Wesley R. Elsberry posted Entry 5 on March 23, 2004 04:33 PM.
Trackback URL: http://www.pandasthumb.org/cgi-bin/mt/mt-tb.fcgi/4

William A. Dembski recently published a book, The Design Revolution: Answering the Toughest Questions About Intelligent Design. The subtitle offers a promissory note, and so do several of the blurbs on the dust jacket and front matter to the effect that Dembski covers herein all the criticisms that have been offered about "intelligent design" and Dembski's particular contribution, "specified complexity". This is untrue, as I will attempt to demonstrate.

Dembski has gone so far as to challenge critics to find criticisms that he missed entirely. I responded earlier, pointing out various difficulties in proving a negative as Dembski had asked for. At that point, I had not read the whole text of TDR carefully, and thus did not wish to offer something that might be discussed on pages I had not yet gotten to.

But now I have read the whole thing, and one rather glaring omission is evident to me. Throughout the book, Dembski touts "specified complexity" as a reliable marker of "intelligent design" and says that certain biological systems, such as the flagellum of E. coli bacteria (please note, Dr. Dembski, that the species name is NOT capitalized) have "specified complexity". This completely sidesteps and ignores a criticism I raised at the CTNS/AAAS "Interpreting Evolution" conference at Haverford College, June 17th, 2001, in a presentation on Dembski's ideas. I pointed out that Dembski's program of submitting events of known design and unknown cause to his "explanatory filter/design inference" were no test at all of the reliability of his apparatus, calling it a "verificationist program". To actually test his methodology Dembski and other ID advocates would have to examine cases where we had biological systems with a sufficiently detailed evidential record that even the ID advocates would agree beforehand that natural causes were sufficient to explain their deveolpment. Only then would running them through Dembski's "explanatory filter/design inference" place the EF/DI at hazard of showing a "false positive" result. I suggested that Dembski and his comrades at the (then) Discovery Institute Center for Renewal of Science and Culture concentrate upon systems like the impedance-matching system of the mammalian middle ear and the Krebs citric acid cycle, which exemplified complex systems for which biologists have accumulated a strong set of empirical data concerning their development.

I reiterated the criticism in a web page critiquing Dembski's book, "No Free Lunch".

TDR is bereft of any hint that this criticism impinged upon Dembski's mental processes at all. At least, there's no hint that I see. To the contrary, there are still the explicit statements that the EF/DI is somehow verified by Dembski's program of feeding it examples that cannot possibly put it at risk of failure. The bogus and completely unfounded "reliability" that Dembski claims thereby is used in several places in TDR to assert strong claims about the need to incorporate Dembskian notions of "design" into biology.

That seems to me to be avoiding a tough question about "intelligent design".

Further links:

Commenters are responsible for the content of comments. The opinions expressed in articles, linked materials, and comments are not necessarily those of PandasThumb.org. See our full disclaimer.

Comment #98

Posted by Paul King on March 25, 2004 12:42 AM (e)

So far as I am aware the discussion between Dembski and Orr following the publication of _No Free Lunch_ ended with Dembski asserting that the filter should ignore all non-design explanations which were not “sufficiently detailed” to calculate the probability.

But if CSI is calculated on this basis then it cannot be taken as a reliable indicator of design. The output in this case is a result of privileging design over any alternatives - it is the result of a methodological bias. Design is even preferred when there is an apparently viable alternative explanation - and where no design-based explanation is on the table.

Unless Dembski has gone back on that assertion and instead accepted that to identify CSI all non-design explanations must be ruled out by actual argument - and actually done so - then it is hard to see how Dembski can claim BOTH the that flagellum of E. coli is CSI AND that CSI is a reliable indicator of design.

As it is it seems that Dembski has “rigged” CSI to favour the “design” category to the point where it cannot be considered a reliable indicator of design.

Comment #139

Posted by Richard Wein on March 25, 2004 10:30 AM (e)

Paul is spot on. To put it another way, Dembski’s method of design inference is based on an argument from ignorance or god-of-the-gaps argument. It infers design whenever science does not have an exhaustively detailed explanation for the phenomenon in question. Of course, if any exhaustively detailed explanations (or “chance hypotheses” as Dembski calls them) are available, it is necessary to eliminate them before inferring design, and that’s where his (flawed) hypothesis testing method comes in. Furthermore, if no chance hypotheses are available, then Dembski invents a spurious hypothesis based on a uniform probability distribution, rejects that and infers design. In the case of the bacterial flagellum his hypothesis assumed that the flagellum arose as a purely random combination of components.