[Editor's Note: I've reworded a couple of the original descriptions in here to make my intention more clear. The ones I edited made me cringe a bit, so I hope the edited versions are more elaborate and incisive.]
When looking through AiG's archives for stuff that I could shred apart for PZ's blog carnival, I characteristically looked for the computer science and mathematics related drivel that is typical of all creationists. And lo and behold, I found it. The first piece deals with Richard Dawkins famous "weasel" word-experiment, and the second deals with genetic algorithms and why they allegedly fail to show that evolution works. A quick rundown of the former: Dawkins experiment sets up a space of 29 character spaces that go through random iterations that produce either a letter in the English alphabet or a space. Through the selection of useful results on individual rings, it eventually produces the sentence "METHINKS IT IS LIKE A WEASEL".
Of course, the process isn't completely isomorphic to the Darwinian process biologists talk about when they invoke evolution. In almost all cases the selective pressure of evolution only operates by winnowing out less beneficial mutations in populations, insofar as they inhibit reproductive success. But the word-selection algorithm Dawkins expounds gives a good indication of how such winnowing helps to find meaningful outcomes among vast ranges of non-meaningful syntax. But that is apparently lost on the AiG crowd, which resorts to inane arguments like this:
Now, progress in reaching the target sentence would be made for any outcome where x > 0. The probability of something useful happening upon spinning all 28 rings is the sum of prob(x = 1) + prob(x = 2) + … + prob(x = 28). This adds up to 0.6524. In other words, the chances are low, only 0.3476, that after a single trial nothing useful happens. And if so, it does not matter, we just try again! (Note that Dawkins’ own ‘random’ initial sequence shows two letters and one space already correctly matched up).
Seeing as though they resort to the argument that there is a "low probability that nothing useful happens", we have to step back for a moment and ask ourselves "are we still arguing over the same subject?" I know of no evolution-supporter who makes the arguments that evolution operates by producing no useful solutions. No one argues, as evolution-supporters have pointed out many times before, that evolution is a random process! It is a highly selective process that culls less useful traits, much like Dawkins word-experiment. Thus this arguments is meaningless:
Therefore, the entire target sentence does not fail to be matched within a small number of trials. How could this prove that life and complex organs could arise by chance?
And of course, it should be noted that AiG is giving a misleading calculation of the space of possible outcomes in comparison to the desired solution. They only give the binomial outcome of one ring producing a successful result, and don't give you the probability of getting the targeted solution across the entire range of rings. When calculating the probability of collective independent events being observed concurrently, you multiply their binomial outcomes together. As an illustrative example, the probability of getting any result from rolling a die is 1/6, but the probability of getting any result from two dice would 1/6 x 1/6, or rather 1/36. It's easy to deduce the probability calcuation for Dawkins word-experiment from here. You simply take the binomial outcomes (1/27) and multiply them together 29 times, which produces a result of approximately 1/3.232 x 1041. In other words, prohibitively low.
This is very illustrative of how creationists like AiG play games with probability. If you simply smash together independent events without taking into account underlying mechanisms that narrow the probability of finding a successful result, you can produce very misleading numbers that appear to probibit evolution from happening.
But now to move on to AiG's treatment of Genetic Algorithms. Here are a few things Don Batten has to show that they in no way show that evolution via selection is efficacious. They do not, of course, show anything of the sort. But they do put his ignorance and dishonesty on display for the rest of the world to see:
A ‘trait’ can only be quantitative so that any move towards the objective can be selected for. Many biological traits are qualitative—it either works or it does not, so there is no step-wise means of getting from no function to the function.
Wrongo, try again. While many biological traits either work or don't work, the same can be true of candidate solutions operated on by genetic algorithms, and the converse is true of both as well. Consider looking at the candidate solutions for a minimum cost-cycle traveling salesman problem; they always operate on connected graphs (making them all operational), but candidate solutions are evaluated on how fast they can be traversed. On the other hand, a 3-CNF Boolean SAT either evaluates to TRUE or does not, putting it squarely in the former camp. Genetic algorithms can be used either way.
A single trait is selected for, whereas any living thing is multidimensional. A GA will not work with three or four different objectives, or I dare say even just two. A GA does not test for survival; it tests for only a single trait.
Wrong again. GA's can and have been designed with parallelism and multicriterial decision making, so this is simply false on it's face.
Something always survives to carry on the process. There is no rule in evolution that says that some organism(s) in the evolving population will remain viable no matter what mutations occur.
Bzzzzzt. Wrong again, thank you for playing. In fact, there is a rule in evolution that makes it far more likely that new candidate solutions will survive into the next generation (albeit not guaranteeing it's continued reproductive advantage, which is also true for GA's). One of the main facets in Darwinian evolution is retention, which provides the base material for selection to act upon.
Perfect selection (selection coefficient, s = 1.0) is often applied so that in each generation only the best survives to ‘reproduce’ to produce the next generation. In the real world, selection coefficients of 0.01 or less are considered realistic, in which case it would take many generations for an information-adding mutation to permeate through a population.
Which probably has something to do with the quickness with which genetic algorithms have to operate in comparison to their biological counterparts. The use of a selection coefficient isn't what is under debate, but the efficacy of selection itself in finding viable solutions.
In real organisms, mutations occur throughout the genome, not just in a gene or section that specifies a given trait. This means that all the deleterious changes to other traits have to be eliminated along with selecting for the rare desirable changes in the trait being selected for. This is ignored in GAs. With genetic algorithms, the program itself is protected from mutations; only target sequences are mutated. Indeed, if it were not quarantined from mutations, the program would very quickly crash. However, the reproduction machinery of an organism is not protected from mutations.
Wrong again. This is a case of equivocation and sloppy use of analogy. That is, the program is analogized to an individual genome. A much more accurate analogy would be to the environment, which is where the populations of candidate solutions are tested against one another for viability and efficiency.
There is no problem of irreducible complexity with GAs (see Behe’s Darwin’s Black Box). Many biological traits require many different components to be present, functioning together, for the trait to exist at all (e.g. protein synthesis, DNA replication, reproduction of a cell, blood clotting, every metabolic pathway, etc.).
This is false, it is conclusively established that irreducible complexity is evolvable via Darwinian mechanisms....thanks to work with genetic algorithms.
The outcome in a GA is ‘pre-ordained’. Evolution is by definition purposeless, so no computer program that has a pre-determined goal can simulate it—period.
Candidate solutions in a GA are selected on certain criteria, and the same is true of biological evolution (reproductive success). "Purpose" is very loosely defined in this case as an end, which biological evolution clearly has (though not a termination point, to state the obvious).
|