This post concludes my recent splurge on Intelligent Design (ID).

I have been trying to argue:

  1. That ID is basically anti-evolutionism
  2. That, as anti-evolutionism, ID is half an argument for creationism
  3. That the question of whether ID is science is actually quite dull
  4. That ID proponents are right to point out the speculative nature of most Darwinian explanations
  5. That science can, in principle, point towards an Intelligent Designer
  6. That arguments about complexity are perhaps quite complex, and
  7. That educating people about all of this might not be a bad idea.

Now, in conclusion, I want to use Bayes’ theorem to settle this once and for all.

We’re going to look at the plausibility (probability) of Darwinism, given the data, and the plausibility of ID, given the data, and see which is more plausible.

To cut a long story short, we are interested in the

\[\mathrm{Pr}(D|M_i)\]

bits in the Bayes factor. This is the probability that the data would be observed under the model in question.

Now, for a Darwinian model, assuming everything the ID proponents say is true, the probability of observing the data that we do observe is extremely small, because of irreducible complexity and the like.

But what is the probability of observing the data under the ID model? Here it gets interesting. The ID model is that some Intelligent Designer(s) has/have done some stuff to our genomes between the origin of life and the present day. The model therefore has a rather large number of free parameters, at least one parameter for each opportunity the Intelligent Designer(s) had to make one change to a creature’s genome. Okay, the model has an infinite number of free parameters. Now, the probability that we observe the data that we do observe, with some specific values for these free parameters, is, of course, 1.0. The Intelligent Designer(s) just made it like that. But we don’t know in advance what the values of those free parameters are. So we have to take an average. And the average probability of the data being observed, considering all the possible values for this infinite number of free parameters, is, of course, zero (or infinitesimally close to zero, if you want to be pedantic).

Now, to find out which model is favoured by the data, whether Darwinism or ID, we simply divide the first number by the second. So we have the odds in favour of Darwinism over ID being: (an extremely small number) / 0 = infinity. That is, given the choice between Darwinism or ID, the data suggest overwhelmingly that we should root for Darwinism.

All of the above is just a very long way of saying that ID, being utterly vague about the nature of the Intelligent Designer(s), has essentially no explanatory power, and therefore offers no reason for anyone to believe it.

So, in summary, the anti-evolution arguments employed by ID proponents are helpful, and suggest that we should explore other possibilities. But ID proponents should turn their attention to constructing models: who/what is/are the Intelligent Designer(s)? When did they alter the genomes of living creatures, to what end, and in what way? It is only when we move from vague ethereal ideas to specific models that it is possible to make specific predictions, and it is only when we have specific predictions that there is any hope of coming up with something more compelling than Darwinism.