William Dembski is one of the leading thinkers behind the Intelligent Design movement (ID). He has written both scholarly texts, such as The Design Inference, and popular books, such as Understanding Intelligent Design (which we co-wrote in 2008).
Although he has recently shifted his focus from ID to other projects, Dembski has recently released a book called Introduction to Evolutionary Informatics, which is co-written with Robert Marks and Winston Ewert. This may be his final book-length contribution to the discussion over Darwinism and intelligent design.
Essentially, the goal of the book is to demonstrate two things. First, that computer simulations of Darwinian evolution fail to show how blind, unguided processes can generate increased fitness and information. And second, that evolutionary computer simulations only function if information is added along the way, which is the very thing not available in nature. Thus, ironically, according to the authors, if the simulations do work, they (unwittingly) support intelligent design:
“If these models do indeed capture the Darwinian process, then we must conclude that evolution is guided by an intelligence. Without the application of this intelligence, evolutionary models simply do not work.”
How can Dembski et al. make such claims? They carefully examine various popular evolutionary computer programs, such as EV and Avida. In each case, Dembski et al argue that the creators include information sources in their programs. And they do so in (at least) three ways:
First, they include a “man-in-the-loop.” In other words, the software programs are designed, tested, adjusted, and further tested until they receive the desired results. Dembski et al specifically cite the ID critic David Thomas who challenged ID supporters to find the information source in his program, which he claimed toppled ID. And so they did. In fact, at one point in the code Thomas specifically wrote, “over-ride!!!” which shows that the program requires a guiding hand and does not mirror the blind processes of evolution.
Second, evolutionary computer programs include “stair step active information.” In his book Climbing Mount Improbable, Dawkins argues that evolution can occur in an incremental, step-by-step process. As a result, for the evolution of a worm to a whale, for instance, each intermediate step must be functional. This raises challenges in nature, but computer simulations are often less restricted. The Avida program only works because the creators have carefully designed a “staircase” where each step is functional along the way. According to Dembski, without this designed process, the simulation fails.
Third, the mutation rates of various programs are carefully fine tuned for success. In nature, mutations are far more often damaging than beneficial. Here is what Dembski et al. write:
"Cornell University geneticist John C. Sanford documents the chance of a beneficial mutation in a complicated organism is essentially zero and that mutation has a greater chance of extinguishing a species than of advancing it. If mutation is generally beneficial in an evolutionary program, there must be a resident source of information that guides mutation away from being a detriment."
Once again, information input is required for the computer simulation to work. To be sure, they don’t argue that Darwinian processes are logically impossible. Rather, given the creative capacity of Darwinian processes and the available resources of our entire universe, evolution is functionally impossible. In fact, according to the authors, blind processes are so limited that the information necessary for the Gettysburg Address alone would require 10792 universes!
Introduction to Evolutionary Informatics is an introductory text, but it is certainly not meant for the mathematical novice. The authors claim that it is accessible, but it definitely requires some mathematical training to grasp its depths. I am not a mathematician, and so I will leave the particulars of the book to the experts.
But if Dembski et al. are right (and my suspicion is that they are), then evolutionary computer simulations fail to provide evidence for Darwinism. In fact, they may even provide indirect evidence for intelligent design.
 William Dembski, Robert Marks II, and Winston Ewert, Introduction to Evolutionary Informatics (Hackensack, NJ: World Scientific, 2017), 1.