Thursday, February 04, 2010

Meyer's Interview of Berlinski

Here's an interview of David Berlinski by creationist Stephen Meyer, author of Signature in the Cell.

It's really fascinating for all the mistakes and false claims made by Berlinski, none of which are corrected by Meyer. Here are just a few:

1. Despite the fact that he claims to be intimate with Marco Schützenberger, Berlinski mispronounces his name consistently as ending in "berger" with a hard g, instead of soft g as in French.

2. He claims that Schützenberger was a "professional biologist". This is not accurate. Schützenberger's first doctorate was in medicine, not biology. He apparently did publish at least one paper on a biological topic, but one paper does not make someone a "professional biologist". (I've published a paper in a philosophy journal, but no one would call me a "professional philosopher".) In this interview, Schützenberger admits forthrightly that "Biology is, of course, not my specialty."

3. He claims that Schützenberger was a "world-famous physician". As far as I can tell, this is not true.

4. Berlinski draws an analogy between evolution and the difference in computational power of finite automata and pushdown automata. In finite automata, only simple outcomes are possible because they have no memory; pushdown automata can do more interesting things because they have an unbounded stack. Evolution has no memory, Berlinski says, and therefore is analogous to finite automata.

But this analogy is simple-minded for a variety of reasons. First, while finite automata are limited in their computational power, in practical applications this can sometimes be remedied simply by increasing the number of states. Second, neither finite automata nor pushdown automata are self-replicating models. Third, if you consider that replication is based on DNA, which is potentially unbounded in size, then evolution does have access to a form of memory that is, for all practical purposes, unbounded in size.

5. Berlinski claims that "Until about 1950 or 1960, the mathematicians had not really interested themselves in Darwin's theory of evolution". This is false. The Hardy-Weinberg theorem, for example, dates from 1908. Population genetics was developed by biologists and statisticians in the 1920's and 1930's. Population dynamics goes back to the 1800's.

6. Berlinski claims that "Every mathematician that I've known ... they all had the same reaction [about evolution]: it's kind of nutty." Berlinski knows me - we have corresponded - and I never expressed this reaction to him. There are many books and papers about mathematical biology, and essentially none of the authors express this view.

But then, we already know that Berlinski's claims about what mathematicians believe about evolution are not reliable.


Bill said...

Berlinski is my favorite creationist poseur! If he lived in Texas we'd refer to him as all hat and no cattle.

I listened to an interview with Berlinski in which he explained, mathematically, why a "cow" (his words) could not evolve into a whale. Is argument was that he started to count the "engineering changes" that would have to be made to the cow and stopped counting at 50,000.

Really, Berlinski, and how many Big Chief tablets did that take? Oh, and, gee, Berlinski, you didn't keep a record of the "engineering changes." What a shame!

RBH said...

Evolution has no memory, Berlinski says, and therefore is analogous to finite automata.

That is hip deep bullshit. A population's genome is a memory system, a selective memory for what worked in prior selective environments.

Joshua said...

I find the claim about finite automata particularly hard to understand. Even if I had some sort of finite automata, if they are being repeatedly rerun and their data is impacting the world around them, then they are effectively not going to act like finite automata. To use a silly example:

consider the finite automata that duplicate Conway's Game of Life. The input space is at most 18 bits (based on whether that square is alive and whether the surrounding 8 squares are alive) and then can send out an accept or reject after running where accept is live and reject is die or stay dead. These automata don't have many states (to do this rigorously one needs to decide how the input data comes in an ordered fashion on your tape, but this is just a matter of convention). We now have a collection of finite automata that is effectively Turing complete.

In fact, if automata can reproduce we can just start out with automata surrounding the finite, living region and have each turn them reproduce into their neighboring squares. So we can do this with even a finite number of DFAs.

There are some details that would need to be filled in here to be rigorous but there's no reason this wouldn't be an exercise for an intro theoretic comp sci class for undergrads or even smart high school students.

I really have to wonder if Berlinksi even bothered thinking about this claim at all.

Barry said...

Re: 5. I suppose Lotka and Rashevsky didn't count until the Dover editions of their books came out?!

der_hammerman said...

I appreciate your work in refuting Meyers et al. I'm not a biologist and your words help me in arguing against creationists. Keep it up!

stvs said...

we already know that Berlinski's claims about what mathematicians believe about evolution are not reliable.

R.A. Fisher! Fisher, the famous statistician and mathematician, also began the modern evolutionary synthesis with his 1918 paper "The Correlation Between Relatives on the Supposition of Mendelian Inheritance."

In this paper Fisher also introduced the word "variance" in its modern mathematical context.

Fisher also had this to say about the futility of any "intelligent designer":

"If we imagine, then, some extra-natural agency endeavouring to influence the organic evolution of mammals and birds by the production, on millions of different occasions, of this single mutation, we can recognise that its efforts were futile and inoperative."

As usual, Berlinski's bizarre assertions are easily refuted.