You may Have seen William Dembsk’s assertion, or repeated it, that he has proved that natural selection working on random chance is not an efficient way of generating mutations or that it’s less efficient than random chance alone. This somewhat nonsensical conclusion was supposedly proved by a computer program written by Dembski and Robert J. Marks. They put up an essay, “The Unacknowledged Costs of Evolution,” criticizing an evolutionary program called ev, on an unofficial web page. Unfortunately, obvious errors in basic computing make their results invalid. The errors were such that the program would never find a solution, such as filling a table with 1’s and then randomly changing some of the cells to 1’s and waiting for some zeroes to appear. With programming skills like those, they would wait forever. And then they used the wrong size of matrix, compounding their error.
Wesley R. Elsberry has the details in an article called “Unacknowledged Errors in ‘The Unacknowledged Costs’” over on the Panda’s Thumb. It covers who was able to find the errors and why, what they wrote to the authors, what they did or didn’t do in response, what has happened when other errors were pointed out, and so on. In the badinage of the comments the standard size of a Dembski error was discussed and 65 orders of magnitude was calculated. That’s 1 followed by 65 zeroes. Someone else pointed out that creationists, in taking the age of the earth to be 6,000 years instead of 14 billion, were only 63 orders of magnitude off target–a centi-Dembski!