This essay was originally posted at http://www.kud-fp.si/~karlo/intview/uslanda.html, and has now been archived for the Manuel DeLanda Annotated Bibliography. Please note that some links may thus no longer work.
by Karlo Pirc
Manuel De Landa : I began as a filmmaker. When I was 22 years old, I moved to New York to study film, and that was 20 years ago. Then I was an independent film maker for 5 or 6 years, and I made several films that were shown all over the country and in Europe. In 1980 I bought my first computer, taught myself programming and created a 3D animation system for IBM PC, which I sold out later to a company. That's how I got into computer graphics, which is what I do for living now. As a filmmaker in New York, you are supposed to be very informed about theory, particularly theory of semiotics. So, from the seventies I was involved in reading philosophy and I published a few essays on film theory. Only when I got into computers, I began studying philosophy seriously for 5 or 6 years. I'm an auto-didact; I taught myself all these things.
In 1986 I began researching my book on warfare (War in the age of intelligent machines) and finished writing in at the end of 1989 and finally my book came out in 1992. Then in 1990 I began researching my new book, which is on history of economics from different points of view. My first book was written from a point of view of a robot, and now my history of economics is written from various points of view - one chapter is written from a point of view of microorganisms, germs and viruses; another chapter is written from the point of view of rocks and mountains. It doesn't seem to make sense, but it actually works. For instance, rocks tell the history of humanity - they only care about our mineral part, our buildings and our roads - so it's urban history and economy of the cities. And microorganisms - all they care about is the great plagues and diseases. When Europeans came to America, they brought all these diseases that killed the Indians - how has that affected economics? That book will be out some time in '95. So now I'm a computer artist to make a living and fund my research, but my real passion is to be a philosopher.
Q : You are lecturing about a shift of paradigm in the postmodern world - the old, analytic way is replaced by new, synthetic one.
A : Let's take an ecosystem - like a jungle for example. When you try to approach it, when you study an ecosystem, using top-down approach, you start with an ecosystem as a whole and then you begin dissecting it until you get to the final units, which are the animals and the plants. That is the method that science has used for 400 years now and it's called analysis or top-down analysis. The problem is that many of the properties of an ecosystem rise from the interaction between the animals - for instance the interaction between the predators and the preys, the parasites and hosts or between simbiots. When you dissect things and take them apart, the first thing you lose is these interactions. You reach the final units by dissecting things, but then at the end you end up with units that are separated from each other. In an ecosystem, society or any other system, many of the properties are what is called synergetic or synergistic properties, that are more than the sum of the parts. But when you do analysis, you end up with a bunch of units and then you want to add them up - everything that was more than the sum gets lost - almost by definition.
So, to complement analysis, we need synthesis, and that's what artificial life does. In artificial life, you do not analyze an ecosystem, you synthesize it. I we begin with several populations of virtual animals inside a virtual environment and set them to interact with each other, the synergistic properties of an ecosystem emerge from those interactions. So instead of using top-down, starting at the top of the whole ecosystem and working your way down to the animals and the plants, you start with the animals and the plants - at the bottom - and work your way up. The advantage is that you do not lose the properties of interactions because you created these virtual animals and put them together to interact with each other. So, an ecosystem should emerge from those interactions.
Another example would be a flock of birds or an insect colony. In an insect colony, the whole colony has a kind of swarm intelligence. The colony as a whole is kind of like a computer. One little ant finds food and then the others follow him as if the whole colony was an intelligent being. Or if you have a flock of birds - there are a few rules when flying ; keep the same speed as the bird next to you, if you're too close get farther away and if you're too far away, get closer. With those few rules - as long as you put enough birds together - flock behavior emerges. And the whole flock has a kind of gracefulness of its own. That is more than the sum of its parts, it's more than the sum of the birds.
Q : What you're saying is that it's possible to apply such approach to cars and traffic.
A : Yes, if we gave enough artificial intelligence to the cars. That's what I was saying in my lecture (at the ARS Electronica symposium) two days ago. There are two ways of applying artificial intelligence to traffic. One is the top-down, centralized, in which you take an artificial intelligence and put it in a center for the direction of traffic. So the cars remain stupid - just cars - and whole intelligence is in center, where the traffic lights are switched on and off. That's the centralized, top-down approach. Decentralized approach, which is bottom-up, would be to make the cars as if they were ants. Just as an ant leaves hormones or pheromones, which attract other ants to follow him, the cars should be able to attract or repel one another. So when a car is in a big traffic bottleneck, it begins non-attracting other cars - so the cars do not go into the bottleneck. Similarly, when there's a big traffic jam and a car finds a way out of the jam, it begins attracting other cars to follow it. So that way, the traffic could have that kind of swarm intelligence that emerges from the interactions - instead of being imposed on the traffic by the traffic center.
Q : But our traveling is deterministic - we travel from one predefined point to another. So you cannot just let the car do the thinking, it can just choose the path to a pre-defined point.
A : That's the idea exactly. Of course you should be able to turn this intelligence on or off so that you can also make some decisions. So the car, instead of taking over, should just be an adviser. Instead of automatically following other cars - which could be quite disconcerting if you're going your way and suddenly the car begins driving in a completely different direction - it should just advise you. It could say : "We're approaching a bottleneck and I just detected a car that found a way out. Let's follow him." And ultimately, you should be the one to make the decision whether to stay in the bottleneck or follow the advice of your car. So the car should take over only when you're drunk or something and you cannot drive.
Q : This could also be done with a centralized network, with radio control, computerized maps and head mounted displays for driving. A mainframe can crunch the traffic data and compute the optimal route for the car.
A : Yes, but the problem there is that if computer analyses - then it's going to dissect the traffic. And all emergent properties are going to get lost there (in the process). The traffic flow of cars has some emergent properties. Certain patterns tend to form, like platoons of cars, small groups of cars that are bunched up together, with long distances between the groups. And some patterns are more efficient than others. If you let the computer analyze, analysis is only good when everything is clear - but what if there are traffic accidents, pedestrians that are trying to cross the roads and don't care (about traffic rules), all the noise and friction of everyday life? Analysis is very bad in considering all this noise, bottlenecks, friction and other problems - but it is very good when you're talking about logic, when a system is very logical.
Q : What about other simulations?
A : Simulations can be top-down or bottom up. Take for example economics. Economic analysts all over the world do simulations to try to predict what is going to happen in the market. Most simulations that exist are top-down. That's because human - or at least western society - has learned to think that way only. It's very hard for us to think bottom-up. The top-down simulations begin with things like for example overall rate of unemployment, overall rate of inflation or GNP, big numbers that apply to the whole country, and then they begin analyzing things until they get to factories and people and so on. But again, once you dissect the things, all the interactions between the factories, companies and so on, are lost. At MIT, they have been developing bottom-up economic simulations, which are much more realistic. Precisely, they begin with little agents, that have enough artificial intelligence with them to make transactions, they have little businesses - so there is a whole world of virtual companies and businesses. There are of course also virtual governments and unions of workers and so on.
Q : Something like SimCity...
A : Yes, something like SimCity. I do not know to what extent is SimCity accurate (compared to MIT project), but I think that it's also bottom-up. It's a good example. Of course, at MIT things are more complex than SimCity, but what matters here is concept. And the concept is that you get much more realism in simulations if you start bottom-up.
Q : Is the idea of intelligent cars still just a concept or are there any research projects dealing with it?
A : It's still just an idea, but there is a Belgian researcher Denebourg, who is working on it. The artificial intelligence for traffic in USA right now is all top-down, it is what is called an expert system. All of expertise from the traffic managers has been extracted from them through interviews and then put into central computer, which then advises traffic managers what decisions to make. But it is centralized, as opposed to a decentralized and anarchic kind of world like the Internet, which grows by itself without any central director - it's more like a bureaucracy.
Q : You are talking about "modern science", which uses synthesis in reasoning and bottom-up approach in making computer simulations. How can things as complex systems and artificial life be applied to social sciences and humanities?
A : My first book (War in the age of intelligent machines) was one of the first ones that began to apply chaos science to society - in this case to the military. I see war as a kind of turbulent, chaotic state of humanity.
The first application to social sciences was done by a physicist Arthur Iberall in the 1970s. The basic idea is very interesting - we tend to think of the history of humanity as if we were climbing a ladder of progress. First there were the hunter gatherers and then we invented agriculture and we stepped up, then we invented the state in Mesopotamia and Egypt and climbed up, then the state became democratic and we kept going up this ladder of progress.
What these new ideas suggest is that we are actually not going up any ladder, we are just going horizontally, exploring different zones. This new science imagines humanity as if it was in different states (gas, liquid and solid), like water. So when hunter gatherers existed - we existed in that state for a long time - we were like a gas, like vapor. Small groups of hunter gatherers - like 50 per group - existed and they did not really interact much with each other. They lived about 75 kilometers from each other - humans walk 25 kilometers a day - so even if they walked 25 kilometers, they would still be far away from each other. They would have to walk two days to reach each other. Hunter gatherer tribes did interact, but very seldomly.
So they were like a gas, but when we invented agriculture and settled down, we liquefied, became more like a liquid - like a pool of water standing there. And when the state was invented, we crystallized - everything became regular, like a kind of crystal. So pyramids began emerging and the state began regulating everything with rules and dividing everything into grids, like street grids and regular patterns.
Q : But what about increasing mobility?
A : You're right, it's never really just like that (just regular patterns). For instance, the state always has markets, and markets are about (symbolize) mobility. So there's always a co-existence between crystal and liquid. The main point is to see humanity as being able to exist in different states of organization. Just as a crystal is no better than a liquid and a liquid is no better than a gas, we are no better than hunter gatherers or agricultural villages because we are civilized - we are just different. So we are not going up the ladder of progress, leaving behind all these primitive states, we have just been changing state, as if humanity was like a substance. And that's important because that means that we should not dismiss what today Pygmies or hunter gatherers do, because they probably have wisdom that we have lost. And what we need to do now is to liquefy a little, because we have become too solidified, too crystallized. Maybe we need humanity to build a more liquefied, flexible system of organization. And that would not be going back down the ladder, we would simply go sideways to another area.
So one of the first applications (of bottom-up approach) to social sciences is to begin to see humanity not in terms of stages of development, but as phases - in gas, liquid or solid phase. From that, other developments have followed. The first social science that this was applied to, was economics - the first bottom-up models were economic. Now there are new applications to management science. Right now, every manager is top-down, they create plans for the company at the top and then they impose them on the workers at the bottom. The whole (new) idea would be now that the creativity, the initiative, would emerge from the bottom, and the role of the managers at the top would be to make sure that self-organizing teams of workers are created at the bottom, a sort of orchestrating (teams).
Q : I'm not really sure that this idea would work. I come from a former socialist country, where similar experiments were made and they failed miserably.
A : Yes, but the problem with socialism was that it was also top-down because the government told everybody else how to do things. (My idea is different.) There is an area in Italy, between Bologna and Venice, I think they call it "third Italy" sometimes, where they took three very big textile companies, dismantled them and transformed them into a lot of small textile companies. I'm not sure if this is still going on, but many writers have written about it, because it was a very successful experiment. Instead of having few huge, large corporations, you have hundreds of little ones, and all of the important decisions are being made by each of the little ones. Of course they are networked together with computers. But again, it is not socialism, it's a new thing. The whole idea of socialism is that you can plan things, whereas here a whole network of firms sort of drifts together creatively and grows together, but in a creative drift - instead of : "we're going there and that's our goal for next two years, so let's meet that goal".
Q : That sounds a bit like anarchism (particularly ideas of P. Kropotkin, Russian philosopher of intellectual anarchy, also called the Prince of Anarchy).
A : I think it's more like an intermediate state, not necessarily an anarchism. Anarchism is a movement that started in 19th century and has certain ideas that I respect, but this is a new thing. It may sound like anarchism, but it is not. The whole idea is not to have a bunch of little firms, disconnected from each other, but again the idea of liquid, solid and gas. A big corporation is too solid, too rigid. That's why IBM came late to personal computer. It was a small company like Apple that invented it. The creativity always comes from the small companies. And all that big companies do - because they have so much power - is appropriate those things. The idea is to take those crystal companies and liquefy them a little, put some flexibility in them. But it's not anarchism, because anarchism is everyone on his own. Here you have a network, so you can have some kind of swarm intelligence emerging from all these interactions between firms. That's what I think is new about it. Anarchism does not consider emergent properties, that are the key concept here.
Q : But first we need a shift of paradigm in human thinking. The only way for this to happen on global scale is in building a cyberspace of electronic (many to many) communications. But people still don't communicate enough by means of (mostly computer networking) technology to create a network of exchanging ideas in real time.
A : Yes, and that's why there's only a few areas in the world, where it works. The experiment in Italy was just that, an experiment. In USA, there are two areas, where computer industry grew - Silicon Walley, the Santa Clara Walley, and so called Route 128 in Boston, where all the minicomputer companies were, like DEC. Silicon Walley grew spontaneously, by this kind of bottom-up process and in Boston, the Route 128, was a little bit more top-down, (there are) big companies. And Boston is doing very bad now, Route 128 has lost its creativity, while Silicon Walley keeps growing, the creativity is still there.
So humanity has just begun to experiment with this (bottom-up approach), we don's trust it yet and still would want everything to be planned. Because plans give you the security - like : "if the government is planning this, they must know, what they're doing". But now we know - after the collapse of the Soviet Union - that if you give too much power to the government, if you centralize things, everything becomes crystal. So what we need is to liquefy society a little bit more.
Q : But what about the uprise of the consumer society? Many technologies today are used for debilitating instead of communicating. If our society becomes mediatized society, this won't be an easy task, because media are in the hands of governments and corporations and they are not eager to forget about consumer culture they're trying to create.
A : I agree with you, that is an obstacle. The consumer culture was invented by big corporations, not by small producers and businesses. The cola-wars are between Coca Cola and Pepsi Cola, the burger wars are between Burger King and McDonald's. And these are the huge corporations that are producing this consumer vision.
Q : And that's like programming (or planning)...
A : Yes, exactly, it is top-down programming, it's completely analytical. Every corporation has department of analysis, called "operations research department" or "linear analysis department". We need another option, to break with the past. It has been proved in computer simulations that "meshworks", as I call them, where there are a lot of people meshing together and they sort of interlock or complement each other, could compete successfully with corporations. Because a meshwork of producers could enjoy the same economies of scale like corporations.
Of course, there's not going to be one small company competing against corporation, but a whole swarm of small producers competing against a big corporation. I think they would have a chance and we need to give them a chance. It's going to be a long struggle because we're already too much a consumer society, we're too dominated by big corporations. This is a process that could take 50 or 100 years. I think it could help society if we put the swarms of small producers to compete with big corporations and make them obsolete. Our grandchildren or grand-grandchildren could live in a non-consumer society, but more of a thinking society.
Q : Japanese zaibatsus (big corporations) are doing just that (creating some kind of meshworks). They're incorporating many different small producers inside a large corporation, but they still use a central, top-down management.
A : You're right, the Japanese are an interesting case. They use combination of the two (principles). However, Japan is now having problems. I think Japan was more flexible 10 years ago. Now the corporations are becoming too top-down and the same thing that happened to USA is going to happen to them. Just as General Motors, Chrysler and Ford eventually became too rigid and they were not producing good cars, because their creativity was gone. All they could do is put fins on the cars, which are just cosmetic adjustments, but they were not able to respond with a fuel efficient car after the 1973 oil price rise. It was the Japanese, that were smaller, that were able to respond fast.
But now the Japanese are becoming too big and eventually they're going to sink under that weight. In other words, they're going to become fossilized just like the other ones. They're going to become too rigid. And there will be smaller countries, like Singapore, or perhaps South Korea, that are going to have the activity and they're going to replace Japan. That's my thinking. Although of course it's not simple as that, it's a struggle. I wish that the meshworks of small producers could win and I think they have a chance. But big corporations already have a lead and it's very hard to say if that lead is too much for meshworks to overcome it. I am optimistic and I think we can do it.
Q : What about the "third world" and eastern Europe - do you see the possibility of creating meshworks in these areas (and environments)?
A : I think so. In the past 1000 years, it has always been the third world of the time where this meshworks existed. For instance, about 900 years ago, when Venice was still a part of Bysantium, of the east Roman empire, and Constantinopolis was the center of the world, Venice was the third world. Meshworks of small producers existed in Venice at that time. Then Venice with Florence, Milan and Genova became the center of European economy and then it was London, that was the third world for Venice. Meshworks of producers existed in London and they began competing successfully with big firms in Florence or Venice. Then London became the center, where the big companies were. Then it was New York, Boston and Philadelphia that were the third world for Britain, where the meshworks of small producers were.
So the answer is yes, I think that in fact the third world has a better chance at doing this if they are smart enough. I mean if their governments are smart enough to see this, because the problem with the third world is that they are just trying to copy the first world. But they are never going to catch up that way because they are way behind. They should start paying attention to history to see what has happened in the past.
Q : So it's like the race between the Archimedes and the turtle now.
A : Yes, exactly.
Q : Another important aspect is communication. Instruments of mass communication are in the hands of large corporations and governments and that is very important for further development of situation. (However, there is an alternative media vision - Internet.)
A : Yes, but right now Internet does not have any big businesses in it. And that (Internet) is what we would consider the most futuristic form of telecommunications. Of course, big businesses like Time-Warner, all big cable and telephone companies are about to come in, but I have the feeling that enough of the grassroots community has formed on the Internet that would be able to resist this (possibility of corporate takeover). The Internet could become one of those meshworks, that I am talking about - an international meshwork. If we manage to do it right, everyone who owns a computer could actually become a small producer of information. You would be able to provide information services from your home, without being a big corporation.
We are now at a threshold, (so far) the Internet was able to flourish as a meshwork, but now the big, top-down companies are about to come in and try to transform it into a highway - the information superhighway. The difference between a meshwork and a highway is that a highway is a one lying (fixed) thing. Howard Rheingold said that important thing about the Internet, as opposed to telephone or TV, is that telephone is a one-to-one communication (medium), TV is one or few-to-many, whereas the Internet is many-to-many medium. So that's another example of top-down versus bottom-up idea. The Internet is completely bottom-up; it grew by itself from small nodes all over the world. Now the big corporations want to transform it into one-to-many medium. And that's idea of highways, (predefined) network of roads, as opposed to a meshwork of interconnections.
But I think that Internet has grown enough and will be able to resist that - so there will be a competition. There will be parts of the net that will be crystallized and other parts that will remain liquid. The big crystals are going to try to come in, but the people that savored the freedom of the net are not going to become (just) consumers again. The know that they can be producers too.
Q : But isn't that just the story of television, repeating itself in new environment - the digital media? They said just the same about the cable TV (although it was just a liberal dogma), that everybody will have his own channel...
A : I agree with you, but the cable TV companies said that they would have a few channels, where people could come in and do their own programs. That was just a little bit of freedom - it was a sort of like a big crystal, with a little portion of liquid. The new idea is that the liquid is as big as the crystal. And television was never like that, whereas the Internet is in a liquid, dynamic state, and it's big and huge. So when big crystal corporations come in, they are going to have to fight against a pool of liquid - and it's going to be different.
Q : A major obstacle in popular TV development was that only few - the corporations - had the expensive equipment to broadcast and enough money to create an attractive program. And if they build multi-media super highways, how can the small producers obtain necessary technology and enough money for program - to compete with media giants? Many small producers will not even want or try to do that.
A : I'm not saying that this is a magic solution for our problems - I do not think there are any magic solutions. Humanity is in a mess and it is very hard to think about how to get out of this mess. Ecological and populational problems, racism and sexism - we have too many problems. Unlike Howard Rheingold, who think that Internet is the big solution for everything, I do not believe that. I do not believe there is a big solution.
But it can open up spaces, where we can experiment with new solutions. And perhaps two or three generations down the line will be able to solve our problems. I certainly do not think that these problems will be solved in this generation just because the Net is there. But at least we can get started, learn new tricks and experiment with new solutions - and then pass that knowledge to next generation, and if it's smart enough - or revolutionary and willing enough - it will put together the next pieces of the puzzle and continue the struggle. And perhaps two or three generations down the line - the humanity will be able to self-organize, live without centralized governments and corporations.
Q : There is another interesting aspect of complex systems and artificial life, the self-organizing processes - by exploring them, we can get to the origin of our thinking and then perhaps use that knowledge to improve our culture, for example linguistics. What do you think about such possibilities?
A : One thing that I wrote in my article on virtual environments is that all linguistics that now exists, is top-down. Chomsky or Saussure's linguistics - they all start on the top and say that language is a system, but they forgot that language is in fact a fluid thing. They began by thinking that language has a synchronic structure, a mental structure like grammar, that does not change. But if you examine languages in history, (you will find out that) they are constantly changing. Five hundred years ago, language existed in much more fluid state. And then - all of a sudden - standard languages began appearing. The language of Toscana (in Italy) became standard Italian, the language of Castille became standard Spanish, the language of Paris became standard French and so on. If you could - 500 years ago - take a picture of the languages of Italy or the languages of Britain - they were languages in continuous variation, because everywhere you'd go, there would be different accents, slightly different grammars. Languages were continuously changing.
Then governments got paranoid because of that changes and begun trying to fix it - they wanted to solidify the language. And they in fact succeeded to a certain extent. Standard English and standard Italian has changed very little in last 200 years. In Italy, you can still go to different regions and find different dialects, but dominant dialect now is the dialect of Florence (that region). Language existed in a liquid state and governments tried to solidify it. And then linguists came in 19th and 20th century. Saussure learned at school standard French and Chomsky learned standard English - and then they mistook the artificial solidity of these artificial languages - the standard English and standard French - for language itself. That's why they saw it as an eternal thing that didn't change - because governments 200 or 300 years before had artificially solidified it.
So the new linguistics would have to be bottom-up. They would have to start with a population of virtual human beings. In an artificial life experiment you start with a population of virtual animals in which two kinds of flow occur, a flow of genes through generations of animals and a flow of flesh, meat, so to speak, in food chains. One animal eats other animals and is eaten by another one, so the fleshy stuff is also flowing. To do experimental linguistics or bottom-up linguistics, we would have to add two more flows to the flow of food and the genes. The first one is flow of memes, patterns of behavior that transmit through imitation, as bird songs that birds acquire from each other; and of course we also have memes, like fashion - whenever we imitate one another. The other one is a flow of norms, which are patterns that are not transmitted by imitation but by enforced repetition, for instance the sounds of a language. Everybody expects you to pronounce things right - I'm not imitating the sounds of English, I'm repeating them. However, anything that reproduces, like genes, memes or norms, can evolve.
The whole idea is to put a population of virtual human beings - very simple ones - they wouldn't have to be spectacularly intelligent and be able to enjoy literature, they can be simple human beings, a kind of peasant human beings with simple way of life and routine type of everyday living. And then you can have these flows flowing through them. When English was born, there were the Anglo-Saxons, which were Germanic tribe, that has invaded England, and they brought with them Germanic dialects - and of course you can add some Scandinavian elements because the Vikings also tried to invade England. Basically you would have a soup of Germanic languages. Then the Normans invaded them and they spoke French; they conquered England - this was about 1000 years ago - and they imposed French as language of aristocracy. The peasants, who were the only ones that still spoke the soup of Germanic languages, trying to resist this kind of linguistic imperialism, in about 200 years - from year 900 to 1100, roughly - created English. Of course it wasn't any of the peasants that created it, it was all of them together. There was a flow of norms through them, through several generations of peasants - in 200 years you have about 10 generations - so there were ten generations of peasants, passing the norms to their kids and so on. They transformed that soup of Germanic languages into something that we would recognize today as English, old English to be precise. So when the Normans left, the aristocracy went back to English - they adopted this language from the masses. It was in the masses that English created itself. And they were peasants, with very routine way of life, not like urban kind of people.
They would be easy to model in a computer, because a routine day-to-day living can be (easily) simulated. But you would have to have a flow of memes and norms, flowing through them, and then some way of representing the pressure from above, like aristocracy speaking French and seeing English as the lower class language. That would be bottom-up linguistics. It does not exist yet, I do not know of anybody who is doing it. I write about it, but it's more a philosophical idea right now, but I have a feeling that it will exist in the future. And we will look back at our current linguistics and see it as a very primitive attempt.