Eminent philosopher of science Elliott Sober is always worth reading. He takes ID seriously and tries to offer principled critiques–and he’s even willing, if need be, to let his critiques slice both ways.
Case in point: Sober recently spoke at the University of Montana, criticizing both ID and Darwinian arguments for

allegedly smuggling in assumptions about the designer’s purposes. He believes that ID proponents cannot say something is designed unless they know the purpose for which the thing is designed.

Likewise, Sober criticizes Darwinian dysteleological arguments for claiming knowledge of a designer’s purposes. Stephen Jay Gould’s argument that the panda’s thumb is poorly designed, for instance, depends upon knowing the designer’s purposes. After all, the natural question is: For what is the panda’s thumb badly designed? According to Sober, Darwinists cannot claim a designer would not have designed the panda’s thumb if they do not know what purpose the designer had in mind. Pandas’ thumbs may be badly designed if they were designed for typing; but they may be well designed if they were designed to strip bamboo.

At the risk of sounding biased, I think his critique of dysteleological arguments holds while his critique of ID does not. Here’s why:

Dysteleological arguments are inherently theological. They amount to “a designer would not have done things this way.” It is a theological objection which requires a theological response. Christians usually claim natural evil comes from the Fall; Buddhists claim evil is an illusion; Zoroastrians claim Angra Mainyu is responsible, and so on.

But do ID arguments claim that things are designed because a designer would have done things a certain way? No, and they do not need to. Designed objects generally exhibit certain hallmarks of intelligent activity (e.g. specified complexity or irreducible complexity), and these can be identified regardless of the designer’s motives or purposes. (We are not even “assuming” the designer is intelligent; we are inferring it from evidence.)

ID merely studies natural objects to see if they exhibit these features. It does not try to study or identify the designer or designer’s motives. This focus on identifying empirically detectable features is what makes ID a scientifically testable concept. We are inferring intelligence–not postulating motives and purposes.

Finally, as another way to see Sober’s error, think of things that we know are designed–the arrangement of stones at Stonehenge, say, or the statues at Easter Island. Now if you are like me, you have no idea who made these things or why. And yet we know they are designed. This clearly shows that design inferences can be independent of assumptions about designers’ identities, motives, and purposes.