August 15, 2011

## The improbability of ID

ID proponents are frequently appealing to probabilistic arguments as evidence for their ID claims.  Unfortunately, most of the presented arguments are wrong.  There’s a particularly egregious example posted today at the Uncommon Descent blog.  The author of that post, JonathanM, apparently managed to get into a debate with Massimo Pigliucci.  He quotes Pigliucci as saying:

No evolutionary biologist I know…actually attaches probabilities to specific evolutionary events of the type you are talking about. There is no way to do that.

JonathanM then goes on to cite places where evolutionists have used probabilistic reasoning.  Apparently, JonathanM has no understanding of the difference between probabilities of specific events, and the use of probabilities over populations.

Here’s an illustration of the problem.  If I shuffle a deck of cards, and then deal out a bridge hand, I will have produced a highly improbable event.  If you were to list a particular hand before I had shuffled and dealt the card, then the probability calculation would show that the hand you listed was very unlikely.  If I had then dealt that actual hand, you would have reason to question whether I had been cheating.  However, once a hand has already been dealt, it makes no sense to compute the probability for that hand.  It does not tell us anything useful.

If you really wanted to look at a few hands that I had dealt, to find evidence of cheating, there is a way to do that.  You would need an alternative explanation as to how those hands were dealt.  And then you could calculate the conditional probability:  given that this hand was dealt, what is the conditional probability that it was dealt by method X (say, standard shuffling) rather than by method Y (your alternative).

It isn’t the direct probability of the hand that matters, it is that conditional probability.  And we can only use that method if we have sufficient data to realistically estimate the condition probability.

Unfortunately, the ID proponents don’t seem to understand this.  They do not use conditional probabilities in their arguments.  Perhaps this is because an estimate based on conditional probabilities would show that natural causes are far more probable than supernatural causes.

It is not just JonathanM who is confused about this.  His blog post has been made into a “sticky” and thus highlighted on the Uncommon Descent blog.  So whoever makes the decisions about such highlighting is presumably just as confused.

After citing his examples of statistics applied to population genetics, JonathanM comments “To this, I received no response.”  That, I can understand.  By this time, Pigliucci must have recognized that Jonathan was driven by ideology, and unwilling to learn anything.