A few months ago I reviewed Joan Roughgarden's book "Evolutions Rainbow". Now that SEED magazine has published an interview with her, I thought about writing about it again (or just republishing the old one), but now I see that I do not have to, because PZ Myers did a much better job at it than I could ever dream of doing, so go and read it.
The only sentence I did not like was: "There are objections that this requires group selection, which always puts an idea on shaky ground...." As someone who has studied group selection (both biological and philosophical literature) intensely over the past few years, I do not think it is on a shaky ground at all, though some people (mostly those who believe in mathematical models more than real data) may tell you so.
Comments
I'd really like to bitch about this statement as it applies to mathematical modeling - "though some people (mostly those who believe in mathematical models more than real data)may tell you so." A good mathematical model is based on real data and produces results that mirror real world observations. A good mathematical model will, if that is it's point, have good predictive value. However if you build a model on nothing but assumptions fed to you or a lack of reliable data, the resulting model will probably be pretty much trash. *stepping off soapbox*
Posted by: JacquiG | June 13, 2006 6:09 PM
How is group selection not BS?
Posted by: Nymphalidae | June 13, 2006 6:22 PM
I'd like to hear more about your views on group selection. (Let's you and PZ fight!)
Posted by: Bill Hooker | June 13, 2006 6:24 PM
Mathematical models are useful, as long as you remember that a model can be general, or precise, or realistic, but not all three. I am taking a stab (friendly hazing between Razib and me for along time now) at those who reify math, especially the ancient pop-gen math, which has more caveats and assumptions than a useful math model should have.
Sober and Wilson, in the first half of "Unto Others", show why classical math models cannot even see group selection if it hits them in the face - they are not designed to do so - and provide alternative models that are capable of detecting it and that are useful in the lab/field work as well as for conceptual thinking on the issue.
Bob Brandon's chapter on Units of Selection in his "Adaptation and Environment" is really what made it clear to me that group selection is not just possible, but probable and certainly not on any kind of "shaky ground". That is a MUST READ for everyone interested in the topic. One of these days I may write a post about this, but not right now - too much other stuff interests me at the moment.
Posted by: coturnix | June 13, 2006 6:32 PM
Coturnix - you're showing a much deeper understanding of mathematical modeling than I generally see when you say that a model "can be general, or precise, or realistic, but not all three." Ultimately, a person needs to decide what you want to model and what you want to get from the model. I'm often left with the feeling, however, that the general public doesn't understand what a model is and goes more to the "it's math, so it must be right" position (after all us mathy types are really smart if we can do math LOL).
Posted by: JacquiG | June 13, 2006 7:31 PM
Oh, I'm not dead set against group selection. I just think it demands a much higher standard of evidence than most people give it, and I don't think that Roughgarden has met that standard.
Posted by: PZ Myers![[TypeKey Profile Page]](http://scienceblogs.com/clock/nav-commenters.gif)
|
June 13, 2006 11:41 PM
I know you are not - I was objecting to the wording, as if you were giving in to the framing by the selfishgenians. I agree that Joan is far from showing group selection in her study of homosexuality, but here I was talking about group selection as a phenomenon in general.
Posted by: coturnix | June 13, 2006 11:49 PM