Nathan Myhrvold looms as one of today’s great polymaths. Master’s degrees in geophysics and space physics at age 19, doctorate in mathematical and theoretical physics and an apprenticeship under Stephen Hawking, presidency of a software company-and all this before becoming Microsoft’s chief technology officer. He spearheaded the founding of Microsoft Research, one of the world’s most influential computer science labs, and played a leading role in a number of the company’s development projects, including some that contributed to Windows NT and Windows CE. Along the way he found time to train as a gourmet chef and learn to drive race cars. More recently, Myhrvold’s been digging for dinosaurs and mastering photography: his office is adorned with photos from trips to Hawaiian volcanoes, Alaskan tundra and the California desert.
But for Myhrvold, now 42 and with a fortune of several hundred million dollars, it is only a beginning. In January 2000, even before formally leaving Bill Gates’s fold, he cofounded Intellectual Ventures with former Microsoft chief software architect Edward Jung. Not exactly a venture capital firm because it’s funded by the founders, mainly to pursue their own ideas, the company is exploring everything from new forms of computing to biotech and genomics. Myhrvold is also thinking about launching the Invention Factory, an effort to unite leading inventors and change the way inventing is done (see “The Invention Factory,” TR May 2002). TR editor at large Robert Buderi visited Myhrvold in Bellevue, WA, to learn about his vision of a world in the midst of an unprecedented explosion of technological growth.
TR: I’ve seen Intellectual Ventures described as a biotech company, an incubator, a venture capital firm. What are you doing?
NM: The reason I decided to leave Microsoft is, I wanted to do whatever I wanted to do. So our mission here is about as eclectic as I am. One of the areas I’ve been very interested in is biotech, and no matter how broad a scope I can carve out for our research group at Microsoft, biotech was just a little too far out. So I’ve met lots of people, I’ve gotten smart on a bunch of areas, I’ve invested in some companies, I’ve talked about forming other companies, and that’s the stage I’m in.
I’ve [also] gotten very interested in the economics of technological revolutions, and how is it that we’ve had such enormous growth in technology over the past 30 years or so. I’ve spent a lot of time comparing the differences between the technological revolutions of the current century and latter part of the 20th century to the revolutions in, say, the 19th century. What I’ve found is it wasn’t just the technology that caused enormous economic change, because in the 19th century, between Edison and a whole host of other inventions and tremendous industrial growth, there were huge, huge changes. I believe there is something more fundamentally different. And it’s fundamentally different because of the emergence of exponentially growing technologies, where the next step is as big as all the previous steps put together, and you are dealing with changes that happen by factors of millions or billions. In the tech industry, we talk about Moore’s Law and the drastic changes in the price/performance ratio of processing power every 18 months or so. However, I’ve found that exponential growth can happen in other industries when certain criteria are in place. So I’ve been trying to put some order to this notion, since no aspect of classical economics really captures this idea.
TR: Can you explain this further? I would have thought that if you’re looking at electric power or telephony it would have exhibited exponential growth for a time.
NM: If you look at the number of homes connected to electricity, or the number of homes connected to a telephone, yes, you could say that the growth was exponential. [But] if you look at it from the perspective of the price/performance ratio, the cost of electrical power per kilowatt-hour or the cost of a minute of phone service, those costs really didn’t decline very much.
The cost of shipping was dramatically affected by the railroad, but we’ve hunted up all the records on what was the cost of shipping a ton of grain, all throughout the 19th century. It improved by about a factor of ten. The price of steel improved by maybe a factor of three. All of those things went through a price/performance ratio change. But typically less than a factor of 10. Whereas from the first transistor to today, the improvement was something like a factor of a billion.
There are dozens of areas undergoing a similar kind of exponential growth. The technological revolution for the 21st century is going to be based on which areas those kinds of exponential growth rates catch hold in, and which ones don’t. That is the key issue for the 21st century in technology.
TR: Besides semiconductors, what other technology areas have you identified as undergoing exponential growth?
NM: If you look at hard disks, the number of bits you get for a buck on a hard disk grows at about 125 percent per year. Mass-market software has an exponential price/performance curve. If you bought Microsoft Windows or Adobe PageMaker or-pick any other application you could possibly imagine-and you look at it over a period of time, you’re paying almost the same amount or even less in real dollars for an increasing amount of technology-an exponentially increasing amount.
The tools of molecular biology are on that kind of a path. Genomics is going through an exponential revolution. So regardless of whether you’re a doctor trying to save somebody’s life, or you’re trying to make new cosmetics, genomics and proteomics and the tools associated with them become indispensable. The Human Genome Project was a gigantic project, eight years, $12 billion, to sequence the entire human genome. And I claim it’ll be down to the $10 range to sequence individual genomes.
TR: To sequence your own genome?
NM: Not just your own genome, but every economically important plant and animal needs to be sequenced. Everything that eats us, all these disease organisms and parasites need to be sequenced so that we can develop cures and better protect ourselves. We’re on the cusp of so many interesting advances in so many parts of science and industry, but based largely on the fact that you have these incredible technologies that continue to be cheaper and cheaper and cheaper and cheaper.
TR: What areas will Intellectual Ventures-or the Invention Factory-be targeting?
NM: I’m very open minded as to what topics we’re going to look at. I have a background in computing. And I was once a physicist. So those areas are near and dear to my heart. Biology is interesting because you have a variety of breakthroughs, conceptual and instrumentation breakthroughs, that have made biology a symbolic, information-rich science.
TR: What is a symbolic science?
NM: Something with deep abstractions described by lots of data. Vast amounts of data-and analyzing abstract data is one of the most important frontiers in biology and medicine. So understanding which of your genes have this, that and the other thing, or which things are being expressed in your body right now. What proteins are in over- or undersupply. Where is there a feedback control system that’s screwed up. We’re on the verge of figuring out that or a million other very complicated systems. A key tool in that’s computing. So bioinformatics, bioinformatics algorithms. Most of that stuff is at its complete infancy. One thing that’s amusing to me is that when I visited proteomics companies, you get people, although they use computers, they use them in completely boneheaded ways. So everybody has big SQL databases, big Oracle databases, under the faith that that’s a good thing to do, when it’s completely ill suited. The relational database was designed for tasks such as tracking stock room inventory or managing employee information; it was not designed for manipulating genomic base pairs and genetic information. So somebody needs to invent a bunch of stuff there. But more than that, biology and medicine are about reverse-engineering a very complicated machine. The detailed understanding of all the mechanisms and pathways by which things are regulated and controlled, the ways in which disease disrupts those regulations and how we can put them right, that’s all incredibly complicated. Well, that suggests all kinds of opportunity. What tools are missing? What are the analysis techniques that you need to do? There are a million things.
TR: We have all this growth in technology, but you have been quite vocal in also pressing for more basic science.
NM: Basic science is the fundamental well from which all this stuff is watered. Ironically, basic science is being given increasingly short shrift. DARPA [the U.S. Defense Advanced Research Projects Agency] funding for computer science is probably the single most successful government program in the history of governments-it led to this entire revolution in computing. Yet most Silicon Valley companies that are the beneficiary of that don’t invest in fundamental research. Then you get the ludicrous thing of people in Congress saying they want more relevant research. No, you should have less relevant research.
I’ve done extensive modeling of all of this. If you’re a company that lives hand to mouth, don’t do research, okay. You don’t need me to tell you that. If you’re a company that has steady cash flows, then you should work at whatever level you can afford. So if you’re a company that intends to be around 20 years from now, like a Microsoft, you are losing money if you don’t do research. It is an incredibly profitable investment only open to a limited club-the people who can afford to take a long-term view. And that’s an industrial research context. At the government level, you really should swing for the fences.
You could make a case that research funding really won the Cold War, because it was those economic things that stoked the economy. As soon as the Soviets went from being our enemies to being potentially our friends, [people said,] now let’s stop giving lots of money to science. Well, that doesn’t make any sense. Fundamental science has been the best investment the government’s ever made.
TR: A big mark against basic research in industry is that the firms who support it don’t always capture the benefit of it-Bell Labs with the transistor, Xerox with so much of modern computing.
NM: Whether you’re expanding overseas or you’re doing any business decision, you can find someone who screwed it up and caused lots of hurt to their company. It hasn’t stopped people from doing it.
So take Xerox as an example. The same era that they started PARC [the Palo Alto Research Center, birthplace of the graphical user interface, Ethernet and other elements of digital computing], they bought a company called Scientific Data Systems. They lost a billion dollars in 1970 dollars on that. More money than they’ve spent on PARC the entire time they’ve had PARC. Nobody gives them any shit for that anymore. Everyone says, oh, Xerox screwed up PARC. They didn’t screw up PARC. PARC invented the laser printer. That one invention alone paid for PARC many times over. Yet people give Xerox a black eye for this. Why? Because they think, “But they should have done more.” Well, if you do shoulda, woulda, coulda, you’re going to drive yourself crazy. The problem that Xerox had-the fundamental problem-is that Xerox didn’t understand computers. That’s why they lost the billion dollars in that other merger. That’s also why they couldn’t commercialize any of the other computer inventions.
So you add it up, investing in basic research makes huge sense for companies. But it makes even more sense for the government. By the way, I’d love to have the rest of the world join us, because research is the kind of thing that feeds on other research. The fundamental researcher in China that isn’t being funded today might be the one who if he was funded would find the cure to the disease I’ll get in 20 years.
TR: You mentioned Microsoft as one company doing fundamental research, which you had a role in. What about the rap that Microsoft is not able to innovate?
NM: When I first got into computers, there was no Microsoft. IBM was considered this big company that dominated the industry and wasn’t very innovative, yet if you look at their patents or the history of the first things that they did, IBM was the most innovative company in large computers. IBM got a reputation for not being such because the computers they sold generally made sense, and their innovation was packaged in something that was incredibly pragmatic and practical.
Fast-forward 30 years, Microsoft’s in the same position. Microsoft tends to package its innovation in things which are incredibly practical. Yet they often are very, very innovative. Sometimes in incremental ways, sometimes in revolutionary ways. Let me take my favorite example of a Microsoft program that was way ahead of its time. Windows. I was development manager when it was 2.0. Everybody now acts like Windows was part of the firmament, destined to be a success. No, it was an incredibly hard battle to convince the industry that the graphical user interface was good. The key ideas were invented at Xerox; Apple and Microsoft both commercialized it. And both Apple and Microsoft deserve a tremendous amount of credit for that.
TR: Let’s come back to something you mentioned right off the bat-a theory that accounts for this new period of exponential growth. Can you elaborate?
NM: People want to have the next Silicon Valley. The interesting thing is not so much the next Silicon Valley in a geographic sense, it’s what are the next [technological] areas that will undergo this kind of growth two, five, 10 years and 50 years from now? And how does that reshape the world? I think it’s possible to go about that in a more deliberate way. To actually say, this is what you should look for. This is how you should do it-and then nurturing them over that hump.
But I’m still working on it.
This new data poisoning tool lets artists fight back against generative AI
The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models.
The Biggest Questions: What is death?
New neuroscience is challenging our understanding of the dying process—bringing opportunities for the living.
Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist
An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.
How to fix the internet
If we want online discourse to improve, we need to move beyond the big platforms.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.