Richard Dawkins

A rather bright and witty fellow who delights in being controversial and pleasant at the same time.

Currently the Charles Simonyi Professor of the Public Understanding of Science, Oxford University. The chair was endowed in 1995 by CharlesSimonyi.

Books by Dawkins include...

The Selfish Gene introduces the notion of natural selection applied to the individual gene. If one considers an organism to be simply a vehicle that genes construct to carry themselves about then the survival value of any particular gene product must be calculated based on the propagation of any identical copy. This allows Dawkins to apply cold Darwinian logic to previously difficult to explain behaviors, like altruism.

The Extended Phenotype is Dawkins response to selfish gene criticism from his peers. He makes more careful arguments and doesn't shy away from references. I found it handy to have a dictionary of biological terms on my lap as I read.

The Blind Watchmaker is, among other things, serious attempt to refute creationist literature by offering clear and compelling explanations of the many facets of evolutionary theory. In one chapter he describes a program in which the evolution of a parametrically described figure is directed by manually selecting the most desired of several randomly adjusted progeny. The program's rapid convergence surprised even Dawkins. (Dawkins had constructed what JohnHolland would call a GeneticAlgorithm.)

This book is also an excellent introduction to Evolutionary Biology for anyone who is wondering how it might work, but isn't sure. The book is intended for a wide audience and is not at all technical.


Dawkins introduced the notion of memes, the mental equivalent of genes, intending it merely as an illustration of a selection processes taking place in other than a strictly biological environment. The concept has infected many diciplines and led to the recent founding of the JournalOfMemetics. We can think of PatternsAsMemes.

It should be noted that Dawkins probably doesn't think of "memetics" as a scientific theory with proper standing, but more as a good analogy to clarify our thinking about "real" evolution, that of genetic systems. See MemesShmemes for some critical views on memes.


And it should really be noted that he doesn't think of genes as actually being selfish either, just that they exhibit behaviour which could be regarded as selfish. (Being a logical positivist myself, I'd say he probably should.) Also, lots of great ideas about the workings of evolution are brought up in the book 'Erehwon' by SamuelButler? ( which I mention here because I think he deserves a lot of the credit that has been distributed to others ) --RichardCordova


When my eldest daughter was about two I gave her the BlindWatchmaker? program and asked her to "make a big circle", "make a long thin line", etc. She was able to make the requested figures in a few generations. Of course she (or I, for that matter) couldn't have manipulated the genome to create the same figure. This led me to postulate a new kind of programming, AnalyticProgramming?, where programs are generated automatically and the "programmer" says whether they are correct or not. I haven't done much with the idea since, but I'd love to explore it in the future.

  --KentBeck

Isn't that what they call GeneticProgramming ?

Yes and No. It's the distinction between supervised learning (what Kent calls AnalyticProgramming?) and unsupervised learning (see e.g. http://en.wikipedia.org/wiki/Unsupervised_learning ). You can apply it to GeneticAlgorithms or to many other learning methods like SupportVectorMachine?s and others. It depends on the way the learning algorithm gets feedback (explicit if supervised). The unsupervised case really is the more difficult one because the question whether an answer is OK is needed too, but must be inferred somehow (statistics or analytic).


CategoryAuthor


EditText of this page (last edited June 3, 2008) or FindPage with title or text search