Intelligent Design Creationists Still Abuse NFL Theorems

Creationists rarely come up with any new arguments. Rather, they keep repeating the same flawed assertions that have been disproved thousands of times in the past. Sometimes, however, they attempt to reinvent themselves. Not by discovering evidence or presenting new arguments, but by dressing up previous arguments in a cheap tuxedo. Claims about “what use is half a wing or half an eye?” gets changed to “what use is half a flagella?”, claims about evolution somehow contradicting the second law of thermodynamics gets replaced by appeals to an imaginary conservation law about information and so on. Another common creationist trope is asserting that evolution is just “random chance”. Since random chance cannot produce complex adaptations, creationists argue that there has to be an intelligent designer behind life. In reality, selection is a non-random process and it is the generation of genetic variation that is essentially random. Because this creationist trope can readily be debunked, they had to throw out some smokescreens in an effort to rehabilitate this approach.

That smokescreen is the abuse of the so-called No Free Lunch (NFL) theorems. Simplified, intelligent design creationists claim that these show that no evolutionary algorithm can outperform a random search, so therefore, evolution is not better than “random chance” and since “random chance” cannot produce complex adaptations, evolution cannot do it either. However, this is based on a key misunderstanding of the NFL theorems. Mark Perakh (2004, p. 102) explains:

The NFL theorems establish that performance of all algorithms is the same if averaged over all possible fitness functions. Dembski illegitimately applies this results to the algorithms’ performance on specific fitness functions where different algorithms can (and do) perform very differently. Dembski’s assertion that no evolutionary algorithm can outperform a random search because of the NFL theorems and that therefore Darwinian evolution is impossible is absurd. The NFL theorems in no way prohibit Darwinian evolution.

In other words, one evolutionary algorithms cannot outperform another averaged over all possible fitness landscapes. However, real-world evolution occurs on a specific subset of fitness landscapes not averaged over all theoretically possible fitness landscapes.

Even David Wolpert, one of the discoverer of the original NFL theorems, rejects Dembski’s false characterization:

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.

The fitness landscape is thus not independent of the evolutionary algorithm, and the NFL theorems do not apply.

Recently, the pseudonym scordova wrote a post on the intelligent design creationist blog Uncommon Descent (UD) about the NFL theorems and Dawkins’ Weasel algorithm (another classic creationist obsession). Because the UD post shows (1) that intelligent design creationists, like their ideological predecessors, continue to appeal to claims that have long since been debunked and (2) that there is a large overlap between scientific creationists and intelligent design creationists in terms of what kind of arguments they use, let us go through it point-by-point.

Evolution of proteins is not like random changes of a password query

Creationists of the 1950s and 1960s liked to use books or other kind of texts as analogies for genomes. An adaptive evolutionary process would, in their view require one human-readable text to “randomly change” to another human-readable text. Since this seems very implausible, they argue, so is evolution. The fallacy here is obvious: nucleotide and amino acid sequences are not analogous to written words in a book. This is because most words lose their specific meaning after modifications, where as a nucleotide sequence can tolerate some changes without affecting amino acid sequence (called synonymous mutations) and amino acid sequences can tolerate some changes without affecting protein function (e. g. the replacement of one amino acid by another of the similar size and chemical properties). The word “can” might “randomly change” to “cax”, but that is not a word that would work in the textual context. However, the nucleotide sequence “CTT” codes for the amino acid leucine, but if this sequence changes to “CTC”, “CTA” or “CTG” it still encodes leucine. This means that the genome can tolerate changes and still function in more or less the same way, whereas written text cannot.

Scordova claims that the evolution of proteins is “akin to taking a functional password for one account and presuming we could evolve it in steps to become a functional password for another account”. However, this is false because computer systems that require passwords do not allow any mismatches, whereas the genetic code allows some mismatches and still produce the same amino acid sequence and the protein allows some amino acid substitutions without major changes in protein function.

Dawkins’ Weasel algorithm shows the difference between chance and cumulative selection

After gotten extremely tired of creationists trying to make evolution look like “random chance”, Oxford biologist Richard Dawkins wrote a simple algorithm that compared the efficacy of chance + selection (cumulative selection) to chance alone. He did this by subjecting an arbitrary string of texts to “mutations” i.e. pseudorandom modifications and then keeping any letter that matched a sample sentence (in this case the Shakespeare phrase “Me thinks it is like a weasel) was kept. The cumulative selection made the phrase in a little over 40 “generations”, while the chance alone method would take a very long time. The goal was not to model evolution as it occurs in nature, only to demonstrate the power of cumulative selection over the creationist straw man of “random chance”.

It is a very simple argument: cumulative selection is nothing like “random chance”. Yet creationists have become completely obsessed with this argument. Scordova claims that the algorithm “is a misleading picture of how natural selection in the wild really works when trying to solve problems like protein evolution”, clearly misunderstanding the purpose of the example, which was to show that cumulative selection can accomplish things that “blind chance” cannot. This statement was true then and it is true now, even for protein evolution. Yet this news has apparently not reached scordova when he falsely claims that: “you have a functioning password that works for your account, it may even share extreme similarities to other passwords that people have for their accounts. Does that fact give you a better chance of solving their passwords over blind luck? No.”

Abusing “No Free Lunch” theorems

As if these trivial conceptual errors were not enough, Scordova moves on to discussing NFL theorems:

No Free Lunch theorems are the formalization that shows that Darwinian search is no better than blind search for cases like solving passwords.

As we saw before, the relevant NFL theorem states that no search algorithm can outperform chance averaged over all possible fitness landscapes. However, this is not what happens during evolution as evolution occurs on very specific fitness landscapes and the fitness landscape is not independent from the search algorithm.

It never ceases to amaze me how creationists and other science denialists continue to repeat the same assertions over and over again. They are virtually immune to refutations. When they are forced to concede points, they act as if they are not retreating, just advancing in a different direction. Sometimes, though, you catch them admitting that their position is either false or irrelevant for evolution. Here is Dembski talking about conservation of “information” and bacterial evolution were he admits that evolution can produce complex adaptations:

So where’s the problem for evolution in all this? Granted, the selection regime here is a case of artificial selection — the experimenter is carefully controlling the bacterial environment, deciding which bacteria get to live or die. But nature seems quite capable of doing something similar. Nylon, for instance, is a synthetic product invented by humans in 1935, and thus was absent from bacteria for most of their history. And yet, bacteria have evolved the ability to digest nylon by developing the enzyme nylonase. Yes, these bacteria are gaining new information, but they are gaining it from their environments, environments that, presumably, need not be subject to intelligent guidance. No experimenter, applying artificial selection, for instance, set out to produce nylonase.

Although Dembski goes on and attempt to level his NFL theorem misunderstandings against this example, he clearly accepts that natural processes can lead to complex adaptations. He keeps insisting that “information” is conserved, but if organisms that exploit “information” from the environment, then the intelligent design creationist ramblings about conservation of information as an obstacle to Darwinian evolution falls apart.

References:

Perakh, M. (2004). Unintelligent Design. New York: Prometheus Books.