CODE

Conversation : TECHNOLOGY

CODE

A Dialogue
John Brockman, George Dyson, J. Doyne Farmer, Steven R. Quartz, Tom Standage, Charles Simonyi [6.11.98]
Introduction by:
John Brockman, George Dyson, J. Doyne Farmer, Steven R. Quartz, Tom Standage, Charles Simonyi

CODE is an attempt to get at the foundational issues of the digital age. George Dyson has a biological approach and I have my own points to make based on an engagement with the founders of cybernetics that began in 1965. We are joined in the conversation by chaos and complexity theorist J, Doyne Farmer, biologist Steven R. Quartz, digital edtior of The Economist Tom Standage, and mathematician and software pioneer Charles Simonyi, creator of the WYSYWYG word processor.

—JB
GEORGE DYSON is a leading authority in the field of Aleut-Russian kayaks, and his work has been a subject of the PBS television showScientific American Frontiers. He is the author of Baidarka, Darwin Among The Machines, and Turing's Cathedral.

 

 

GEORGE DYSON: Everybody is worrying about Microsoft, and I think they're more or less missing the point. It's not whether a monopoly is good or bad, or whether it's breaking some rules to merge the browser with the operating system. Turning this into a political issue-Government versus Microsoft-is diverting attention from something much more significant: the growth of multi-cellular forms of organization on the Net. You have the same code-Windows-running on all the chips, and when you merge the Browser with that you get the same code running on all the chips, but also in communication, the way the cells of a metazoan are in communication. I don't think it's something we can stop-nor is it necessarily something we should stop. Nobody complains about UNIX. The development of multi-cellular operating systems is a separate issue from the question of whether what Microsoft does is fair or legal in a business sense.

BROCKMAN: Go back-first you mention the same code is running on all the chips...

DYSON: Not all, but we're talking 80-90 percent.

BROCKMAN: Second you're talking about multi-cellular digital organisms. How did we get to where we are now?

DYSON: The analogy with biological organisms is highly tenuous-as EDGE readers will be flooding your inbox to point out. It's just the beginnings of something, in a faintly metazoan sense. The operating system used to be the system that operated a computer. Now it is becoming something else. This all started with one computer, whichever one you choose, whether it was ENIAC, or the computer at the Institute for Advanced Study, or the machine in Manchester-you had one of these machines and it turns out it can do very useful stuff.

BROCKMAN: Was David Farber involved in ENIAC?

DYSON: No. But he's Alfred Moore professor at the University of Pennsylvania, where ENIAC was built. He's carrying on the tradition-it's like holding the Lucasian Chair.

BROCKMAN: Back in the '60s none of us had ever seen a computer. I remember leading a crew of artists to Harvard/MIT in '65-we went to see "the" computer. It wasn't about computers at all. It was about communications. Walter Rosenblith's field was sensory communications. Harold Edgerton was an electrical engineer; A.K. Soloman was a biophysicist. I don't recall meeting anyone who called himself a "computer scientist." Something important was lost when we started talking about hardware.

DYSON: So these things immediately started to communicate, by cards and paper tape and phone lines, nothing new or mysterious about that. But what's happening at Microsoft-and elsewhere-is a coalescence towards the complete communication of everything. As Farber would tell you-if you read his list, [IP, a mailing list that's a good way for someone outside the industry to keep up]-there are moves afoot to get the same code-Windows, or Windows CE, or Windows NT or whatever, not to mention underlying protocols-running everywhere. Running on your desktop, running on your network, running in your car, running in your toaster, running on the credit card you have in your wallet-it's all going to run this same code. And if it's not Windows it'll be something else. The thing is, it's happening. Which is very much what's gone on in the world of biology. In biology there is one operating system, and it's the one we're stuck with-the DNA/RNA operating system. All living organisms, with very rare exceptions, run that same system. That's not necessarily a bad thing, but...

BROCKMAN: So can I call this conversation "Life as an Operating System"?

DYSON: Maybe, but then you'll offend the biologists who say, "Oh, but it's much more complicated than that."

BROCKMAN: "Life as an Operating System, Sort Of."

DYSON: Or just "Operating System"-period. The power that Microsoft represents goes far beyond what we can ever imagine. Don't forget money-not the Microsoft Money program but real money-represented digitally, and incorporated into the operating system. It's inevitable. Most of the hard stuff is already in place. Money is cross-platform information, in a very powerful, fluid form. And a small percentage of it filters back to Redmond. It's like an ant hill or a termite nest. The ants collect crumbs, but the crumbs add up. You can take the view that it's dark and sinister, or you can say it's the coming of Utopia or whatever. I don't really advocate either position, I just think it needs to be treated as much more serious than the business of an oil monopoly or something like that.

BROCKMAN: More important than most of the players in the industry or justice department realize. We become the tools we create. In 1965 John Cage handed me a book to read. It was Cybernetics by Norbert Wiener. Then Marshall McLuhan turned me on to The Mathematical Theory of Communication by Bell Labs scientist Claude Shannon and Warren Weaver, which began: "The word communication will be used here in a very broad sense to include all of the procedures by which one mind may affect another." For Cage, mind had become socialized. By inventing electric technology, we had externalized our central nervous systems, and he wanted to tap into this by creating "a global utilities network."

DYSON: And that's exactly what happened. 1965 was the beginning of the time- sharing revolution, when one computer could be shared by many users. Now we have time-sharing turned inside out-when one user can be shared by many computers. Microsoft's "Digital_Nervous_System" isn't some cybernetic vision-it's a product with an advertising campaign.

BROCKMAN: It's on the mark in a nineties kind of way. And the big issue has nothing whatsoever to do with business, or government regulation. It's about who we are and who we will become.

DYSON: The question is, who does it belong to? We are all going to end up owning computers, but will we all end up owning shares?

BROCKMAN: Let's go back to ENIAC.

DYSON: OK. So you've got one computer alone that can be very powerful, but when they're in communication they become more powerful. It's the same way that a colony of cells with no nervous system at all can become a starfish or a sponge or something like that just simply by chemical communication.

BROCKMAN: By communication you're talking about a network such as the Internet?

DYSON: Yes, but you have to have all sorts of other communication to make an organism happen: chemical, hormonal, mechanical. We are still immersed in the metaphor of fifty years ago, the computer as brain, the brain as electrical network, etc. The metaphor we haven't quite got to yet will come from molecular biology, when we start to see the digital universe less as an electrical switching network or giant computer and more as an environment swimming with different levels of code. How these increasingly complex one-dimensional strings of code actually do things, interacting with each other and with the three-dimensional world we live in, has more in common with the code-string and protein-folding world of molecular biology, where molecules interact with each other-and do things-by means of templates, rather than by reference to some fault-intolerant system of numerical address.

BROCKMAN: There is no Internet-there is only a process. When you stop a process to name it, it becomes dead. What we think of as the Internet is only a measure of its effect.

DYSON: Look at it from the point of view of the code itself, not the end user sitting at a terminal, which is either a synapse to some other coded process, or the means to some formalizable end. In ancient (computer) times code would run, be executed, and be terminated, that was the end of it. On the Internet code can keep moving around; it may escape termination by the local CPU, and when it arrives at a terminal, that doesn't mean it stops.

BROCKMAN: How do you define "code"?

DYSON: Sequences of instructions, or data, that form either patterns in time or patterns in space. It's a very broad definition. For instance a sequence that when decoded by your machine turns into a song that you make copies of and thereby reproduce. When you write it to your disk it stops being a pattern in time and becomes a pattern in space. Computers transform patterns in time into patterns in space and back again, and they do it very fast-that's the whole Turing machine concept, the ability to make transformations between these two kinds of patterns, by formalizing a relationship between bit-to-bit (coding) on tape, and moment-to-moment (processing) in time. It's a symbiosis-the hardware doesn't make any sense without the code, and the code wouldn't exist without the hardware.

BROCKMAN: Multi-cellular?

DYSON: Danny Hillis has a good explanation of that-from when he started to do massively parallel computing. There's two kinds-single instruction multiple data, and multiple instruction multiple data. What you have in biology is sort of single instruction-you have one seed, which is one string of code, and then it divides and becomes all these different cells that differentiate into things-from cells to individuals to species-and they are all running this original mother code, but doing different things with it. That's what Windows is trying to do, to become this one seed of code that allows you to do all these different things-balance your check book, play your games, do your income tax, and everything else. And of course it has become bloated by trying to do all that. But then code in biology is bloated as well-that's one thing we've learned. We thought DNA must be so efficiently coded; but it's actually full of all this redundancy, because molecules are cheap, and editing is expensive.

BROCKMAN: So you say that this is not just a monopoly such as an oil monopoly?

DYSON: I think it's more serious. Because it is infiltrating everything.

BROCKMAN: There is an essential feedback process in which a technology relays back signals telling us what to do/who we are. Government is out of this feedback loop. Until only very recently no democratic populace, no legislative body, ever voted for what kind of information it desires. We didn't vote for the telephone, for the automobile, for printing, for airplanes, for the birth control pill, for antibiotics, for television, for xerography, for transistors, for space travel, for electricity. Governments play catch-up in terms of legal code. The other role government plays is to muscle in on the action and shake down the successful technologists. That's what we're seeing happen today.

DYSON: It's puzzling to me, as a historian, that government suddenly feels left out. From the 1890 census (the origins of the punched card industry and IBM), through the 1940s and 50s and right up into the 1990s, most of the critical innovations in computing (time-sharing, packet switching, HTML, etc.) were instigated by the government, or at least incubated with government support.

BROCKMAN: Right, and Buckminster Fuller and his colleague John McHale, rarely missed an opportunity to note that current military technology has a way of winding up in your dishwasher twenty years later.

But let's move on and talk about Jaron Lanier's thinking, i.e. that the architecture of the operating system is becoming embedded for a thousand years. Would you agree with that?

DYSON: Yes. The Year 3000 Problem! And the issue of monoculture vs. biodiversity in the software world. It has parallels with religion. Once established, they tend to last a very long time. We live in a world with many different religions, we've had some of the most vicious wars fought over issues of religion, and we've had no end of government involvement in religion. Yet we still have a world of diverse religions. With operating systems it looks like we may be losing that diversity.

BROCKMAN: And there have been quite a few up to now-Unix, etc.

DYSON: But the growth now is favoring Windows and Windows NT. And in the next generation those two are going to merge. And perhaps become much larger than Microsoft is today.

BROCKMAN: Is there something inherently sinister in this process? We both know a lot of people at Microsoft. They're not at all sinister.

DYSON: Which is why it's so wrong to treat this as simply a legal or business conflict-it isn't. It's the incorporation, by one corporation, of collective behavior that's moving at an unprecedented pace.

BROCKMAN: What does it mean?

DYSON: I don't know. What's remarkable is that we're not going to have to wait that long to find out. It used to be that you'd say "I sure wish I'd be alive in a hundred years to see what happens"-if we live five more years we're going to see what happens.

BROCKMAN: Is it going to be a good thing if and when there will be no Netscape? You will be limited to accessing the universe of information through Microsoft's eyes.

DYSON: At the beginning, the browser and the operating system were symbiotic bodies of code. But then one swallowed the other. That's probably how we have the modern living cell, with all its embedded subsystems, because free-living symbionts were absorbed into the cell. That's what's happening with the browser, it's gone from being an outside symbiotic body of code to something that's swallowed by the operating system and become the nucleus of it. It's a very sensible way to do it, just to be able to browse everything, whether it's on your disk or on somebody else's. The problem here is that Netscape got incorporated not by symbiosis but by imitation, and people sense that somehow this isn't fair. (And then you hear, "But who imitated Mosaic?")

BROCKMAN: Any advice to the Justice Department?

DYSON: Lay off this question of whether you can merge your browser and your operating system and these other vague things-all they can possibly lead to is being argued about in court for ten years. Send a bunch of hard-nosed lawyers in there who understand business deals and can crack down on some of the details-any number of smaller cases where Microsoft has pushed their weight around-but not these big religious issues that can't be solved. Make sure they obey the absolute letter of the law.

BROCKMAN: What's a religious issue?

DYSON: Well, the issue of whether Microsoft is a monopoly or not, or where you draw the line between applications and operating systems. Those are tough things to legally decide. And can you really do anything if you decide them?

BROCKMAN: Are you saying that there's no point in breaking up Microsoft and having an operating system company and an applications company that compete?

DYSON: Right, because the only way you can break it up is by forcing some larger government administrative structure upon it, so the cure is worse than the disease. One thing we know about regulation is that it's very, very slow, and it's usually about ten years behind. Microsoft may exercise its power unwisely, but government inefficiency may be worse.

BROCKMAN: The Justice department's involvement on the technological level is off the mark. There are issues to consider that are more important than Microsoft, Netscape, "the consumers", or the today's economy. We don't need Justice, Congress, the lawyers for this.

DYSON: We need biologists. Molecular biologists and field biologists. Entomologists. Immunologists. Viral geneticists-they can tell you how to write (or evolve) robust code. As far as I know, there's almost no biologists at Microsoft. Lots of physicists, and four-dimensional topologists even, and of course Nathan's work with dinosaurs, but not much else. Maybe they're keeping it quiet. It reminds me of Von Neumann's computer group at the Institute in the 1950s-Charney's meteorology group was a convenient smoke screen for all the calculations being done on thermonuclear bombs. But the bombs were sort of an open secret. There was a much deeper secret, however: Nils Barricelli's numerical symbioorganisms. No one dared draw attention to that.

BROCKMAN: Have you discussed this with Microsoft?

DYSON: I was invited to visit Microsoft-and gave my pitch for software evolution as a somewhat haphazard symbiogenetic process, and some of the programmers seemed to take this as a criticism of their work. Programmers write code, code doesn't self-evolve.

BROCKMAN: What was your pitch?

DYSON: In nature, every possible variation of code is tried sooner or later and nature selects what works. You throw code at the universe and see what grows. That in a very crude sense is what I see happening at Microsoft. There are 13,000 people, many of them writing code. Whole divisions write code for a year and if it doesn't work and the market doesn't buy it, it's dead-if it's something that works, if something's successful, it grows. You throw money like grass seed in a park and watch where the paths form. There are some very clever programmers but can anyone predict ahead of time what's going to work? I think it's much more an element of chance. It's not random- you see the successful things because they're the ones that get to market, but it doesn't take thousands of people to write-even to write an operating system.

BROCKMAN: How does it happen?

DYSON: Systems grow by symbiosis. Remember the System Development Corporation, which was started in the early 1950s as a small subdivision of RAND, to write operating systems for air defense. By the mid-1950s it had grown to twice the size of the rest of RAND. Ashby's Law of Requisite Variety says that effective control systems have to be at least as complex as the systems they control. So you have to use components--and hierarchical languages. No one could engineer something as complicated as Windows 95 from scratch; it has to be built up from other autonomous things that are known to work. The code has a life of its own-it has to go out in the world like biological code and do something, and then the response goes back to the source and if it's successful it gets reproduced-or imitated, which gives digital evolution a faintly Lamarckian quality that's absent in the natural world.

BROCKMAN: Have you had this discussion with Charles Simonyi?

DYSON: Only in snippets. His project on intentional programming is way ahead of the curve. He's a mathematician, and he can think in more than two or three dimensions. There's always a higher dimension than the one in which you are writing the code. There's always another level-the language above the language-and this IP-Intentional Programming -project is a way of opening a doorway so that something successful at one level can be extended to the other levels without this incredibly laborious process. It becomes less brittle. But it's not just another language. Languages form layers, whereas IP, as I understand it, has depth.

BROCKMAN: Software is the only business today.

DYSON: In the 1930s it took a visionary to see this coming. Turing (and Goedel) said that everything can be coded-people laughed and said, oh, those romantic mathematicians are imagining this unreal stuff. In the 1940s it started to happen.

BROCKMAN: A notion that descended directly to the logic of By The Late John Brockman . Everything's being coded. And now, it's going to be coded through Windows.

DYSON: Exactly. That's the amazing thing-technically Windows is just a number. One very long number. You buy Windows, it's on a compact disk, it's just one long string of bits. If you tried to type it out as a book, you would be typing for a very long time. In Turing's day this all seemed ridiculously abstract-the idea that you could have some kind of universal number, and here Windows is the idea of a universal number, carried to reality and shrink-wrapped. If you took somebody 50 years ago and tried to tell them this is going to happen in 50 years they wouldn't believe it.

BROCKMAN: But it's just a string of bits.

DYSON: Yes, it is. But let me remind you of something "which might interest biologists more than artificial intelligencers," as logician John Myhill put it in 1964. "The possibility of producing an infinite sequence of varieties of descendants from a single program... suggests the possibility of encoding a potentially infinite number of directions to posterity on a finitely long chromosomal tape."

BROCKMAN: Who owns the tape?

DYSON: Good question.

Reality Club Discussion

charles_simonyi's picture
Charles Simonyi
Software Engineer, Computer Scientist, Entrepreneur, Philanthropist

It is evident that two different metaphors can be read from the biological / evolutionary analogy that George Dyson has advanced.

People concerned with monopolies rely on classic examples such as the power utility grid. I view utilities as entities that occupy difficult-to-acquire land, accommodate expected future growth, that operate at a very high level of efficiency (well over 50% and probably more like 95%), and where technological advances (superconduction, or hydrogen-economy) would not be expected before the investment is completely amortized. Similarly old-time telecommunications operated in the difficult-to-acquire lower frequencies, and until recently did not have serious technological competition. Government created and government enforced monopolies (such as AT&T was until recently) required government regulation.

My view of Dyson's analogy is that the standards are necessary and useful but they are as dynamic and as ephemeral as what I jokingly surmised the first "eleven" steps of the creation of DNA must have been. Compared with the utility, there is no limited resource of any kind, future growth is expected to be exponential, efficiency (measured against theoretical limits of performance) is very low (yes, I believe that), and the product will be obsolete relatively quickly. So in this sense it is important to locate ourselves in the evolutionary map: are we at some pinnacle where we are so flush with our incredible luck that we crave security and conservatism, or are we at the bottom of Ïmount improbableÓ (to borrow Dawkins' phrase) where we feel good about the progress but want more. I submit the latter is the case in software. Look at the movie 2001: we are not even 5% there. Until a few years ago the government was worrying about DOS. The biggest problem with nationalizing Windows would be having to use Windows 98 in 2010! Windows 98 will be obsolete in a few years. It will be replaced by a Microsoft product, if we do our job right, or by a competitor's product if we don't. Or do you really believe that millions of consumers would shun brand X's superior voice-recognizing natural language based operating system because of some former network efficiency?

One alternative to nationalization that is proposed is to "regulate" software development because of the control software producers have. I have to admit I am unsure what is meant by control: I understand the control the utility company exercises: I can not trespass on their right-of-way to build my transmission lines, I can not send energy or even signals up or down their lines, and if I do not pay my bills they shut off the supply at my doorstep, and there is only one line to the house. Now software is a simple one-time transaction: you pay the bill, you get the license for the bits on the diskette, very much like a newspaper. You can keep it, run it, replace it. It will do what it does forever, it will never be shut off. And many can and will write software at the least provocation and incentive, for example a Finnish hacker called Linus Torvalds (who wrote Linux). How the software can control anything that YOU the user did not explicitly put in charge of, has always escaped me. Now I understand the general lament that YOU feel obligated to have to put it in charge of things and YOU feel obligated to update to a newer version because that is how network efficiencies work out and that you may resent that obligation. But are your feelings of obligation absolute? If the competition saved you money, offered you a box that was more stylish, had a friendlier user interface, more direct or wider access to the Internet, and it handled data formats which mattered to you the most, would you still go for some abstract network efficiency instead? Of course not: but in a sense network efficiency IS having all those things, so I am not denying that network efficiency exists, just claiming that network efficiency is not different than customer satisfaction and therefore that network efficiency can not be maintained by ANY other means than by maintaining customer satisfaction. That is a kind of control, and a legitimate one.

The large market share some software enjoys is very similar to the extreme popularity of a relatively few entertainment and sport stars which started when broadcast radio and broadcast TV became available. See "When Winners Take All," The Economist, November 95. In case of entertainment, it is not argued that popularity somehow implies control and that it may be unrelated to performance. Yet the same kind of mechanisms are at work: the second best in any category gets nonlinearly less attention if the supply of the best is unlimited as is the case in broadcast media

j_doyne_farmer's picture
J. Doyne Farmer
Professor of Mathematics, Oxford University; Director, Complexity Economics, Institute for New Economic Thinking, Oxford Martin School; Co-founder, Prediction Company

I don't think that the connection that George Dyson has made between cells and operating systems is far fetched at all. In any case it doesn't really matter; the key point is that there are enormous advantages to having a standardized platform that all applications run on, and it is bound to happen. I agree that the implications and consequences of this monopoly are much more far reaching than Rockefeller's control of the oil industry. This is much more than a monopoly — it is control of society's replication machinery for ideas.

To me, this makes it vital that the government step in and block Microsoft from gaining absolute control. If this were a public domain, free-ware product such as Linux, this would be a different story. Nobody complains about UNIX because it is not really a commercial product. But to give total control of such a key element of society's information processing apparatus to a private company is very dangerous. Even if the government is likely to be inept, they are at least subject to the checks and balances of a democracy. Ineptness is the conservative course of action. Of course it is true that this is much more than a monopoly, and the anti-trust laws were not designed to deal with this kind of thing. But laws are just like blocks of DNA code: In typical evolutionary style, entirely new, fully formed parts are rarely created from scratch; rather, existing parts are adapted to perform new functions. An analogy is the endangered species act: What is really needed is an "endangered ecosystems act", but since no such thing exists, environmentalists stretch the endangered species act to its limits in an effort to make do.

George's analogy breaks down because the innovation mechanism for memes is not random variation — it is conscious design. It is one thing to settle on a standard mechanism for replicating RNA — and entirely another to give this control to a consciously directed, profit making entity.

george_dyson's picture
George Dyson
Science Historian; Author, Analogia

Charles and Tom are right. It's a woolly analogy and a metaphysical discussion.
To Tom, I would say that I think "random mutation" is overplayed in natural evolution, and underplayed in technological evolution-but treacherous ground, there. I'm not being teleological, just arguing that less-than-random recombination does the heavy lifting in biology-and in technology-it seems to me.

To Charles, I would confirm that yes, the use of "cellular" and "multicellular" is used to convey the notion of distributed process, not the suggestion that even our most complex machines (Windows coded or otherwise) resemble biological cells in any but the most rudimentary ways. Charles mentions genetic takeover, from clay crystal templates to nucleotides, or however the unknown steps played out-and this, I think, is one of the keys to imagining how the otherwise unimaginable future might unfold.

George

tom_standage's picture
Tom Standage
Business Affairs Editor, The Economist; Author, The Edible History of the Humanity

I was very interested to see exactly how George Dyson likened software development to biological processes. But I can't help thinking that it's a rather woolly analogy; the "programmers write code, code doesn't self-evolve" objection is valid, because although code has to prove it is "fit" in the marketplace, it doesn't arise randomly (despite appearances). There isn't really any mutation going on, either.

I've often wondered whether the three pillars of Darwinism (mutation, inheritance, competition) could somehow be aligned with the three pillars of Object Oriented Programming (encapsulation, inheritance, polymorphism). I haven't figured out a satisfactory way of doing this, though.

But if OOP is not Darwinian, perhaps some day a form of software engineering will be devised that is (and I'm ignoring the various artificial life experiments here, because they tend not to result in commercial software). One possible candidate is Genetix, a programming system based on machine code fragments called "genes"; see http://www.ieee.ca/genetix/ for more information. In the mean time, I remain not entirely convinced by software/biology comparisons.

Tom.

charles_simonyi's picture
Charles Simonyi
Software Engineer, Computer Scientist, Entrepreneur, Philanthropist

With the Appelate Court siding with Microsoft's position it becomes much easier to discuss these issues in the abstract. I mean that when the Law is engaged in metaphysical matters, abstract metaphysical discussions can easily have very concrete physical results.

I thank George for his very generous comments regarding my project, Intentional Programming. I employ a lot of biological metaphors, for example when I talk about an "ecology of abstractions". I compare trying to separate abstractions from computer languages to Dawkins' insight to focus on the reproduction of genes instead of the more obvious reproduction of gene-carrying organisms. But one has to be very careful about extrapolating from metaphors, and this is where I would advice caution.

George makes a very good point that the old metaphors of cyberspace being a computer or a communications network are not helpful. There IS something organic about how cyberspace is developing and George describes very well why we should feel that way.

However I feel that the specific connection between cells and Windows-based machines is far fetched. I suspect that one reason for George's metaphor is that the evolutionary steps leading up to DNA themselves are so poorly understood (I understand little documentation was done during development and then most of the documentation was lost during the last 5 billion years and now we have to try to reverse-engineer the bootstrap process from code fragments). So we may have had some clay crystals for templates, amino acids forming at the edges of temperature gradients, RNA, DNA, what have you. It was probably a 12 step process. Because none of these is agreed on, George HAD to chose cells as the simplest biological representative in his metaphor.

Since we all believe in the eventual emergence of cyberspace, by continuity we know that Windows will have to have a place in its evolution, just as Jacquard, Babbage, Boole, Hollerith, Zuse and v.Neumann will have. I just do not think that it would be helpful to draw parallel lines from each of these to some specific complexity of organization in the evolution of life especially given our current state of ignorance of the initial stages.

steven_r_quartz's picture
Steven R. Quartz
Neuroscientist; Professor of Philosophy, Caltech; Co-author, Cool

BIOLOGICAL METAPHORS, MICROSOFT, AND THE ECONOMICS OF IDEAS

George Dyson's intriguing comments on biological systems and code development have helped me think about the issues in the news lately involving Microsoft and the Justice Dept. Several commentators have already raised one issue George anticipated, problems in the analogy between random genetic change and the directed or Lamarkian change governing programming. These are essentially the same sorts of objections raised against memes and the process of their discovery and innovation. In mulling the dialogue over, I think there's a more basic difficulty with the use of biological metaphors as a framework for thinking about code in general and the more specific issue of antitrust law and monopolization, one that distorts much of the issue.

What I've read in the press has colored the antitrust probes of Microsoft with the language of classical commodity-production economics. We've thus seen comparisons to Rockefeller, Standard Oil, and cartels. Even Robert Kuttner suggests that "though the industries and technologies change, the economic fundamentals do not" (http://www.epn.org/kuttner/bk980518.html). In contrast, Bill Gates has tried to argue that the assumptions behind the antirust laws are inapplicable to the realities of a new economy. Although he's met limited success in terms of public persuasion, after mulling it over, I think he's right on some key measures. In fact, as I'll suggest, I think the biological metaphor that's at the core of George Dyson's intriguing comments also belongs to classical economics. Like much of the debate in the press, I think it misdiagnoses the key issues involved. As interesting as Dyson's comments are, the biological metaphor commits us to a mental model that distorts the principles of an emerging economics of ideas. To understand it, we should either choose new metaphors or approach it on its own terms.

One of the cardinal principles of commodity-production economics is the law of diminishing returns, a growth-limiting principle that among other things shapes a model of competition. As voiced by such economists as Paul Romer and Brian Arthur, a key proposal concerning the emergence of a knowledge economy is whether the law of diminishing returns applies. One of the most intriguing proposals is that in a knowledge economy a new rule applies, the law of increasing returns. In terms of the process of discovery, this basically says that ideas lead to more new ideas without diminishing some fixed pool of ideas. In terms of production and distribution, it means that costs per unit are low and go lower per unit produced (but not to be confused with a scaling principle). An example is the cost of distributing software over the Internet, where the cost per copy to the company is almost nil.

What rules apply to biological systems? As physical things, they face a number of diminishing returns, ranging from scarce resources in the environment to parental investment in offspring. Although I don't know the history well enough to try, a case could probably be made that diminishing returns is the central notion behind both Darwin and Wallace's theory. They were, for example, both influenced by Malthus' Essay on the Principle of Population in which diminishing returns figures centrally. In fact, in biological systems there's diminishing returns with a vengeance: gaining too much "market share" by becoming too successful can actually decrease fitness as per frequency dependent selection.

Seeing that biological systems and diminishing returns are highly intertwined isn't a minor point. It reveals that the biological metaphor is only applicable to a physical economy of resource extraction and commodity production. As I intimated briefly with the influence of Malthus on both Darwin and Wallace, the biological metaphor has had a major role in shaping how we envision economics. In a physical economy, for example, diminishing returns generally ensures that some equilibrium will be reached among competitors. But this is just Spencer's justification for laissez faire policies on the basis of Darwinism. If, however, a knowledge economy is characterized by increasing returns, then the biologically-derived notions of competition no longer apply and the dynamics of competition will be substantially different. And so, biological metaphors don't help us understand the nature of a knowledge economy. The lessons of Ford and GM as seen through a biological metaphor don't hold for Netscape and Microsoft.

Indeed, as Romer, Arthur, and others note, increasing returns predicts the emergence of monopolies and one should expect them. For one thing, the impediments to market dominance stemming from diminishing returns no longer apply. Increasing returns make it reasonable to pursue first-in strategies like gaining market dominance by flooding a product into a market for free (to leverage another product). You'd never expect GM to give away their cars to gain marketshare. For software, there's a further push toward market domination: bandwagon, or network, effects like becoming the market standard increase the momentum of a product.

If that were the end of it, then things would be pretty pessimistic-someone comes along and establishes themselves as a monopoly and no one can budge them from their dominance by introducing a similar product. This is the typical image of Microsoft in the press. But this seems to be an artifact of applying the biological metaphor and its model of competition. Increasing returns suggests a different model, one known as monopolistic competition. As I mentioned, increasing returns has two senses, one regarding innovation and the other regarding production and distribution. The feasibility of monopolistic competition rests on the increasing return of ideas. Monopolistic competition doesn't involve the competition among similar products for marketshare on the basis of pricing and other competitive factors. That's the biologically-derived one based on diminishing returns. Instead, it occurs when a new innovation results in a new product or piece of code that makes a dominant one obsolete-they're not called killer apps for nothing. This replaces one monopoly, or market-dominating product, with a new monopoly. Integrating a browser into an operating system seems like a classic case of monopolistic competition. But that's OK in the new economy. And wasn't Netscape trying to integrate an operating system into their browser? That's monopolistic competition too.

This brings up another point against the standard biological metaphor. The biological metaphor embraces incremental change, the gradualism Dyson's comments frequently invoked. In physical economies, incremental change is the kind of continual improvement production methods strive for. Schumpeter's phrase "creative destruction" gets tossed around a lot in discussions of an emerging knowledge economy. But if it approximates the kind of instability and dynamics we can expect in a knowledge economy, then gradualism is the wrong model of change. If one were tempted by biological metaphors, punctuated equilibria and speciation would seem better ones, operating in what's sometimes called Internet time. Netscape's phenomenal growth was akin to a speciation event, owing to their first-in position and increasing returns. But the lack of stable equilibria in such environments implies equally fast declines. What that suggests is that the successful corporation will be one that doesn't adopt a continual improvement model of product development, but instead sees their product line as a succession of replacement events. Historically, companies whose products have market dominance tend toward conservatism, since short-term returns typically dominate longer-term interests. Tom Peters has been screaming for years now against this conservatism. What it suggests to me is that the whole issue of the "fairness" of integrating a browser with an operating system misses the point-it assumes a model of incremental change and doesn't recognize the requisite model of change in an idea economy as akin to speciation (though I'm wary of the biological metaphor). If Microsoft stuck to a model of incremental change the way Ford changes the headlights of a pickup on a new model year, you'd wonder how long it would take to lose their market position. If I were Netscape, I'd take the fact that its stock is trading below its IPO price as a pretty good signal that it's time for their own "speciation" event. But that also makes practicing creative destruction easier than defending market position, and a viable corporation over the long run will be one that is ready to replace its own products. That seems to me the kind of normal life cycle to expect. Though jarring from the perspective of a physical economy, it also enables the growth rates that make an idea economy so enticing. And it points to the kinds of instabilities indicating why monopolies will be temporary and that a succession of monopolies will be the norm.

I don't mean to suggest that there aren't facets of antitrust legislation that pertain to this case. My only point is that much of the discourse in the press seems distorted by inappropriate economic assumptions, which the biological metaphor plays into. The flipside of this is that some of the really difficult and pertinent issues like intellectual property rights have been neglected. Actually, if I were a Fed, I'd be most concerned about the acquisition of startups. That's where the innovative ideas that make monopolistic competition a feasible idea are likely to come from.

There's lots of intriguing things about this new economy. The impact of information technology on productivity, for example, is still poorly understood, perhaps because traditional GDP accounting misses their contribution. My own interest lies in the intellectual capital challenges the new economy presents and how organizations from schools to corporations should organize themselves to maximize that capital. Rather than limiting our understanding with potentially misleading metaphors, though, responding to the new economy's challenges and opportunities requires understanding it on its own terms. The dialogue that George Dyson has prompted is the kind needed to sharpen our understanding

j_doyne_farmer's picture
J. Doyne Farmer
Professor of Mathematics, Oxford University; Director, Complexity Economics, Institute for New Economic Thinking, Oxford Martin School; Co-founder, Prediction Company

In response to Charles' latest:

My memory of the match monopoly is dredged from a high school social studies class. I only remember that there was someone who was called "The Match King" who managed to corner the market, I think around the turn of the century. A trip to my Encyclopedia Britannica fails to help on this point. However, it does offer the Swedish as the inventors of safety matches (1844), and credits an American, William Gates, Jr., as the inventor of the first mechanized match manufacturing machine!

Looking up monopolies is even more interesting. They point out that "corners and combinations were prohibited by the most ancient laws of China, India, and Babylonia". They then trace the history of monopolies and their regulation from Greece, Egypt, and Rome through medieval times and this century. Particularly interesting is the discussion of the emergence of monopolies in the US in the last part of the 19th century. Apparently it was originally impossible for US corporations to own stock in other corporations, which suppressed the formation of monopolistic combinations. But in 1882 Standard Oil figured out a legal maneuver getting around this, and by 1990 monopolies were widespread; they mention sugar and distilling as examples.

This makes it clear that the emergence of monopolies is very sensitive to the nuances of the legal structure and in particular the legal definition of a corporation. It is also clear that monopolies and their regulation by governments are almost as old as commercial civilization, and the proper way to regulate monopolies has always been controversial. So the debate we are having is nothing new.

Regarding the Microsoft situation, my only assertion is that there is a proper regulatory role for the government. The mere fact that there have been a series of court cases puts heat on Microsoft not to take advantage of their dominant role to squash all the competition. Personally I am quite happy with integration as long as the hooks are made available so the competition can integrate their products as well. I don't know if the appellate court agreed with me on this, but the fact that the decision went first one way and then the other illustrates that it is not a simple issue.

It seems that Microsoft is just the latest chapter in a 4000 year old debate.

charles_simonyi's picture
Charles Simonyi
Software Engineer, Computer Scientist, Entrepreneur, Philanthropist

Dear Doyne,

I recall your example of utilities as an area where government control is important. I extrapolated that you might think that utilities and operating systems share some traits since you imply that the arguments are transferable otherwise the example would have been irrelevant.

I admit I have never heard of the match monopoly. I am naturally interested in the fiendish devices they must have used to coerce the public into buying their monopoly matches and to frighten away the competition from dipping pieces of wood into phosphorus. :-) I must also add that in my elementary school I was told that safety matches were, yes, a Hungarian invention.

Your message then departs from the evolutionary issue and becomes a straightforward restatement of the DOJ's anti-Microsoft case which has been rebuked recently in large part by the Court of Appeals. The details are on the net (e.g. microsoft.com/presspass/doj lists all the rulings). Maybe you are against bundling but accept the need for integration of needed functionality - then you would be agreeing with the Microsoft position. DOJ is against bundling, claims - wrongly, according to the Court of Appeals - that we bundle, and just to be on the safe side they are also against integration, i.e. extended or evolved functionality, too. We admit to having committed integration and claim that it is good and legal. And evolutionary. Where do you stand?

Best, Charles

j_doyne_farmer's picture
J. Doyne Farmer
Professor of Mathematics, Oxford University; Director, Complexity Economics, Institute for New Economic Thinking, Oxford Martin School; Co-founder, Prediction Company

I don't disagree with a lot of what Charles Simonyi says. But I never claimed standards were static, or any such silly thing. This doesn't mean that dynamic and evolving standards don't exist; WindowsXX, provides a good example. This is reminiscent of the standard joke: Question: "What will the computer language of 2010 be?" Answer: "I don't know, but it will be called Fortran."
To get at the meat of what is going on here, I think we need to discuss the specifics of this particular situation, which I don't think is very complicated.

"Control" for a software provider comes about when they have such a large fraction of the market share that a large fraction of software products run only in that software environment. For example, at the moment I am stuck using Windows for some purposes because the only terminal emulator that can really keep up with Xwindows, GoGlobal, runs in Windows and not on a Mac. This product is not better because it is running on Windows; rather, it just isn't worth it for them to write a version for other operating systems with much lower market share. If there were a viable non-Windows alternative I would take it. Its dominant market share forces me into a situation where dealing with Windows is unavoidable unless I am willing to take a sacrifice in performance.

Fine. This is one of the rewards reaped by Microsoft for getting the dominant market share. What is bothersome is when that same company begins to use its monopolistic position to its advantage by doing things like bundling their own net browser, and making it difficult to use other net browsers. I don't deny this is a smart business tactic — if you can get away with it. But I don't see how one can argue that it helps consumers. Okay, you say, the browser is free. But not really — why couldn't you just lower the price on the operating system and sell them separately. (Answer: This wouldn't help Microsoft get the business away from Netscape). From a distance, it looks like the classic tactic used by a monopoly to strengthen its position as a monopoly. Use the monopoly to lower the price, drive the competition out of business, and then raise the price again. Monopolies have a long history — this strategy has succeeded many times, for matches, oil, and many other "non-government created" monopolies, and without government intervention would have gone on working, to the detriment of the consumer. To an external observer this interpretation of the microsoft situation just seems like common sense. But perhaps I am missing something.

No one is talking about "nationalization" when a judge rules that Microsoft has to unbundle their net browser. It is just a simple regulation of a monopolistic business practice. And I don't think it hurts anyone, except Microsoft.

I don't begrudge Microsoft their large market share, or even their attempts to take advantage of their monopoly. But what bothers me is the disingenuous whining and moaning about how unfair and backward the government is when they enforce the laws that have been created to prevent just the kind of scam that they are attempting.

To bring this back to the theatre of evolution, I think that what the government is doing is precisely what is needed to keep us out of the evolutionary rut that exclusive dominance of everything in the software business by Microsoft would drive us into.

j_doyne_farmer's picture
J. Doyne Farmer
Professor of Mathematics, Oxford University; Director, Complexity Economics, Institute for New Economic Thinking, Oxford Martin School; Co-founder, Prediction Company

The question is how to achieve this standardization. To what extent and in what manner should it be regulated? Here we are necessarily thrust into politics, as government is the mechanism that society normally uses for regulation. Regulation is necessary in situations where the incentive mechanism of capitalism fails to do its job. Global warming provides a good example. There is no incentive for a corporation to address the problem of global warming on its own. Action in this area will necessarily have to be driven by governments.

Arguing for government regulation is by no means arguing against evolution. Governments have evolved through precisely the same cultural evolution mechanisms that have created everything else in our society, from financial markets to computer software companies. Governments may be imperfect and inefficient, but they play a very important role. I'm sure we all have opinions about how government could work better than it does. But I doubt that any of us want to argue that we want it to go away completely. There are many historical examples where too much government has caused problems, but there are also many examples where too little has caused problems. This is all to say that arguing for a larger role of the government is not per se in any sense to argue against evolution. It is a matter of judgement to say what level of government will produce the best results in the future of evolution of culture in any particular situation.

Nor would I in any way suggest that we are at some kind of evolutionary pinnacle. This is silly. It has nothing to do with the point I am trying to make.

The anti-trust laws were developed precisely to deal with a situation where capitalism breaks down. When one company gains exclusive control over any given good or service the mechanisms of competition that normally regulate capitalism are removed. In this regard the problems with monopolies in oil or matches are very similar to the problem of a monopoly in computer software. Only as George Dyson has pointed out, the stakes are even higher.

There are many examples, such as telecommunications, where the government plays a an important, and I believe necessary, regulatory role. My wife, Letty Belin, happens to run the division of the New Mexico Attorney General's office that regulates our telephone and utility companies. So I have heard some horror stories about the things these companies would do to the consumers were it not for the regulatory oversight of the government. Again, these companies are singled out for regulation by the government because they are monopolies.

As I mentioned before, there is a fundamental difference between cultural and biological evolution: Where the innovations in biological systems are generated by random variation, in cultural systems they are based on designs originating from groups of human beings. On thinking about it some more, however, I don't think this is the main point.

Perhaps more important, in nature there is no global entity whose survival is enhanced every time a DNA molecule replicates. I don't disagree that in some analogy cells can be viewed as profit making entities. But they don't have to give energy to "MicroWet" to reproduce themselves. The code that they use to replicate is, so to speak, in the public domain. All organisms use it, they own it themselves, and no species can lay any special claim to it.

So, in response to the issue at hand, I am stating a political opinion. I see the need for standards. They will happen no matter what we do. But because of the dangers and known problems with exclusive ownership of standards, this is an area where I want my government to perform the regulatory function that it has evolved to perform. I don't doubt that they will do this in a less than optimal manner. But I think this is a situation where imperfection is better than nothing. I don't know what the right approach to this particular problem is; my guess that the right answer is not to break the monopoly up, but rather to regulate it, as we already do for telecommunications and utility companies.

In the long run, I think we all agree that we do not want the internet to get stuck in a local maximum. The question is how to avoid this. My opinion is that giving exclusive control of the replicating machinery to a profit making entity without any regulation would have a very negative result.