I believe I understand how William Dembski's Design Inference works -- it basically detects patterns that are too good (on some scale, and the relevant one here is fitness) to have occurred by a purely random process such as mutation (unaffected by natural selection), even once in the history of the universe.