The C-Files: William Dembski
Our Favorite William Dembski
Intelligent Design is "An argument from ignorance"...
In an article by Richard John Neuhaus from First Things 121, March 2002, Neuhaus writes "With respect to the origin and complex development of life forms, clear thinking begins with recognizing what we do not know. Dembski puts it nicely: 'An argument from ignorance is still better than a pipe dream in which youre deluding yourself. Im at least admitting to ignorance as opposed to pretending that youve solved the problem when you havent.'
Dembski talks on "Darwin's Unpaid Debt" -- but it's Dembski's note that's in arrears...
[Adapted from NMSR Reports December, 2001, Vol. 7, No. 12]
by Dave Thomas : nmsrdaveATswcp.com (Help fight SPAM! Please replace the AT with an @ )
By many accounts, William Demsbki is the mathematical genius of the Intelligent Design (ID) "Movement." Here are his vita as given by Access Research Network (ARN), a prominent ID web site (http://www.arn.org): "A mathematician and a philosopher, William A. Dembski is associate research professor in the conceptual foundations of science at Baylor University and a senior fellow with Discovery Institute's Center for the Renewal of Science and Culture in Seattle. Dr. Dembski previously taught at Northwestern University, the University of Notre Dame, and the University of Dallas. He has done postdoctoral work in mathematics at MIT, in physics at the University of Chicago, and in computer science at Princeton University. A graduate of the University of Illinois at Chicago where he earned a B.A. in psychology, an M.S. in statistics, and a Ph.D. in philosophy, he also received a doctorate in mathematics from the University of Chicago in 1988 and a master of divinity degree from Princeton Theological Seminary in 1996. He has held National Science Foundation graduate and postdoctoral fellowships. His articles have appeared in mathematics journals such as the Journal of Theoretical Probability, in philosophy journals such as Nous, in theology journals such as Epiphany, and in journals such as Perspectives that deal with science/faith interaction. He has published three books. In The Design Inference: Eliminating Chance Through Small Probabilities (Cambridge University Press, 1998), he examines the design argument in a post-Darwinian context and analyzes the connections linking chance, probability, and intelligent causation. His most recent book is Intelligent Design: The Bridge Between Science and Theology, which appeared November 1999 with InterVarsity Press." Dembski has a new book coming out next year called "No Free Lunch."
Dembski spoke at the Continuing Education Center on November 12th, 2001, in a talk titled "Darwin's Unpaid Debt." He started by talking about cause and effect, quoting John Stuart Mill on needing different causes to find different effects. He implied evolution doesn't work because the causes -- replicability, heritability, random variation, and natural selection -- don't always produce the same effects. To show this, Dembski mentioned cases where all these elements were on hand, and yet "nothing happened," obviously confusing "if cause, then HAS to happen" with "if cause, then CAN happen."
Dembski admitted that it's hard to prove something [evolution] could not possibly happen, especially if all you had was an exhaustive search. So, he said, you need to find "proscriptive generalizations" or "invariants." Dembski mentioned perpetual motion machines as an example - we don't have to test every candidate machine [exhaustive search] because the laws of thermodynamics show such devices are impossible [proscriptive generalization]. He showed a slide of a chessboard with a pair of missing adjacent squares; clearly the missing squares must include one black and one white square, and so you wouldn't even need to test candidate tiles with two adjacent white squares. In this case, the invariant (number of black squares minus number of white squares equals zero) is clearly violated by adding a tile with two white squares.
Dembski discussed Michael Behe and his "irreducible complexity" approach to ID, and mentioned that Behe is coming to Albuquerque next March. Amazingly, Dembski described several effective critiques of Behe's ideas, including scaffolding, co-aptation, redundant complexity, ignorance of the scientific literature, appeals to ignorance, incremental indispensability, and reducible complexity. But Dembski blew off all the criticisms, as if simply mentioning their existence effectively counters them. Here's an example: Dembski mentioned scaffolding and the Roman arch example, which Massimo Pigliucci happened to have presented to NMSR at his talk on August 11th. Here, a mound is built, then stones are place on top to make the arch; once the last stone is in place, the mound is removed, leaving the arch in apparent "irreducible complexity" (take away any stone and the whole thing falls down). Dembski's counter to this example was to claim you can't have "delayed gratification," i.e. that in living systems, you would need the arch to work right away, and wouldn't possibly have a "mound" stage.
Dembski described his addition to Behe's "irreducible complexity" required to tighten it up: it's a system where removal of a single part destroys the original function, removal of multiple parts destroys the original function, the system comprises numerous and diverse parts, and (Dembski's innovation) the system is minimally complex for the required functionality. He calls this "Irreducible Complexity 2.0 (read two-point-oh), and claims this is the "Darwin-stopping invariant." In five to ten years, Dembski said, "Darwinism" will stand alongside failed theories like alchemy and perpetual motion. He said Darwin's unpaid debt is an unpayable debt, and that leaves only Design.
It's important to note that Dembski argument at this can sound quite compelling, especially to non-scientific listeners who just want assurance that science supports their religion. But it's just low-density popcorn; it looks filling, but there's no substance. One cannot live on cotton candy alone, sweet or no. All that Dembski has here is his unfounded claim that no biological feature can ever exist without a single, specific function, and that co-opting just doesn't happen in biology; by analogy, feathers must have always been feathers for flight, and could not have once been used only for warmth. Cotton candy, indeed.
Demsbki went on to describe his "Explanatory Filter," which he claims can robustly detect design. First, you decide if a process can have a natural explanation - is it the result of a natural law? If it's not, then you consider chance - can the process be the result of random events? If the process is not random, the third element of the filter is "specified complexity" - does it provide a definite function? Dembski argued that specified complexity is not a necessary condition for design - a Designer could, after all, mimic natural processes - but that it was a sufficient condition. If you have specified complexity, he said, you've got Design.
Dembski then discussed genetic algorithms, and even showed the result of a remarkable study in which herds of wire antenna models were bred, and those that provided more uniform antenna patterns were bred some more. A simple antenna - a straight piece of wire, for example - has a non-uniform radiation pattern, looking much like a donut, uniform on the sides, and vanishing along both ends of the wire. Dembski showed a curious antenna that looped this way and that that was developed in a genetic algorithm, and which actually produces a very uniform antenna pattern. Clearly, there is specified complexity in the loops and twists of the antenna wire, but Dembski says this was snuck in, front-loaded, in the fitness function - the requirement that a uniform pattern is preferred.
This is wrong. I have written genetic algorithms, and discussed these with Dembski after the Tuesday debate. I presented him with results of my algorithms that bred solutions to Steiner's problem, involving efficient networks, and some of which displayed striking geometric patterns - design. Dembski said I had simply front-loaded that design into my fitness function, which took the simple form "shorter is better, as long as nodes are connected."
Dembski is obviously confusing the environment (the fitness function) with the adaptation (the twists of the antenna, or the geometry of a network solution). Design is an attribute of the adaptation, not of the environment. In biology, the fitness function has a simple form indeed: given the current environment, does the organism live long enough to have children? William Dembski does not understand this simple concept. The mathematical foundation of the Intelligent Design Movement is a facade - it rests on confusion between the environment, and adaptive responses to that environment.
In the question/answer period following the talk, I asked Dembski if the similar defects in the genetic makeup of apes and humans that make us the only creatures incapable of making our own ascorbic acid (vitamin C) weren't possible evidence of common ancestry, and Dembski astonished the creationists in the audience by admitting that, yes, it was possible that humans and apes share common ancestry. However, he interpreted the genetic flaw preventing production of vitamin C as "degradation," the opposite of innovation through evolution.
Stuart Kauffman dismantles Dembski's arguments
William Dembski debated Stuart Kauffman at UNM's Woodward Hall on Tuesday, November 13th. His presentation was more restrained for this academic setting; Alan Gishlick of NCSE, who saw both talks, described the Dembski at Monday's talk as "Bad Bill," and the Dembski who said nice-sounding things about skepticism and scientific inquiry at Tuesday's debate as "Good Bill." Dembski's two presentations did share some things in common, but he used more technical terms in Tuesday's appearance, such as "degrees of freedom of DNA."
Kauffman is a well-known member of the Santa Fe Institute, and author of several books on self-organization. He is currently at BiosGroup.
Kauffman began by summarizing the chapters Dembski sent him from Dembski's new book, "No Free Lunch" : (1) Specified complexity exists; (2) Natural Selection cannot explain it; (3) Evolutionary computing (genetic algorithm) uses fitness landscapes, and specification of fitness is really sneaking in the information.
Wrong, Kauffman said. Complexity can arise in nature through symmetry-breaking processes. If a pan of water is at uniform temperature, it's symmetric -- and boring. But apply heat non-uniformly to the bottom of the pan, and convection cells start to form, breaking the symmetry of the water in the pan, and producing interlocking hexagonal cells - designs. Kauffman gave other examples of such processes, and noted that non-equilibrium states can lead to localized complexity. "There is no difficulty in understanding that in our universe, things can get locally complex," he said.
Kauffman went on to discuss problems involving the origin of life itself. Although no one has yet gotten DNA to self-reproduce, workers have produced self-replicating proteins, and destroyed the notion that such replication requires templates. Kauffman gave an example of a ligation reaction, in which no molecule catalyzes itself, but in which pairs of molecules catalyze each other. And he described a delightful analogy with buttons and thread representing catalytic reactions. If you threw several buttons on the floor, and began connecting pairs of buttons with pieces of thread, soon you would have more than just several pairs of buttons tied together. As the number of threads increases, the pairs become entwined, and eventually all of the buttons are joined in one large clump. In the chemical analogy, chemical elements like monomers and dimers take the role of the buttons, and catalysts act like the threads, binding various pairs of simple monomers and dimers. With more catalysts, there arise more clusters, and eventually giant clusters - complicated macromolecules. The giant clusters act as catalysts for more, similar giant clusters. The problem with the small creationist probabilities of the origin of life (such as Hoyle's getting a jet plane out of a whirlwind in a junkyard) is that it's based on one specific system (the plane). In life, what's needed to get started is not one specific system, but any self-replicating system.
Kauffman mentioned how evolutionary computation is not just sneaking in design via a fitness function. In nature, it's the niche that sets the criteria for organisms competing in that niche; no intelligence is required. Organized self-complexity can get life rolling, and natural selection can do marvelous things once that has begun.
Many attendees agreed that Kauffman was the clear winner of the debate. While Dembski stuck to analogies and promises of details in future books, Kauffman cited real research, real results, and real science. It wasn't really a very fair fight. Stuart Kauffman will be speaking to NMSR on February 13th, 2002.
Read Phillip Johnson's Wedge Updates on these talks...
William Dembski's Home Page:
Wesley R. Elsberry's William Dembski Discussion Page (THOROUGH!)
William Dembski: "Intellectual Father #3"
of intelligent design theory.
NMSR Site Map