I have, so to speak, finished Climbing Mount Improbable. In the sense that I have read the book but also that I am at the peak of evolutionary fitness, because I am alive.
On my old blog I wrote about blind search as a hill-climbing metaphor. Richard Dawkins uses this same idea to describe how various features such as wings, eyes, colourful flowers and such could have evolved. The idea is just as I described:
Searching for things in computer science is often likened to climbing a hill, where the
summit represents the goal, and every point on the landscape represents a candidate
for this goal. The idea is that, without knowing where the summit is, one can get there
by always walking up hill. Assuming an ideal (smooth) landscape one could get to the
top of a hill blindfold. (Not recommended.)
Each step is just a simple adjustment, a tiny, unlooked-for variation in the population. This variation will mean the animal that has it is at a tiny advantage and may spread its genes wider than average. Thus we can get from nothing, to the human eye; or from a similar nothing, to the bird’s wing; or from nothing to the intricate ecosystem that is the fruit of the fig tree.
And like I stated in my other post, scaling a mountain blindfold is possible, but can lead to dead ends:
If we can’t see where we’re going then we can never be totally sure if we’re on a minor
summit rather than the very top of the mountain. This is called a local maximum — every
direction you walk goes down, but there is a direction which will eventually take you to a taller peak.
Unfortunately, evolution doesn’t have the advantage of predicting these problems. So if chance chooses a particular short-term solution which later leads to problems (like the blind spot in our own eyes) there is nothing to be done. There is no option to back-track and start again down a different route — only forward and up.
The book’s very interesting, and certainly very readable. One of the middle chapters on segmentation of insects was a bit tedious. I’m not sure if this is because he was explaining something obvious in great detail, or because it was obvious that the detail was being expended on a description of a computer program to demonstrate something obvious, which itself was obvious. The abstractions he described were rather uninteresting, to me at least. Maybe others would find them less apparent.
I would have preferred, instead, that the final chapter be longer, or even split into two. It had some extremely complex relationships laid out without much differentiation. Indeed there were two types of wasps described in this chapter, the ‘true’ fig wasp and the fake fig wasp, both called fig wasp. Some aliases and maybe a diagram of the relationships would have made this chapter much clearer. And possibly more interesting for it.
Overall I was surprised how short the book was. It only took me a few days to finish and I wasn’t devoting much time to it — I was ill and it was a busy Christmas, after all. I would still recommend The Selfish Gene over this any day.