questions that depend only on simple facts. (“How many chromosomes does a blue jay have?”). But the most important elements of human knowledge involve much more sophisticated constructions. Even cut-and-dried knowledge includes rough statements of causality (“Too little sunlight can lead to stunted plants”), generality (“Most birds can fly”), metaphor (“DNA is like a blueprint”), counter-factuals (“If Earth’s gravity were halved, trees could be twice as tall”), rule knowledge (“If a cell dies, its cell membrane disintegrates”), and prediction (“Mutations should increase in the presence of radioactivity”).

  The goal of the Digital Aristotle project is to find ways for computers to grapple with all types of human knowledge, and to manage and manipulate their full range and richness. In order to succeed, it will need to acquire knowledge intelligently, reason through it effectively, and find appropriate answers on a truly massive scale.

  Our Project Halo research program is designed to build the systems that can ultimately lead to a functional Digital Aristotle. We began Project Halo several years ago by targeting biology at the level of a high school Advanced Placement course. This subject area served our purpose because it has significant (but not overwhelming) scale, features a set curriculum with accepted tests for competence, and exhibits many of the more challenging types of knowledge. Thus far, we have analyzed standard biology textbooks line by line in order to categorize each type of knowledge they contain. Now we are working on ways to encode these types of knowledge into Project Halo’s computers, merge them with the knowledge that is already there, and keep everything in a form that will allow our various reasoning systems to respond with the correct answer to a user’s questions.

  The basic challenge in all this work is its pervasive brittleness. Many tough problems in computer knowledge encoding and reasoning have been successfully addressed at a small scale in a laboratory. But when these efforts scale up—even to the amount of knowledge in a single biology textbook—they break. Furthermore, the individual approaches are often incompatible with one another, and so current AI systems can’t match people’s fluid shifts between different ways of using their knowledge.

  The international Project Halo team has made considerable progress in our research. We believe that by 2015 we’ll be able to build a system that includes most of the knowledge required to answer Advanced Placement–level biology questions. This system, in the form of a tabletlike Halobook, will constitute an important step in our pursuit of the Digital Aristotle. Nevertheless, difficult challenges remain; the ultimate solution will require many more breakthroughs. Here are ten areas of knowledge representation that are currently formidable for machines to handle and are of interest to Project Halo, grouped into three tiers of difficulty:

  I. DIFFICULTY TIER 1: PROMISING APPROACHES STILL HALFWAY AT MOST TO A ROBUST SOLUTION

  Human language is powerful and complex. There are many ways of saying the same thing, and many different things communicated in every sentence. For a machine to process the full range of human language, it must “understand” and react appropriately to a huge variety of potential expression. Many promising techniques are being developed using both manual and automated analysis of language, including statistical studies of massive data sets drawn from the Web. T
he intersection of language and knowledge is an area that we have great interest in and are actively pursuing in Project Halo, dealing with the full range of linguistic expression.

  Visual/spatial learning and reasoning. Can seventeen suitcases fit inside the trunk of a typical car? What about an open umbrella? Can a jetliner land on a sidewalk? What information is represented in a diagram? How does DNA uncoil? Humans perform rough-and-ready spatial and visual reasoning tasks and visual simulations with ease. While the computational geometry that is needed for navigation, manufacturing, and architecture exists and is commercially available, progress has been much slower in dealing with the kind of intuitive geometry that we routinely use every day. Project Halo does not currently focus in this area but welcomes new ideas.

  Knowledge about actions, causality, and simulation. If a cup is on a table in a room, and a person enters the room, the cup is unaffected; it will still be on the table. But other things do change as a result of this action: the person will no longer be outside the room; the person’s body and clothes will be in the room; the room will no longer be empty, and so on. Humans effortlessly perform mental simulations in their heads, both in a “forward” direction to predict how events might play out, and in a “backward” direction to identify likely causes. Computationally, however, this is a difficult and long-standing problem for AI. Reasoning about actions, change, and causality is extremely complex, especially when an action’s effects are uncertain and have indirect consequences. The best current solutions are found in business processes, automatic planning, and robotics, but they tend to be highly customized and difficult to apply to new areas. Project Halo has made substantial progress in general reasoning about processes and actions.