I work on developing an understanding of biological complexity and how we can create it, because the limits of software engineering have been clear now for two decades. The biggest programs anyone can build are about ten million lines of code. A real biological object a creature, an ecosystem, a brain is something with the same complexity as ten billion lines of code. And how do we get there?
SOFTWARE IS A CULTURAL SOLVENT
A Talk with Jordan Pollack
The New York Times, in a front page article on August 31st ("Scientists Report They Have Made Robot That Makes Its Own Robots" By Kenneth Chang) reported on the work of Jordan Pollack and his Brandeis University colleague Hod Lipson: "For the first time", The Times reported, "computer scientists have created a robot that designs and builds other robots, almost entirely without human help."
Pollack notes that he "works on this question of self-organization, using evolution, neural networks, games, problem solving, and robotics. And the way that we work on it is by trying to set up non-equilibrium chemical reactions in software which dissipate computer time a form of energy and create structure. Some of that structure we can actually make real in the form of robots, and although robots are much more exciting to cameras and the media than problem-solvers, games and language learning, our fundamental work is in trying to understand where complexity itself comes from, without a designer."
He sees "a merger of bio-informatics, biotechnology, and information processing. As we understand cellular processes, neural representations, and develop microelectronic and nanoscale technologies, our artifacts will be able to interact with our biological forms at a most fundamental level. Unfortunately, we really haven't fathomed the complexity of nature yet to know what to do with it."
JORDAN POLLACK, is a computer science and complex systems professor at Brandeis University. His laboratory's work on AI, Artificial Life, Neural Networks, Evolution, Dynamical Systems, Games, Robotics, Machine Learning, and Educational Technology has been reported on by the New York Times, Time, Science, NPR, Slashdot.org and many other media sources worldwide. Jordan is a prolific inventor, advises several startup companies and incubators, and in his spare time runs Thin Mail, an Internet based service designed to increase the usefulness of wireless email.
IS A CULTURAL SOLVENT
JORDAN POLLACK: It's a marvelous age that we live in right now: The age right before convergence with mechanism, where ultimately we will be wearing our computers as part of our bodies. People are talking about internet, television and telephone convergence to personal wearable devices. But we are also within a century of the merger of bio-informatics, biotechnology, and information processing. As we understand cellular processes, neural representations, and develop microelectronic and nanoscale technologies, our artifacts will be able to interact with our biological forms at a most fundamental level. Unfortunately, we really haven't fathomed the complexity of nature yet to know what to do with it.
on developing an understanding of biological complexity and how we can
create it, because the limits of software engineering have been clear
now for two decades. The biggest programs anyone can build are about
ten million lines of code. A real biological object a creature,
an ecosystem, a brain is something with the same complexity as
ten billion lines of code. And how do we get there? My lab works on
this question of self-organization, using evolution, neural networks,
games, problem solving, and robotics. And the way that we work on it
is by trying to set up non-equilibrium chemical reactions in software
which dissipate computer time a form of energy and create
structure. Some of that structure we can actually make real in the form
of robots, and although robots are much more exciting to cameras and
the media than problem-solvers, games and language learning, our fundamental
work is in trying to understand where complexity itself comes from,
without a designer.
My definition of a robot is a computer program hooked to a real piece of hardware, working 24 hours a day and justifying the investment in its own creation. It may even put people out of work. In the case of ink jet printers it's scriveners, and in the case of ATM machines it's bank tellers. So there is some disruption coming, but the wearable/wireless upheaval is happening before the robotics upheaval!
But I've also been involved since 1976 with microcomputers, the very small computers that are now everywhere; they're inside your cameras, and they're inside tape recorders. Apple's new mouse has a super-computer in it, apparently! We're coming to the age of wearable computers in the age of wireless, and I've been waiting and watching that for a long time, since I fashioned a one-hand keyboard in 1985. I think the new devices like the Blackberry and the PDA/Phones are really the beginning of wearable computers. You've seen people with cellphone buds in their ears all day long. These wearable computers are not what the pundits and nerds said that they would be, but people are carrying them around all day on their belts, using them all the time, and they will evolve into something approximating the science fiction communicator, voice together with videos, MP3's, fax, and e mail. It will all be something that you live with all day long, and we'll become untethered as a society.
EDGE: You're talking about getting untethered, but to some people it might sound like an electronic leash.
POLLACK: What's happening inexorably is that even though supposedly the world is smaller, our social relationship networks are forming wider and wider nets. I have people in Atlanta, Washington, California and New York all over the world in fact who I communicate with, and I become more mobile and I travel more, and at the same time the number of email messages a day that I process has moved up from ten a day to a hundred a day. And in a couple of years it's going to be 300 a day! We see an opportunity for intelligence here, as the email message count and the cell phone message count rise.
For example, I've designed an adaptive e-mail filter. I can tell it that I only want to see 50 messages a day of these 300, but the other 250 I want to know about still. I don't want any machine to throw stuff away unless I've already said I don't want any more mail from this vendor. But I do want the top 50 in order of my priorities. And those are adaptive and change on a daily basis. Who did I recently send an e-mail to? Well that's somebody whose response I want to see. If I get e-mail from the President of the United States, he might not be on my "in list", but I don't want to miss that mail.
These are fairly simple AI techniques that can essentially take those 300 messages a day and dynamically pick out the ones you're most likely to be interested in, while not throwing away anything that you might even have a relevant chance of being interested in.
We are inexorably getting more mail, we are more mobile, and the demand for a wearable communication device that can essentially intelligently deal with all your communication becomes more and more relevant. That's why the Blackberry is so popular now, it's really the first popular wearable computer since it fairly seamlessly replicates the Microsoft Outlook view on your desktop into something that an executive can carry around. In a few years we'll have eye pieces that'll give you full color view of your desktop and a little wireless computer device that you'll wear. That's really where it's going, and I'm excited to be involved in it.
EDGE: The work that you're most famous for on self-replicating robots was recently reported on the front page of The New York Times. This naturally will attract all those people concerned with the human condition. You must have thought about this. In what way is the development of your technology changing our ideas of ourselves?
POLLACK: We can look at the problem of robotics from an industrial point of view, from an economic point of view. Say that we could build you a vacuum cleaner robot and it would cost five million dollars to develop, and each vacuum would cost $5,000. But you can just buy a plain old vacuum for a hundred dollars and push it yourself, or have someone who works for eight dollars an hour push it. There's no money to be made in robots; there's no mass market that would justify the kind of development necessary to have a robot. Until you actually get a general purpose humanoid robot that's cheap enough to hire instead of a human, there's a huge chasm to cross.
Everyone keeps thinking that the robots of science fiction are the robots they want. I say the real robots are ATMs and ink jet printers. They're really robots and they're really here now. And asking them to obey Asimov's laws does not compute. They have a mass distribution, and they justify their own existence. We have an ice cream vending machine at my university. That robot probably costs several thousand dollars, but it earns its own living selling ice cream pops! Humanoid robots might meet our expectations of Science Fiction and ancient dreams of ethical slavery, but it is still several hundred years off.
What my lab is doing to answer your actual question about the human condition is we're saying the thing that we need to do is to take the human engineers, the expensive human fixed-cost talent, out of the process of designing robots, so that we can make robots which are economical in small quantity. A robot design may only be needed in once, or three times, or in five copies. A robot helping in a manufacturing production run may last only 6 months, not enough time to amortize a big investment in engineering.
Only when we have robots which approach the cost of the materials that are needed to make them will they be economicallyvalid without mass production. Our fully automatic design and automatic manufacturing approach is not "self-replicating", but it uses software to design a machine in virtual reality to fulfill a particular purpose, and then automatically builds that machine. It isn't scary, its cost-effective.
If you say I need ten humans earning a hundred thousand dollars a year for five years to do the design of a particular robot, say a horse or a tank, and only ten of those are going to be sold, well then each of those is going to have to sell for ten million dollars in order to justify the expense of the human design. That's really where the economic problem is in robotics. By doing automatic design, by having software which actually invents, we're beginning to broach the issue that, computer chess players, have broached a long time ago. If playing chess is really a human thing and then a computer starts to do it, what does that tell us about humanity?
We think of creative work as a purely human thing, and wrap invention in mystery and legal monopoly of copyrights and patents. I've written a lot about information as property, actually, because software is the most important recent development in human culture, and its been misunderstood as a box containing just a book and a compact disc. Software is a cultural solvent.
EDGE: The astrophysicist Martin Rees argues that 21st century-technology will threaten humanity far more seriously than any natural hazards. One of his concerns is that we could be wiped out by rogue nano-machines that replicate catastrophically. Any comment?
I agree with Rees and with Bill Joy that there are risks, especially when people are playing with artificial chemistries, genetic engineered species, and micro-weaponry. But I think the risk from pollution, disease, and war are much greater risks. Poverty and desperation of the dispossessed are triggers for catastrophes which could be ameliorated with games favoring the middle class.
But at the scale of electromechanics which we work in, out of control self replication is impractical. No robot will be able to eat an old fax machine to make babies. No robot will be in control of massive factories and natural resources for manufacturing themselves. Look as what's really here, not what's in science fiction books.
I do think that there is a very slight risk of networked server and routers becoming self-aware through aggregation of amazing amounts of CPU time. A phase-transition of a computer network to self-awareness is the terminator hypothesis, not self-reproducing robots with Swedish accents. We don't really understand how brains, composed of billions of neurons become coherent personalities, so we have no idea if a million servers on the internet could self-organize. I once proposed a SETI-like project to the Pentagon to watch for signs of artificial life arising in the communications and power infrastructures.
near term, I think the risk of Microsoft Exchange Visual Basic Viruses
crippling critical infrastructure is far more serious than Nanomachines
POLLACK: The notion of computation itself is something I've worked on. John Kolen and I wrote a paper in 1991 called "The Observer's Paradox" which essentially overturned the traditional notion of generative capacity that was brought to the fore by Chomsky and the whole cognitive science movement. The question is whether or not you can infer the power of a machine's internal computation by observing the way it operates. Earlier we thought we could analyse human language, and could say something about the human mind being context-free or context-sensitive. But when you actually look at a physical system, a dynamical system, the choices made by the observer affects the level of computational complexity that the observer actually finds. This means that, in a certain way, the Heisenberg principle actually applies to the traditional notion of computation as it arises naturally. In my view, computation is everywhere, and we could look at life itself as a process which involves information and energy and matter, working together and forming state structures in very complex ways. And we understand them best through our computational notions of process and representation.
EDGE: Last year I talked about similar ideas with Rod Brooks in his Edge feature ("The Deep Question"). Here are a couple of sentences:
"Computer scientist and AI researcher Rodney Brooks is puzzled that 'we've got all these biological metaphors that we're playing around with artificial immunology systems, building robots that appear lifelike but none of them come close to real biological systems in robustness and in performance. They look a little like it, but they're not really like biological systems.' Brooks worries that in looking at biological systems we are missing something that is already there that has always been there. To Brooks, this might be called 'the essence of life,' but he is talking about a biochemical phenomenon, not a metaphysical one." Any comments?
POLLACK: It's an organizational principle and a scale of complexity which dwarfs software engineering. Vitae is not an ingredient that anybody is really looking for. We work on a basic biological question in my lab, which is, "How can a system dissipate energy and create more and more informational structure? How can a computer program write itself simply by wasting computer time?" In some sense it's a computational thermodynamic approach to artificial life, as opposed to a purely software-engineering approach to life. When we get to 10B lines of code, we will know if we succeed or fail.
The traditional notion, the high church computation, the separation of the brain as hardware and the mind as software, is influential but it's ultimately wrong because the computational metaphor of serial programs operating on discrete data structures doesn't really capture the richness of natural systems. To return to your earlier question, our traditional Von Neumann notion is just not rich enough to capture what's going on in the natural world. I'm not saying that the world is Penroseian, post-Turing, or that there's magical soul-power there, just that we must drastically expand the notion of how information is represented beyond traditional symbolic computing which is based on flexible lists and logical forms. Semantics is more than syntax with Capital Letters.
We've been looking at chaotic systems, fractals, dynamical systems, both the attractors and the transients, all of which are very different than the traditional data structure plus algorithm that we've studied in computer science. I still think its computation; everything in nature is fundamentally the interaction between energy and matter and information. The question is: What is that interaction? What are the principles which guide the dynamic evolution of the system? It's not simply a program, some data structures and some polynomial time algorithms. While it's right to think of things in terms of computation, the ideas we've brought to it have been very impoverished compared to what's actually happening in natural systems, and that's really what I've been studying. What is God's cookbook that enables compositions of things to have new and surprising behaviors everywhere in the universe? Why does hydrogen and oxygen combine into something with a long liquid phase and such odd freezing properties? Random mixtures of organic chemicals are dense with behavioral possibilities, while random strings of machine code are 99.9% useless.
EDGE: On one level, software is a machine, a virtual machine. It has physicality. This is the software you buy shrink-wrapped off the shelf or download from the Internet onto your computer. Let's call this 'software machinery.' On a deeper level, our civilization has made a phase transition, the result of which is a new and as yet unnamed state. We might loosely call it software, but it is deeper, broader, more fundamental. In a physical phase transition from a solid, say, to a liquid to a gas molecular content remains the same but the relations among molecules change, as when water takes on a radically different form when it boils off into a gas. We've entered a new state of civilization, and software, the 'engine' of this phase transition, becomes its philosophical essence in the deepest ontological sense."
POLLACK: Software is the embodiment of design, and its been stripped out of the physical manifestion of design. This is going to change traditional industries. Imagine Ford Motor Co losing its factories and being an intellectual property company, developing and licensing their blueprints to the most cost-effective factories. The best car companies will have the most creative humans using the best CAD. Then, like the video game industry, groups of engineers will spin off into little startup car companies so they get a piece of equity in their own creativity.
We could say that traditional notions like the patent system are really places where you could study this transition most effectively. A wheel is a piece of hardware. It's this thing, and we've lathed it, and it's round, and it rolls and it carries a weight. But there's a piece of software which says that "for every i from 1 to 360 plot r and theta, being the angle," and in software you've created a wheel! Now traditionally the patent office has rejected algorithm as something that was of nature, something that was of God and therefore unpatentable, but over the past decade and a half or so, we've allowed the patenting of software. We've even allowed the patenting of a business idea. If the current situation applied 40 years ago when Amway was formed, it could have gotten a patent on the idea of hierarchical multi-level, pyramid scheme marketing, as opposed to the original Ponzi scheme of pyramid scheme investing.
Software is language, and we copyright it; it's all language that describes how a machine is going to work, all the way down to the lowest level of detail. And what a computer, a compiler, or an interpreter for programming languages does is to make our piece of text come alive, and the machine actually operates exactly the way the piece of text describes! So instead of just being a description of the points that form a circle, a computer can take that description and turn it into a virtual simulated wheel inside of a virtual environment. Software is a solvent melting the boundary between what is virtual and what is real, between text and invention.
Some of the excitement over my lab's conception of automatically designed robots was the idea that we made a transition from the virtual world back to the real. Software itself made inventions inside a computer that, in another age, could be patented. We could have patented a triangle, a ratchet, or a breast stroke before people were swimming that way, and then turned that from a software idea in a virtual world into a real piece of hardware that embodied that principle. The end of that boundary between text and invention is something that profoundly affects society and the academy.
I'm working in my university to alert the faculty to what software means to its traditional arrangement in which researchers keep their books while the university gets their patents. What does it actually mean now, when I could write a piece of text using an editor, which then becomes a patentable piece of software? Is it mine or is it the University's? We're all fighting about that, either for software or courseware. My proposal is to end the patent policy and tithe the university 9% of all our IP equity it incubated.
Not only has software destroyed the text/invention distinction, software destroys the boundaries of what we used to own versus rent. We used to buy a book and could own that book or we could own a recording. Property itself is being redefined in the information age.
The book was really three different things fused into one. One part was the words, the information content; the second part was the medium, the physical delivery mechanism, the paper and the ink that captured the words; and the third part was a social and legal contract, the license that said you may buy this book, and after you read it you could keep it in your library, or pass it to a friend, or sell it in a garage sale. But you can't make more copies to sell. Those three components, the media, the content, and the license, have been torn asunder in the information age.
When you think you're buying a piece of software from Microsoft, if you read the license carefully, what you find is that the license is a legal agreement that says you actually didn't buy this. You really have just given us money for a license to use, and here are the conditions of use. You may not sell it separate from your computer. You may not make two copies of it. You can't give a copy to your friend while you're not using it because it's part of your computer. And the actual content changes; it goes through release one, release two, release three. It goes through forced upgrades where you have to buy it again and again even though you've already bought it. And it comes on floppy disc, hard disc, zip drive, compact flash, it comes downloaded off the Internet.
And that breaking apart of an information property into its three essential components media, content and license has rendered it a very, very important moment in understanding the human condition. What we think of as property, as ownership, is about to change. Instead of being gold and jewels, property is just information about usage rights, which is transferred in a non-duplicative and non-lossy way. The only difference between a poor and rich man besides the quality of their dental care is what the bank computers and brokerage computers say!
We're now watching new licensing models arrive, like ESD (electronic software download) in which there is no disk. Like the ASP (the application service provider) where instead of actually buying a shrink-wrapped piece of software, a disc to put in your computer, you go to a Website, give them your credit card, and use some computer time served off the Website. We're seeing rental licenses like Divx, which only failed because the machines were overpriced. . We're seeing "read once" licenses, like what Bill Gibson had envisioned: You buy an "ebook," read it once, and it self-destructs. Jim Phelps's tape in Mission Impossible was an earlier harbinger of the read once license.
The fact that the three components of a book have been broken apart means that there are great opportunities for wealth, in terms of selling the same content over and over and over again, and there are also great opportunities for abuse. Both the Digital Rights Management and the File Sharing movements operate against property (as something you own until you sell it). I think the greatest threat to human condition is not my cheaper robots, but the end of property, when our books, records, videos, software can no longer be owned.
EDGE: Civilization has states, just as matter does. Why haven't we pushed this idea so far?
POLLACK: There is a very interesting thesis by Henry Plotkin, who tries to get Darwinism to operate at the level not just of individuals but of species, ecosystems, cultures and language. It is true that we can understand the states of culture and civilization as we understand dynamical systems.
We see the acceleration both of culture and the communication of cultural change. It's faster than evolution of species. Human language has developed in the blink of an eye from the point of view of the species. All of a sudden language evolved, and then human culture evolved much, much faster. Language and culture co-evolves, and the texts and connections which fuel this arms-race have jumped onto the internet.
Electronic Culture has been growing for 15 or 20 years now. When I talk with my parents and inlaws I tell them that the reason that they should go on the Internet is because all of human culture, every bit of human culture, every idea, every book, every thought, every hobby everything that we do, everything that people think about is now part of the Internet. Now there is somewhere in your house where you can find other people who engage even in the smallest niches of human interests.
What's interesting is because business, industry, and economic activity are parts of culture, now places in the third world that might have wanted to prevent people in their culture from being exposed to other ideas have no choice. If they want to participate in the world economy, then they essentially have to get their economy up on the Internet. And once they get up on the Internet, it's very, very difficult to stop ideas from coming in.
As we move into the age of being wirelessly connected to the internet, I definitely see a new age of civilization.
EDGE: Civilization isn't what it was; it's something quite different. One example: I have an office but I never have to be there. In fact, I am there less than half the time and I believe I am far more effective in this mode. Instead of sitting in one spot waiting something to happen, I'm out talking to people, making things happen. But at the same time while I am sitting in a hotel suite talking to you, I am setting up a video camera and sound, checking email, answering the phone, the door......I've never been so busy. A lot of the work normally done by the staff at my office I now handle myself.
POLLACK: Here's the question: Is this new economy something in which we're all just making more work for each other? Is the great boost in productivity in America just because we've out-psyched ourselves? People working in huge companies acting as if they're working in start-ups and working sixty hours a week instead of forty. And not getting time and half, but getting stock options. Everybody I know is stretched to the limit and missing their families. The opportunities are so great. If you don't take advantage of the opportunities, you're falling behind. Yet I don't think the fundamental opportunity of the technology change is going away anytime soon.
EDGE: What hit us? Drive by any billboard on the highway and you see www.something.com. Life has changed. We seem to be in the middle of a phase transition.
POLLACK: Yes! There are certain periods in history that we might call phase transitions, bifurcations, or critical points. With respect to evolution we discuss how there are certain elements that adapt to one another in some state, and somehow they come together and they create a new function. Its called an exaptation. Flying and seeing didn't just happen. There was no goal for something to fly, but there were some existing appendages to some animals that enabled them to jump further, or others that enabled them to balance in certain ways, and that ultimately became flying. The coming of the Internet, the ability for anyone with a computer at home to get a phone number for an ISP and get on the Internet, has led to the formation of very rapidly available markets. You could put up a website, and with ten million dollars of advertising, you can get tens of millions of people to start knowing and start coming to the website. There's never before been a time in history where a few guys in a garage can all of a sudden take credit cards from hundreds of thousands of people. That is really a new economic opportunity.
In November 1984 I was in Japan and bought a pocket computer from Casio. It had a full chicklet keyboard, a little screen, and a basic programming language inside, and it came with a printer and a cassette deck as peripherals. I was completely amazed. Here was the origin, the beginning of a whole new kind of computing with much smaller form that could do amazing things. And it's been 15 years that pocket computers remained a niche market! And little radio computers have been around for quite some time as pagers and stock quoters. Suddenly you make a pager look more like a pocket computer, and you have everybody in the world using e-mail, and all of a sudden it's a very valuable thing! Sometimes those things just come together; things grow independently and into each other and then you get a critical point, and that's what we really saw with the Internet when gopher and ftp retrieval got graphic buttons, and what we're going to see with wireless data and wearable computers.
EDGE: Where are things going to be in 2005?
POLLACK: They're really not going to be much different than they are now! I am quite a gadgeteer and am comfortable talking about specific kinds of gadgets, rather than about society and culture. For example, I think digital cameras have settled on 3 or 4 megapixel image; when these $700 cameras drop to $250, traditional photography will be toast like LP Records. In computers, we're going to see pretty much the same kind of laptops that we see now, just an incremental increase of power, with more ports built in. In Cellphones, we'll see better integrated PDA's and email systems, and might see some interesting wireless multimedia, especially as Japan and Europe get their investments in the third generation (called 3G) of higher bandwidth devices. But like telegraphy, the entire network will be held back by the slowest common denominator, and text email will be the morse code of this era. I think the telephone system will be radically changed by robotic VOIP technology and unified messaging. I expect disruptions like practical Microdisplays to enable pocket-sized laptops with 10 hour batteries. I think some ad hoc networks might spring up like CB radios for free instant messaging. And some filtering breakthroughs to get my email back under control!
EDGE: What about robotics?
POLLACK: Can I sell you a Lego Bridge? The robotics industry as it exists today caters expensive machinery to very high-profit industries whose production profits can justify such luxuries. I don't see much really changing before 2005 in robotics, but by 2010 I think that we can have some impact. With the right investment and patience, I can see how to create a general purpose robotics industry that can automatically design and manufacture simple machines for industry and entertainment. That's an inversion of the traditional idea, "Let's build a humanoid robot slave that can do everything!" A technology that can very cheaply produce specific dumbots for different kinds of tasks; assembly tasks, military tasks, cleanup tasks, entertainment tasks, even domestic tasks, might actually lead to a profitable and self sustaining industry and a change in culture back towards invention and manufacturing of real stuff instead of just dotcom stuff.
EDGE: You don't expect me to buy this, do you? We know what you're really up to.
POLLACK: No, actually, I'll give you a copy of the business plan. You mentioned earlier that you go to Aspen. Maybe you'll meet Arnold Schwarzenegger up there and he'll be interested!