Principia

Dembski misapplied the NFL theorems. That is to say that he neither came up with the theorems (which he titled his book after), nor did he seem to understand them. Recently he's retreated to saying that NFL isn't a major part of the thesis but rather his "displacement" theory is. Whatever.

The NFL theorems say (roughly speaking, of course) that there is no advantage to any search algorithm when looking across all problems (i.e. all cost functions), especially when compared to a random search. In other words, to claim that an algorithm performs "better" than any other requires some restriction on the problems (search spaces) being considered. Or, put another way, if you want a performance gain over random search, you have to tailor an algorithm with a priori knowledge about the domain of problems that you are working with. So, Dumbski would have the IDiots believe that because "chance" in biology is a random search, it is impossible by NFL for evolution of complex structures to occur, unless some ID tailored evolutionary processes to develop these structures.

Of course, no one is buying it. Not even one of the authors of the NFL theorems, David Wolpert.

Doubting Didymus

Thank you everyone for pointing out to me the independance of no free lunch theory. I feel very sorry for the original authors of the theories that their work is stained by the dembski brush.