Alex Pruss has a very nice post on his blog entitled, "Design, Evolution and Many Worlds." His post prompted some thoughts (which I posted in his combox), which I've posted below:
It be that there is an anthropic selection bias against design strong enough on the evolutionary theorist side that it undercuts justification in a Darwinian (or specifically a kind of mechanistic account of) evolution? That is, one could argue that "not seeing" design in nature might itself be the result of a selection bias, let's say for a view of things on the model of a machine (think of a watch) that had its parts brought together randomly? Perhaps it is no coincidence that the western mind was captivated by the complex clocks devised by those in the Islamic Empire in the Middle Ages before the scientific revolution. So, thinking of the world in a mechanistic way followed from an anthropic predilection to modeling our understanding of things on machines. Support for this theory may lie in the fact that people often think that science explains things mechanistically (bouncing billiard balls), when in fact this is not a correct model for understanding the breadth of scientific theory today (even if only considering physics). I'm specifically thinking of Einstein's denial of mechanism as he defines it in his The Evolution of Physics as well as Burtt's account The Metaphysical Foundations of Modern Science.
One characterization of a selection bias is perhaps as follows: given a de facto scenario in which one makes a judgment one on the basis of that scenario infers that some intelligibly discernible and recognizable pattern accounts for some feature of that scenario. So, from our specific situation and looking back on our history one might infer that there must have been a designer to account for the existence of human beings. In this case the recognizable pattern would be that of an artist fashioning an object. However, justification for such an inference is undermined by the selection bias problem. It would be similar to shooting an arrow into the air (where it lands I know not where), finding its exact resting place then thinking, "Wow! How improbable that it would land in this spot given all of the possible places that it could have landed! I see here an intelligible pattern by which I must infer that something guided it to this spot!"
With that in mind, perhaps my earlier suggestion (some inference against design) could be thought of also as involving a selection bias of this sort. The best way to do this is to apply the characterization of selection bias to an inference against design. The de facto scenario is the existence of human beings given evolutionary descent. From this fact one infers that the process was not a product of design by identifying a recognizable pattern to serve as an explanation; namely, that of a machine that exists in virtue of the accidental joining of preexisting independent parts resulting in a functioning whole in virtue of the intrinsic properties of its parts. (Let's say that the person who sees things this way came across in a forest one day something that could function as a figure-four deadfall animal trap that came out without human agency as the result of a thunderstorm and subsequent flooding). From this recognition, she infers that human beings must not have come about from design. I see this particular case as parallel to the charge of selection bias for the ID theorist.
However, I'm inclined to think that neither scenario actually suffers from bone fide selection bias. One can only commit a selection bias if one knows (or has reason to think) that some particular pattern is at work. On the one hand, it seems to me that the reason why one wouldn't be justified in inferring from Pruss's computer simulation that there was a forward-looking bias in the movement toward the red line is only in the case that one knows that this was one of 23773 unsuccessful attempts. On the other hand, regarding the inference to non-design, let's say that God was running a computer simulation like Pruss's but basing his simulation on the actual factors that fully account for the evolutionary development. And in this case, 23773 simulations had less circuitous paths to the coming to be of human beings! Thus, I don't think that one could charge the person with a selection bias unless she had some idea of what the results of God's computer simulation were.
My suspicion at this point is that we should think that the design explanation (design) is a bad explanation (at least in science) and the non-design (mechanism) inference is a good one. But why should this be so? Its not like design is a bad explanation for something and it does allow the existence of things that mechanism has a hard (or impossible) time accounting for. For example, design might be a good explanation for why biologists find answers when they ask the question, "why does this function work the way that it does?"However, anti-ID folks could respond by saying that the best explanation is one that relies only on the intrinsic properties of tiny things that can come together in various causal ways because these are free from anthropic bias. However, I remember Chisholm mentioning at the end of the "Problem of the Criterion" that the best pattern for thinking about efficient causation is that based on our innate and insider sense of agency in the world--all efficient cause is something like my ability to push my coffee cup across the table. A similar thing can be said for how we envision the properties of physical elements. So it might not be as simple as that.