X-Message-Number: 26000 Date: Mon, 11 Apr 2005 18:26:14 -0400 From: Thomas Donaldson <> Subject: To Yvan Bozzonetti, re "uploading" THE GARDEN IN THE MACHINE: THE EMERGING SCIENCE OF ARTIFICIAL LIFE Claus Emmeche, trans. S. Sampson, Princeton U. Press, Princeton, NJ, 1994 This book tells about the idea of artificial life and just what it may mean. What is artificial life? It's basically life implemented inside one or several (parallel) computers. I personally have doubts about just how useful such a broad definition of life may be; the critical point about life is that it acts in the real world, not a simulated world. It's not even that I doubt the possibility of computer life forms: a computer virus is a primitive life form living in REAL computers. Regardless, Emmeche defines "life" so broadly that simulated life "living" in a simulated world counts in his mind as life. The problems raised by the simulated worlds aren't trivial at all: if we wish to draw conclusions about the actual behavior and evolution of life forms, having them only live in a simulated world omits almost all factors bearing on behavior and evolution of a real life form. Some cryonicists seem to believe in the merits of this kind of artificial life, and actually hope to someday become such a creature in a suitable artificial world. It's not that anyone thinks such creatures or worlds will SOON become real, but with cryonics we don't need our advances to come soon. We can wait in suspension until they occur. Yet to surround yourself completely with an artificial world remains no more than a way to spend the rest of your life gazing at your navel. Still, as a means to learn about real life forms computer simulations have lots of usefulness. Biologists and biochemists have run into one problem: the behavior of life forms, and even the behavior of biochemicals inside our body, works in a far more complex way than, say, gravitation in our Solar System, or even electrical circuits. We can usually easily decribe behavior due to mass, force, and gravity. Yet even our brains use not merely one transmission method (electrical currents) but a wide variety of chemical transmitters, plus electricity too. Computer simulations help us understand such complex systems. The early part of Emmeche's book gives a good account of this work, with discussions of all of the major work done here. Again, if we want not our artificial life not in an artificial setting but in the real world, he tells of work aimed at making computer systems able to do the kinds of things we and other animals do without much thought: walk, pick things up, respond to what we may see. It turns out that here we have some quite different researchers, and in a very broad sense, more imitation of biology: neural networks, evolutionary computing, genetic networks. It's a truism now that work in artificial intelligence has found tasks such as classification and calculation very easy, but making a system with which a robot could walk on a real path in the real world turned out very hard. (Since Emmeche first published his book in 1991 in Denmark, this problem has seen some serious progress, with 4-legged robots able to walk without falling over). Many other such problems await solution. Emmeche himself repeats a quotation: "The dumbest smart thing you can do is to stay alive". He's also clearly aware of the distinction between artificial life in an artificial world and life in the real world. He tells of two ideas of computation (which for at least for artificial life equals computation): one consists of the modification and movement of physical signs only, with its meaning given not by the computation but by someone outside it entirely. The other notion involves processes and change inherent in a physical system. As adult human beings we're normally immersed in symbols, and see even the real world in terms of those symbols. Yet to say that a tree can walk doesn't allow it to walk; many mammals and birds deal with the world without using language at all, and do so with some intelligence. It is this feature of our brains that gives MEANING to all the symbols we use. It also forces us to deal with unexpected events, both minor and major. In the latter part of his book, Emmeche devotes several chapters to discussions of various critiques of artificial life. The strongest argument came from a Peter Cariani, who points out that a consequence of living in the real world, whether a system is artificial or not. The real world never fits neatly and entirely into our symbolic, linguistic account of it. Even a robot endowed with a computer as brain can control its computations but not what it sees or the results of what it does in the real world. In one sense, Cariani's criticism should be quite obvious (but apparently may not be, at least for some). If you can ignore the broad definition of life with which he begins his book, Emmeche has written an interesting book on two subjects: the use of computer simulations of life to give us insights into how living things work, and the methods computer scientists turned to when ordinary digital programs and computers failed to help make robots able to work in the real world. There is a garden in our computers, but only an artificial one, just like everything else in our computers. It still remains beautiful, but not a place in which to live. Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=26000