At Home in the Universe, by Stuart KauffmanPosted: July 12, 2005 at 11:38 pm in nonfiction
This was referenced in the footnotes of some other book that I read, but I can’t remember which one any more (maybe Six Degrees?). Kauffman is a MacArthur Fellow who works at the Santa Fe Institute, which is a center for studying complexity theory (and a place I’ve occasionally dreamed of working at), so his credentials were in place. And I’m fascinated by the idea of self-organizing networks and other aspects of complexity theory, so I tossed the book in my last Amazon order several months ago.
However, it was tough sledding to get started (this was the last remaining book from that Amazon order, so I had to finish it before I let myself make another order), and even tougher to keep going. I got about a third of the way through it on my Boston trip in May, and then it sat on my shelf for a month, before I finished it off over several BART rides the last few weeks.
I don’t particularly like Kauffman’s writing. It’s both dense and uninteresting; this last week, I was skimming large portions of text just so I could finish the book off. I also don’t like his prejudices. He clearly comes to his work with the preconceived notion that the universe can’t be completely random, and that it is not possible for humans to be merely an accident.
We are but accidents, we’re told. Purpose and value are ours alone to make. Without Satan and God, the universe now appears the neutral home of matter, dark and light, and is utterly indifferent. We bustle, but are no longer at home in the ancient sense. (p. 4)
So this book is Kauffman’s attempt to demonstrate that humans are an inevitable result of the universe. That we are “at home in the universe”, whatever that means.
It’s a pity, because he’s done a lot of interesting simulation work, setting up toy models on the computer to gain insight into how systems can evolve increasing amounts of complexity, by being poised on the border between chaos and order. However, I don’t feel that he successfully demonstrates a strong connection between his models and the world at large. Some suggestive correlations, at best.
Things that I found interesting:
- The idea that we can counter the unimaginably small probabilities of life being started by some sort of auto-catalyzing reaction with the unimaginably large numbers of combinatorics. As he points out, if you have ten thousand organic molecules floating in a soup, you have a hundred million different possible one-to-one reactions. So even if it’s a one in a million chance that something interesting happens, it should happen a hundred times in such a mixture. Of course, the one in a million estimate of the probability is pulled shakily from a different experiment, but I’ll ignore that part.
- The idea that evolution by natural selection is a tremendous way to optimize among various conflicting constraints. He makes an appropriate analogy to the difficulties we are now running into with the design of technology, where optimizing among hundreds of different constraints is straining the abilities of our design teams, who can’t keep track of the myriad of combinatorial ways in which things can interact. Some food for thought here.
- He also points out a correlation between natural evolution and technical evolution. In nature, the Cambrian explosion showed organisms evolving in a bunch of different directions, setting the foundation for most of the phyla found today. All of natural evolution since then has been refinements on a theme. In technical evolution, the early days of an invention show all sorts of wild variations; he uses the example of the early days of the bicycle with the pennyfarthing arrangement, etc. Eventually the different variations converge on a “best” design, and from there, only minor refinements are added. He comes up with a clever model to illustrate why the two situations might be parallels, but I think the connection is weak. (p. 202)
- He addresses the interesting problems of trying to optimize for an environment that is constantly changing based on how other organisms are evolving. For example, the ongoing war between predator and prey – traits that select for survivability of the prey in one generation creates a new element of selection for predators in the next generation, etc. It’s a completely nonlinear time-dependent optimization problem.
- I’m glad I punched through to the end, because the second to last chapter was excellent. Given the difficulty of trying to optimize to an ever-changing environment, he tries a simpler approach for the purpose of modelling, which he calls the patch procedure.
The basic idea of the patch procedure is simple: take a hard, conflict-laden task in which many parts interact, and divide it into a quilt of nonoverlapping patches. Try to optimize within each patch. As this occurs, the couplings between parts in two patches across patch boundaries will mean that finding a “good” solution in one patch will change the problem to be solved the parts in the adjacent patches. Since changes in each patch will alter the problems confronted by the neighboring patches, and the adaptive moves by those patches in turn will alter the problem faced by yet other patches, the system is just like our model coevolving systems. (p. 253)
He points out the resemblance to representative democracy and to capitalism. Invisible hands and all that. The idea of using different elements (or “patches”) self-optimizing to create a better environment for all leads to thoughts of anarchy, which I like.
There are definitely some thought-provoking ideas in this book. I think Kauffman asks a lot of the right questions. How do things self-organize? Can we learn from nature how to design things better? But I thought that the author’s prejudices were too apparent, and detracted from his presentation of his modeling work. The models that he did present tended to come across as too simplistic, in what appeared to be an attempt to justify the results he wanted to demonstrate. Admittedly, he’s trying to model some very very difficult things to model. And I think his toy models illustrated some fascinating leads into the study of self-organization, evolution and complexity. But he tries to stretch his suggestive models too far in the service of his over-reaching conclusions, and that detracts too heavily from the book overall for me to be able to recommend it. Give it a pass.