Attacking Complexity.
I am re-reading a book I picked up by chance around 1994 while in high school -- Artificial Life, A Report from the Frontier Where Computers Meet Biology by Steven Levy. I was awestruck back then as a teenager by the possibilities revealed in the book on autonomous, self-organizing systems, complexity emerging out of simple rules and the potential of modeling complex systems in nature inside a computer using cellular automata. Some of the NASA research described in the book on self-replicating factories for space colonization was straight out of a science fiction novel.
A decade and a half later it is interesting to look at the promises presented in print around 1992, and how artificial life seems to have faded from hard science championed by mathematician John von Neumann into a recreational geek hobby following Conway's simplification of von Neumann's universal computers into binary cells of Game of Life. The more promising artifical life research today seems to happen on wetware, in terms of synthetic life, rather than in hardware and software.
Von Neumann was curious about the possibilities of creating self-replicating machines and whether life itself could be reduced to a set of logical rules. That idea remains intriguing today and as relevant as ever. In this regard the book Artificial Life has maintained its power to inspire the mind to ponder upon the possibilities of some of the cellular automata models and how nicely they map to certain phenomena we can observe in nature. Together with the ideas championed by Richard Dawkins in his book Selfish Gene, the ideas built on top of Robert Axelrod's observations on game theory and evolution of cooperation and following some of Daniel Dennett's writings makes one wonder whether it is possible to discover universal automatism governing the universe. It is certainly more attractive idea to me personally than explaining life with some variation of vitalism, a magical incantation of an unknown power we're unable to view, study or repeat.
It's a bit of a shame then than software based artificial life is suffering a reputation of a "non-serious science" status given how fascinating some of the experiments and ideas behind it are. John Maynard Smith at one point called artifical life a "fact-free science". According to Levy's book, one Nobel prize winner described the first ever artifical life conference organized by Christopher Langton at Los Alamos in 1987 as one of the "conferences you might go to but never tell a soul you were there".
One of the problems appears to be the lack of interest on the side of biologists to adopt computer simulations as a tool to study biological processes. The disconnect is with the computer scientist's inability to prove that however life-like their simulation may appear in its results, there's no guarantee that nature is using those very same rules for its own creations. And to be honest, it is a valid criticism. Therefore the biologists still prefer to go out on the field to study their subjects and attempt to explain behavior by observing the complex end result and speculate on the underlying rules in a top-down fashion, as opposed to the bottom-up approach of artificial life.
If biologists are rejecting software based artificial life as a tool to explain life on earth, what then is left for the field of artificial life to accomplish? Going back to von Neumann's original quest to establish the logical rules of life, one would have to first define what is considered living -- a position of hard artificial life that says life as we know it on earth is merely a single variation of it, or more interestingly, that life need not exist in carbon-based physical form but can exist as a set of rules deep inside silicon chips. Of course, life in silico is an idea too many have hard time accepting.
The softer artificial life opportunity is maybe then in entertainment. Following the previous post on Will Wright's speech at TED on using games as toys for learning, if we create the virtual universe ourselves from scratch, then artificial life rules will fit right in -- creating life-like behavior without the burden of having to prove whether mother nature on this planet is following the exact same rules or not. The only kind of biology artificial life needs to answer then is of artificial kind.
Now whether artificial life re-emerges as a hard science or is left as a recreational endeavor is of little consequence to me -- it is far too much fun to play with the rule sets and watch the virtual ants explore the virtual landscapes. And besides, the programmer in me is convinced that the rules are out there, just waiting to be discovered. If life as we know it can be coded in DNA using 4 different acids, if we accept that life on earth is remarkably complex without the need of a highly developed neocortex to design it, and if we refuse to accept the invisible magic touch of vitalism as an explanation for all things living then tinkering with computer simulations may just be what is required to yield interesting and fascinating results, some day.
And it should be a fun and interesting process of discovery on the way there...
A decade and a half later it is interesting to look at the promises presented in print around 1992, and how artificial life seems to have faded from hard science championed by mathematician John von Neumann into a recreational geek hobby following Conway's simplification of von Neumann's universal computers into binary cells of Game of Life. The more promising artifical life research today seems to happen on wetware, in terms of synthetic life, rather than in hardware and software.
Von Neumann was curious about the possibilities of creating self-replicating machines and whether life itself could be reduced to a set of logical rules. That idea remains intriguing today and as relevant as ever. In this regard the book Artificial Life has maintained its power to inspire the mind to ponder upon the possibilities of some of the cellular automata models and how nicely they map to certain phenomena we can observe in nature. Together with the ideas championed by Richard Dawkins in his book Selfish Gene, the ideas built on top of Robert Axelrod's observations on game theory and evolution of cooperation and following some of Daniel Dennett's writings makes one wonder whether it is possible to discover universal automatism governing the universe. It is certainly more attractive idea to me personally than explaining life with some variation of vitalism, a magical incantation of an unknown power we're unable to view, study or repeat.
It's a bit of a shame then than software based artificial life is suffering a reputation of a "non-serious science" status given how fascinating some of the experiments and ideas behind it are. John Maynard Smith at one point called artifical life a "fact-free science". According to Levy's book, one Nobel prize winner described the first ever artifical life conference organized by Christopher Langton at Los Alamos in 1987 as one of the "conferences you might go to but never tell a soul you were there".
One of the problems appears to be the lack of interest on the side of biologists to adopt computer simulations as a tool to study biological processes. The disconnect is with the computer scientist's inability to prove that however life-like their simulation may appear in its results, there's no guarantee that nature is using those very same rules for its own creations. And to be honest, it is a valid criticism. Therefore the biologists still prefer to go out on the field to study their subjects and attempt to explain behavior by observing the complex end result and speculate on the underlying rules in a top-down fashion, as opposed to the bottom-up approach of artificial life.
Most of the time in nature, we can easily see the effect of rules, as classified by Newton. [...] sometimes things are following simple rules when they seem to be following extremely complicated rules, or no rules at all. The problem [is] our own limited vision -- we simply [do] not know these rules. [The] challenge, then [is] to find those uncharted rules -- to discover the natural laws that determine complex systems.The quote above from Artificial Life reminds me once again of Richard Feynman's chess analogy -- attempting to determine the underlying laws of a system by observing the end results is a difficult, arduous process. Sometimes it may seem close to impossible. Artificial life takes the opposite approach, trying to build life-like processes from bottom-up.
If biologists are rejecting software based artificial life as a tool to explain life on earth, what then is left for the field of artificial life to accomplish? Going back to von Neumann's original quest to establish the logical rules of life, one would have to first define what is considered living -- a position of hard artificial life that says life as we know it on earth is merely a single variation of it, or more interestingly, that life need not exist in carbon-based physical form but can exist as a set of rules deep inside silicon chips. Of course, life in silico is an idea too many have hard time accepting.
The softer artificial life opportunity is maybe then in entertainment. Following the previous post on Will Wright's speech at TED on using games as toys for learning, if we create the virtual universe ourselves from scratch, then artificial life rules will fit right in -- creating life-like behavior without the burden of having to prove whether mother nature on this planet is following the exact same rules or not. The only kind of biology artificial life needs to answer then is of artificial kind.
Now whether artificial life re-emerges as a hard science or is left as a recreational endeavor is of little consequence to me -- it is far too much fun to play with the rule sets and watch the virtual ants explore the virtual landscapes. And besides, the programmer in me is convinced that the rules are out there, just waiting to be discovered. If life as we know it can be coded in DNA using 4 different acids, if we accept that life on earth is remarkably complex without the need of a highly developed neocortex to design it, and if we refuse to accept the invisible magic touch of vitalism as an explanation for all things living then tinkering with computer simulations may just be what is required to yield interesting and fascinating results, some day.
And it should be a fun and interesting process of discovery on the way there...
0 comments:
Post a Comment