Dembski's "Explanatory Filter" (EF) claims to be a reliable technique for detecting design. To date, the EF is the only method presented by the "science" of ID. How well does it do? Nobody knows. It has been applied precisely once, by Dembski in his book No Free Lunch. And that application was a dismal failure.
Before going into the reasons that the EF is a psuedo-algorithm, I'd like to present an example of what Dawkins calls a "designoid," that is, something that appears designed but isn't. A "false positive" for the EF, if you will.
Objects almost identical to these were dredged from a shallow bay at the bottom of which some think is the remnants of an ancient city. The objects are made of a concrete-like carbonate, and have a single bore-hole down the center. They are about 4 cm in length.
The natural questions to ask are:
1. Are the objects designed?
2. How would you apply the EF to determine if they are designed?
3. Are they "specified" (whatever that means)?
4. What objective method could one use to determine if they are specified?
5. what other information do you need to apply the EF to these objects? Do you need to know anything of the actual history of their formation?
These objects are just the latest in a long line of designoids which would seem to be natural candidates for the EF (other examples include the visibility of solar eclises, geological formations on Mars and Arctic stone circles.)
So far in all cases these questions have been met with stony silence.
But then again, Dembski may be doing something right: if there's no way to apply your method, you are guaranteed to have no false positives.
(Thanks to Paul Heinrich of the Louisiana Geological Survey. Links to his pages will be posted later, after the design implications of thiese objects have been pondered awhile.)