The question of how we detect design is a question that both atheists and theists can equally well engage in. The question comes up when one is engaged in the project of looking for scientific evidence for the existence of a supernatural God, but the question also comes up when, for example, one ponders what structures one would have to find on a planet for one to conclude that the planet is probably inhabited by intelligent aliens.
William Dembski famously answers that, to infer design, one must find specified complexity. While I won’t try to argue this here, I share the opinion of many that his notion of specification isn’t that well worked out. Del Ratzsch agrees with Dembski that complexity isn’t enough, but Ratzsch argues that what it takes in addition to complexity is mind-affinity, as well as factors that would be relevant to the intent an agent would have to realise a particular value. (Ratzsch doesn’t have a good name for this latter sort of factor; let’s call it “value-affinity”.)
By “mind-affinity”, Ratzsch is getting at patterns that match patterns that a mind would be expected to create. For example, if we saw “John 3:16” spelled out using rocks on the side of a mountain, we would think that that distribution of rocks was the pattern of design, not chance. With value-affinity, Ratzsch is talking about effects that an agent would want to realize in the course of realizing some value. For example, there is evidence that in the distant past there was a nuclear reaction that happened in a river, not as a product of human agency. If that reaction didn’t have any substantive effects, we would probably conclude that the reaction happened as a result of chance processes. But if that reaction was causally efficacious in producing intelligence in human ancestors, and intelligence couldn’t have been produced any other way, we would think it more probable that the reaction happened as a result of design.
In my opinion, Ratzsch is giving a more promising answer than Dembski. Ratzsch’s discussion of mind-affinity and value-affinity on pages 61-69 of his book Nature, Design, and Science is quite brilliant. It’s disappointing that the intelligent design literature focusses on Dembski’s answer and not Ratzsch’s.