Friday, November 17, 2006

Responding to ID -- A Review of Their Positive Arguments with Rebuttals

**UPDATE (2/9/07): See Ken Miller's Youtube presentation on the errors and problems with ID arguments**

Intro: I simply want to lay out a fair representation of the positive case for intelligent design (ID) in this article, and examine why the case has been ruled a failure by the greater scientific community. Because ID is so vague, eg defined by the Discovery Institute as,
The theory of intelligent design holds that certain features of the universe and of living things are best explained by an intelligent cause, not an undirected process such as natural selection.
Therefore, making an argument for such a claim is going to be difficult from the outset. "Some features"? "An intelligent cause"? We see the vagueness of the definition come back to haunt ID advocates in their inability to put forth convincing arguments which line up with "this is best explained by an intelligent cause" versus "random mutation/natural selection". For example, some of ID's cosmological arguments in The Privileged Planet (largely the anthropic principle) do not [obviously] fit this criterion.

And thus lot of what passes for "arguments" coming from ID/Creationism (IDC) these days against biology is simple: find some part of evolution that people don't know a lot about and say, "How do you explain X?!" This argument from ignorance is common among laypeople, because they don't know enough about the science to make positive arguments for the claims of IDC. Therefore, a great deal of the "noise" and political theatre of IDC is just critiques of evolution. As Judge Jones eloquently pointed out, there is a false dichotomy between showing someone else's explanation isn't good and proving your own case [not granting that the former has been done in the case of IDC]. Just because you do the former doesn't mean you've done the latter -- established evidence for IDC. I would call these critiques and arguments from ignorance negative in character, because they do not purport to demonstrate the truth of IDC.

However, there are some positive arguments for IDC, and they will be the focus of this review. I am going to go through and list the positive arguments (that I'm aware of), and link to strong rebuttals/refutations of those arguments.

Almost everyone already knows the basic arguments of IDC: "design detection" (DD) and "irreducible complexity" (IC). Generalized arguments, though very useful in philosophy, don't work in science. You have to pick particular examples (data) and use these to demonstrate your argument works. Until then, you haven't yet crossed the threshold of science. I'll pick back up on this at the bottom.

Responses to the claims IC are abundant, and here are layman-friendly resources to familarize onesself with the mechanisms by which scientists explain complexity and apparent IC: exaptation/cooption, scaffolding, gene duplications, etc:
1) Prof. PZ Myers' Powerpoint presentation, see slides 29-50
2) Prof. Dave Ussery's paper delineating the four different RM/NS pathways to complexity
3) Don Lindsay, How Can Evolution Cause Irreducibly Complex Systems?

Two particular systems are touted as crucial examples of IC: i) the blood-clotting cascade (BCC)/immune response; ii) the flagellum.

i) The IC arguments for IDC are the brain-child of Prof. Michael Behe, a biochemist at Lehigh U. The BCC arguments can be found throughout IDC websites. One of the most well-informed responders to these arguments is Andrea Bottaro. Last year, as more evidence came in that transposons were involved in the human immune response, Dr. Bottaro put together the pieces and completely refuted the claim of IC as it applies to BCC. How did Prof. Behe respond? By moving the goalposts -- not in any way denying the evidence, but demanding a mutation-by-mutation account of the pathways involved. Thus, he undercuts his own argument by rendering the burden of proof unattainable by any scientific pursuit.

Also see Matt Inlay's article on the supposed IC of the BCC. Matt rigorously examines different branches of the tree of life to demonstrate the fallacious nature of the claim by evidencing the reducible nature of the immune response.

The single best place for you to start in looking at IC as it relates to the immune response is here.

These technical papers, as well as others that can be accessed via this long bibliography on the subject, here, here, and peer-reviewed literature like this Nature review and via searches on PubMed and other scientific databases, decisively defeat the claim of IC as it relates to BCC or the immune response. Again, the very important thing to do is focus in on the specific systems that are being examined, and claimed as evidence, and force the burden of proof upon the claimant. In the cases of IC, these claims have been refuted by the research of biologists and biochemists, which have always shown evidence of cooption and homology of these systems -- completely undercutting the identification of IC.

ii) The evolution of the flagellum is probably the keystone argument of IC. IDC proponents use this system because they feel it is best compared to a "machine", is most difficult to reduce, and is the strongest evidence for intelligent design in biology. One of the most comprehensive resources on this argument is from a continually-updated paper written by NCSE staffer Nick Matzke. Another resource is the PandasThumb section here on flagellum evolution.

In addition, Nick Matzke got a peer-reviewed paper published in Nature Reviews which unequivocally demonstrated, for the first time, that of the 42 proteins involved in the flagellar system of a particular bacterium, all but 2 of them had known homologs! This research involves doing BLAST-type searches in the genome of the bacterium being considered, and showing that the raw evidence of co-optation/exaptation is abundant -- there is no good reason to suppose that all 42 of these components originally had the function that they now do. See also Matzke's reader background page.

Moving on to the second prong of the "case for design" involves taking on those claims that design has been "detected" via mathematical research in IC or other biological systems.

iii) Design detection is the specialty area of William Dembski. Dembski took the "No Free Lunch" (NFL) theorems developed by David Wolpert and others (background with citations) and attempted to use these mathematical constructs to argue a few different things. One of his arguments was the the NFLs showed that evolutionary RM/NS would theoretically not be successful in the development of complexity. Although we can find numerous responses and refutations of this claim (here, here, here , here and see references here -- Wein 2002a,b; Shallit 2002; Rosenhouse 2002; Perakh 2001a, 2002a, 2002b, 2003; Young 2002; Orr 2002; Van Till 2002), I think it best to turn to the person who actually developed the NFL theorems in order to take Dembski to task for misrepresenting them and their implications. Wolpert almost immediately refuted Dembski's claims, and it took a few years before Dembski further tweaked his claims:
[David H. Wolpert] The values of the factors arising in the NFL theorems are never properly specified in his analysis. More generally, no consideration is given to whether some of the free lunches in the geometry of induction might be more relevant than the NFL theorems (e.g., those free lunches concerning "head-to-head minimax" distinctions that concern pairs of algorithms considered together rather than single algorithms considered in isolation).

Indeed, throughout there is a marked elision of the formal details of the biological processes under consideration. Perhaps the most glaring example of this is that neo-Darwinian evolution of ecosystems does not involve a set of genomes all searching the same, fixed fitness function, the situation considered by the NFL theorems. Rather it is a co-evolutionary process. Roughly speaking, as each genome changes from one generation to the next, it modifies the surfaces that the other genomes are searching. And recent results indicate that NFL results do not hold in co-evolution. [emphasis mine]

It may well be that there is a major mystery underlying the performance of some search processes that one might impute to the historical transformations of ecosystems. But Dembski has not established this, not by a long shot.

Dembski refined his arguments and published (2003) a response which attempted to show that even in co-evolutionary processes, that the NFL theorems do still hold. A great deal of the problem with Dembski's work is that it is all on his own website and books, and none of it in peer-reviewed literature. That means that mathematical laymen (like me and most of you) are often going to miss the subtleties in Dembski's articles that peer-review would immediately weed out. The most basic mistakes and differences between arguments by Wolpert and Dembski will be caught by other professional mathematicians and fixed in the MSS before publication. This keeps the "certitude" factor on the IDC side to a minimum, because while Wolpert's arguments are accepted by the mathematical community, which immediately lends substantial credibility to them, and evidences Wolpert's authority, none of this can be said for Dembski. An argument from silence may then be made that Dembski's work cannot clear the bar of legitimacy. Why else would he not want it published academically, if it is valid?

Dembski may possibly complain that he can't get his work published due to discrimination, but this complaint makes little sense: Behe has published work since he published his Darwin's Black Box, and Dembski's work, being mathematical in nature, need not even address the question of evolution directly. Therefore, if any case can be made about anti-creationist bias, it would be much more likely for Behe's work to be "censored", [as they love to claim (without evidence)] due to its intrinsically anti-evolutionary content, versus Dembski's abstract math. Dembski's work may or may not apply to biological systems. Behe's is directly about biology. Therefore, which is more likely to be "censored" by anti-creationist bias, and so does Dembski have any excuse for not publishing his work in a respected academic format?

Conversely, Wolpert has recently published, via peer-reviewed literature, about the co-evolution which accurately models biological systems:
Abstract: Recent work on the foundational underpinnings of black-box optimization has begun to uncover a rich mathematical structure. In particular, it is now known that an inner product between the optimization algorithm and the distribution of optimization problems likely to be encountered fixes the distribution over likely performances in running that algorithm. One ramification of this is the "No Free Lunch" (NFL) theorems, which state that any two algorithms are equivalent when their performance is averaged across all possible problems. This highlights the need for exploiting problem-specific knowledge to achieve better than random performance. In this paper, we present a general framework covering most optimization scenarios. In addition to the optimization scenarios addressed in the NFL results, this framework covers multiarmed bandit problems and evolution of multiple coevolving players. As a particular instance of the latter, it covers "self-play" problems. In these problems, the set of players work together to produce a champion, who then engages one or more antagonists in a subsequent multiplayer game. In contrast to the traditional optimization case where the NFL results hold, we show that in self-play there are free lunches: in coevolution some algorithms have better performance than other algorithms, averaged across all possible problems. However, in the typical coevolutionary scenarios encountered in biology, where there is no champion, the NFL theorems still hold.
Dembski has seized on this last sentence as evidence that he is right. It isn't true. Note that the problem here is still that the NFL theorems are about "algorithms averaged across all possible problems".

It is thus hardly convincing to say that the specific ecological "fitness landscapes", or "search space", is not beautifully searched by RM/NS versus a blind search. Simply put, in order to model biological systems properly, the algorithm itself would have to be altered such that it was not specific-target-directed [multiple positive adaptations are possible], such that each "score" improved the algorithm [the scope of the organism's ability to adapt further], and such that each "score" altered the competitiveness of the landscape [the domain in which fitness is evaluated, here, the ecosystem, which coevolves with the players]. This is the true nature of biological co-evolution -- as organisms adapt, the machinery by which they acquire adaptations itself adapts (consider that increasing surface area for sunlight is useless to animals, but not to plants), and the environment around them is full of other "game players", co-evolving just like them. This is, so far as I am able to tell, the definition of an "open algorithm" during the permutations. This is not what Wolpert has even considered at this point.

Dembski has not modeled a system in this way as of now. Thus, in the shortest way to respond to the "postive case" Dembski has laid forth, it is simply a strawman representation of biological evolution. Modeling real evolution in simple mathematical terms is probably one of the most daunting and complicated of tasks. The NFL theorems do not take into account an algorithm that itself adapts with increased fitness of the player, and changes the landscape with each successful target. Wolpert explained this in his earliest response to Dembski, and published his findings that take this into account w.r.t. NFL theorems. This intrinsic flaw is continually overlooked and undermines the positive case for design completely.

To his credit, Dembski's formalistic abilities are not in question. That is, his ability to evaluate a given model, and perform the correct analysis, is A-ok. The problem is the relevance and correlation of this model to anything resembling reality. "Let the reader judge," as Dembski says in response.

The real problem here is that Dembski has never addressed the most substantial critiques of his work. As Mark Perakh points out in this extensively referenced article (a good place to start for an overview of the status of Dembski's claims:
When encountering critique of his work, Dembski is selective in choosing when to reply to his critics and when to ignore their critique. His preferred targets for replies are those critics who do not boast comparable long lists of formal credentials – this enables him to contemptuously dismiss the critical comments by pointing to the alleged lack of qualification of his opponents while avoiding answering the essence of their critical remarks. (See, for example, Dembski’s replies to some of his opponents [3]) This type of behavior provides certain hints at Dembski’s overriding quest for winning debate at anycost rather than striving to arrive at the truth. For example, in his book No Free Lunch [4] Dembski devoted many pages to a misuse of Wolpert and Macready’s No Free Lunch (NFL) theorems [5]. (Some early critique of Dembski’s interpretation of the NFL theorems appeared already in [6 a, b]. A detailed analysis of Dembski’s misuse of the NFL theorems is given, in particular, in [6 c].)

Dembski’s faulty interpretation of the NFL theorems was strongly criticized by Richard Wein [7] and by David Wolpert, the originator of these theorems [2]. Dembski spared no effort in rebutting Wein’s critique, devoting to it two lengthy essays. [3] However, he did not utter a single word in regard to Wolpert’s critique. It is not hard to see why. Wein, as Dembski points out, has only a bachelor’s degree in statistics – and Dembski uses this irrelevant factoid to deflect Wein’s well substantiated criticism. He does not, though, really answer the essence of Wein’s comments and resorts instead to ad hominem remarks and a contemptuous tone. He can’t do the same with Wolpert who enjoys a sterling reputation as a brilliant mathematician and who is obviously much superior to Dembski in the understanding of the NFL theorems of which he is a coauthor.

Dembski pretends that Wolpert’s critique does not exist.

Dembski has behaved similarly in a number of other situations. For example, the extensive index in his latest book The Design Revolution: Answering the Toughest Questions About Intelligent Design [8] completely omits the names of most of the prominent critics of Dembski’s ideas.

We don’t see in that index the following names:
Rich Baldwin, Eli Chiprout, Taner Edis, Ellery Eels, Branden Fitelson, Philip Kitcher, Peter Milne, Massimo Pigliucci, Del Ratzsch, Jeff Shallit, Niall Shanks, Jordan H. Sobel, Jason Rosenhouse, Christopher Stephenson, Richard Wein, and Matt Young.

All these writers have analyzed in detail Dembski’s literary output and demonstrated multiple errors, fallacious concepts and inconsistencies which are a trademark of his prolific production. (I have not mentioned myself in this list although I have extensively criticized Dembski both in web postings [9] and in print [10]; he never uttered a single word in response to my critique, while it is known for fact that he is familiar with my critique; the above list shows that I am in good company.)

Thomas D. Schneider, another strong critic of Dembski’s ideas, is mentioned in the index of [8] but the extent of the reference is as follows:
"Evolutionary biologists regularly claim to obtain specified complexity for free or from scratch. (Richard Dawkins and Thomas Schneider are some of the worst offenders in this regard.)"
Contrary to the subtitle of Dembski’s book [8], this reference can hardly be construed as an answer to Schneider’s questions. Essentially, all the listed writers have asked Dembski a number of questions regarding his concepts. The absence of any replies to the listed authors makes the title of Dembski’s new book [8], sound like a parody. It should have properly been titled The Design Revolution: Answering the Toughest Questions About Intelligent Design. Of course we already know that Dembski is a stubborn purveyor of half-baked ideas [10]. Is he also of the opinion that selectivity in choosing when to respond to opponents and when to pretend they do not exist is compatible with intellectual honesty? [edit: Check out the article for the references: http://www.talkdesign.org/people/mperakh/perakh_ddq.pdf]
Welsley R. Elsberry has an impressive set of links (left sidebar) on Dembski's work and its ramifications. Mark Chu-Carroll has an entire section examining some of Dembski's work, Richard Wein has written extensively about the problems with it, also TalkDesign, as does TalkOrigins, on the NFL and his work in information detection, for more reading. Another awesome resource is the PandasThumb IDC archive. Furthermore, this post at the PT lays out some good refs on the background arguments of Dembski involving complexity via RM/NS:
  1. Peter Schuster, How does complexity arise in evolution? Complexity, 2:22-30 (1996)
  2. Christoph Adami, Charles Ofria, and Travis C. Collier Evolution of biological complexity, PNAS | April 25, 2000 | vol. 97 | no. 9 | 4463-4468
  3. Lenski RE, Ofria C, Pennock RT, and Adami C, The Evolutionary Origin of Complex Features Nature, 423:139-144 (2003).
  4. Tom Schneider, Rebuttal to William A. Dembski’s Posting and to His Book “No Free Lunch”
  5. Tom Schneider ev: Evolution of Biological Information Nucleic Acids Res, 28:14, 2794-2799, 2000
_____________________________

As we discussed at our freethought group's 11/9 meeting, there are actually some good philosophical arguments in ID. Teleological and anthropic principle-type arguments certainly aren't invalid, although the veracity of their premises is, of course, crucial. However, they simply aren't science, or scientific arguments, without meeting some pretty strict criteria, and without going through the process of peer review, experiment, and eventual scientific consensus. As of now, they are completely philosophical in nature. That only means that ID still has to pass the standard hurdles before including itself into the scientific community as a valid idea to teach in science classes.

As Prof. Joe Meert pointed out at that night's meeting, Einstein didn't take out ads in newspapers asking people to write their congressional representatives to get relativity included in high school curricula. Einstein didn't try to get a relativity-sympathetic school board voted in, and include his ideas in high school textbooks. Einstein wanted legitimacy among his scientific peers and to establish his ideas via empiricism and the method of science, and then, he knew, getting into textbooks would follow. Those sorts of political tactics by IDCists are what undermines their claim to legitimacy in the scientific community. They want their philosophical verbiage smuggled into science classrooms since they can't get a single scientific argument going for them.

Conclusion: irreducible complexity fails as a critique against evolution, as the proposed systems have viable, published explanations for their evolution; and Dembski's work in design detection fails as it is a critique against quasi-evolution -- a strawman, which does not even intersect with the robust work done by geneticists and computer scientists in mathematically modeling evolutionary theory.

The good philosophical ideas intrinsic within teleological arguments is lost amidst the "culture war" that IDC's are waging against evolution and materialism, in their own words. Hopefully, once the dust settles, and cooler heads prevail, some of their ideas can be incorporated into places that they belong (history, philosophy, etc.), and continue to be excluded from the science classroom until they become scientific...hopefully.
________________
Technorati tags: , ,