Wednesday, February 25, 2009

Further thoughts on Chapters IX-X and the suppl. reading

Good discussion today, and a good supplementary reading. Most of the issues I wanted to raise were touched on in class, so I don't have too much more to add. I'll recapitulate some points I enjoyed chewing on, and folks can comment if they have further thoughts.

1. Dealing with the unknown in science. The debate between Darwin and Kelvin on the age of the earth is a fascinating case study of the informal, heuristic strategies scientists use to estimate unknown quantities and bound the uncertainty of those estimates. The strategies on display include:

(a) 'Back of the envelope' calculations to roughly estimate an unknown quantity. This is how Darwin estimated the time required to denude the Weald. The trouble is that it's hard to say how much confidence we should have in such estimates. What's the range of plausible estimates?

(b) One way to determine the plausibility of a back of the envelope calculation is via a second, independent back of the envelope calculation. If both calculations give you the same answer, it's either a huge coincidence, or it's because both calculations are giving an approximately-correct answer. This was Kelvin's strategy for estimating the age of the earth. I had known Kelvin estimated the age of the earth, but hadn't known he'd made two independent estimates. So I now have much more respect for him as a scientist and a much greater appreciation for the pressure his calculations put on Darwin. Of course, Kelvin's calculations weren't entirely independent; they were based on some shared assumptions, which perhaps should've been cause for caution (though recognizing shared, implicit assumptions is often hard, as William Wimsatt's discussions of "independent" models of group selection point out). In hindsight, we know that Kelvin's calculations were way off, because he didn't know about radioactivity. But don't be too hard on Kelvin; there was no way for him to know his picture of how the world works omitted a hugely important, unknown factor (radioactivity). Such factors can never be ruled out, but it's never reasonable, or even possible, to allow for the possibility of them in our science. Telling scientists to "always keep in mind that your theory could be totally wrong because of some hugely-important factor nobody's ever thought of" isn't a useful guide to action.

(c) Simply admitting ignorance and calling for more research. This is basically the strategy Darwin ended up with by the 6th edition of the Origin. It seems to me to be a reasonable strategy, given that there were many reasons for treating evolution by natural selection as a working hypothesis useful for guiding further research, even lacking a good estimate of the age of the earth (lack of viable alternatives, independent evidence favoring natural selection, curiosity about how evolution by natural selection might proceed in the future, etc.) Part of being a good scientist is being a good judge of when to keep pursuing an imperfect or incomplete hypothesis, and when to drop it (either in favor of a competing alternative, or in favor of pursuing a completely different research topic). Philosopher of science Thomas Kuhn (in)famously thought that such decisions were entirely irrational, but this claim can be questioned on various grounds.

3. Ironically, debate about the correctness of uncertain calculations (regarding the age of the earth) was itself predicated on an (implicit) uncertain calculation (regarding the time required for natural selection to produce certain results). I'm a little surprised that apparently no one recognized this at the time. Or perhaps Darwin discussed in his letters why he thought natural selection needed more time than Kelvin's calculations allowed?

4. Turns out the hierarchy of sciences (aka "physics envy") goes back at least as far as Victorian times. So does the intimidating effect of mathematical calculations on those who don't know math.

5. Darwin's writing in the Origin hints at an implicit philosophy of theory testing, but it's not a philosophy Darwin actually adhered to in practice. Darwin says of many objections to his theory that they would be "absolutely fatal" if true. I assume he writes this way because these objections really did seriously concern him; it's not just a rhetorical technique he's deploying to make himself look brilliant when he refutes the objections. Darwin often sounds like a fairly strict Popperian falsificationist when he's discussing objections to his theory. That is, he sounds like he'd be prepared to junk the whole theory if even one fact could be shown to be inconsistent with the theory. The history of science shows that most scientists don't operate that way, and for good reason (science needs working hypotheses and so can't go chucking entire theories at the first sign of trouble). And arguably, Darwin himself didn't operate that way. His response to Kelvin was to modify his theory to include a greater role for rapid Lamarkian evolution, not to throw his theory out and start fresh.

6. Many of Darwin's arguments about the imperfection of the geological record are exactly the same as those deployed against Gould and Eldredge's modern idea of punctuated equilibrium (roughly speaking, the idea that evolution is characterized by bursts of rapid change associated with speciation events, separated by periods of stasis). We may talk about this more in the session on the Modern Synthesis and its critics.

7. A tangential issue: You might think there's a tension between my suggestion that scientists need working hypotheses (even imperfect ones), and my suggestion that it's ok to just admit ignorance on the question of the age of the earth, pending further research. Doesn't further research on the age of the earth itself need to be based on some working hypothesis, so that confessing ignorance on this question is a non-starter? I don't think so (write a comment if you disagree!) To estimate the age of the earth, I don't think you need a working hypothesis about its age. The question is a purely descriptive, empirical one, rather than a theoretically-motivated one. It's like estimating the height of a mountain, or the average depth of the ocean. To pursue those kinds of purely descriptive questions, you need to have working hypotheses about the sources of error and bias in one's estimation method, so that you can minimize or eliminate those errors and biases. But to identify and study the mechanisms of evolution and work out the consequences of those mechanisms, one needs to have some working hypothesis about the identity of those mechanisms. That is, one needs to have an underlying theory of evolution. (We could argue about whether it's ok to think of that theory as a purely instrumental predictive tool rather than as a true description of how the world actually is, but that's a whole other ballgame...)

No comments: