The Unpublished Challenge

Late last month, Casey Luskin wrote an article advocating a positive case for ID creationism. Jack has already done a good job of refuting Casey’s wishful hand waving.

However, in the comments of the post, Casey and I had a little exchange in which I tried to get him to commit to developing a positive model of common design, and he just, well, assured me that his legal education prepared him to deal with the statistics that he was butchering left and right. Yeah, umm . . . anyways, my last comment was never published, and now enough time has passed that I feel I can share it with you and let you decide why it never showed up on the DI’s website.

Mr. Luskin, my counter to your article is a simple one. If ID is testable, where is the model that allows me to test its fit to molecular data? Until one exists, your article is nothing but spin. Evolution can do this, why can’t ID?

I am being intentionally conservative in how I respond because I don’t want to get side tracked into a long debate over Dr. Theobald’s paper. I mentioned him because I believe his methodology is the best way to test your claims of the utility of “common design.” Thus much of your responses on the quality of his paper are immaterial to the challenge, which is the main point of my argument. (I see them as being off topic. On topic criticisms would be related to the quality of Bayesian analyses.) My reason for bringing up his research is that he developed a robust statistical test for universal common ancestry and executed it. That is more than the fellows of the Discovery Institute have developed in two decades of trying to disprove common descent. His research isn’t perfect, but it is still better than hot air.

If you don’t like how Dr. Theobald constructed his models, his research provides a framework for doing what you see as a proper test. For instance, you criticize him for not using a proper model of “common design”, yet the Discovery Institute has yet to produce a model to of how molecular data should look under common design. In other words, you criticize him for not including a hypothesis that not been elucidated in either scientific or anti-evolution literature. In my professional opinion, I do believe that his independent origin models are a good approximation to “common design” and much more rigorous than anything you or your comrades have proposed. But you disagree, which is why I ask you to educate me on how molecular data should look under a ID-approved model of common design. (Which is another reason why I can’t develop such a model. It won’t be ID-approved.)

If I sound disingenuous, it is because I believe that neither your nor any other fellow of the Discovery Institute has any interest in doing any work to develop a positive model of ID and test whether molecular data fits it. (That doesn’t mean that I won’t encourage you to prove me wrong.) You’d rather complain about Dr. Theobald not including a model of common design approved by you, than actually showing us what that model should look like. This is why I’m trying to keep us on track. Because I want to see an alternative model, not complaints about what Dr. Theobald used.

While it is true that you don’t have to use Bayes-factor analysis for every hypothesis test, it is also true that the body of evolutionary research is rich enough and well developed enough to support such models. In other words, in evolution we have models that allow us to calculate “P(data common descent)”. These models are core to our research and teaching programs. Despite all the rhetoric, ID lacks even rudimentary models for “P(data common design)”. The gauntlet has been thrown down. Care to pick it up?

By challenging you to develop something that can be inserted into a Bayes-factor analyses, I’m challenging you to develop a new, positive approach to ID, one that doesn’t rely on negative argumentation. You seem rather confident on your statistical training from law school and before that, thus I think you should at least try to put something together for the community.

I find it ironic that you insist on citing Dr. Martin’s comments, and I suspect that you have failed to accurately parse his convoluted prose. Paraphrasing him: “Theobald’s paper is trivial because we already know that universal common descent is true. Thus it is not novel and should not have been published in Nature.” Despite claiming to be aligned with Dr. Martin, you clearly are not making the same argument as him, unless you are really willing to claim that universal common descent is so obviously true that statistical tests of it are not needed. I will note here that in my professional opinion, I disagree with Dr. Martin’s views on the novelty of Dr. Theobald’s research.

And finally, the point about “null hypotheses” is an important one. When I teach, it is an important criterion to distinguish those who understand the differences in statistical methodologies that those who don’t. This isn’t a semantic issue; it is a fundamental test to whether you understand what Dr. Theobald did or not. In Fisherian hypothesis testing, a null hypothesis is given special treatment, and picking the null is in reality a matter of taste and convention. In what Dr. Theobald did, there is no hypothesis that was given special treatment, all were treated equally. This is why criticisms that focus on his “null hypothesis” are completely off the mark.

The advantage of treating everything equally is that any critic can develop a better model than Dr. Theobald and test it alongside his models. Instead of criticizing what he did, it would be more productive if you could work on developing a model and show me how it can be done better.