The author, Darko Roglic, presents the intriguing idea that biological evolution uses concepts that go beyond traditional algorithms. Roglic argues that super-recursive algorithms are necessary and sufficient to explain biological processes. This is definitely an interesting idea. Similar ideas have been presented in the past by other authors. They are usually followed by strong rebuttals and a general descent into ad hominem mudslinging in both directions.

I started reading this article with the hope of finding something new, different, and more interesting, but my hopes were dashed by the start of page 3, when I read such zingers as “the notion of mutation has no meaning for programming theory,” based on a citation from a biology-centric article. I do not have the cited article, so cannot comment about its general value. However, this line, perhaps taken out of context, appears to claim that decades of genetic programming lie outside the body of legitimate programming. Huh?

But perhaps I misunderstood. So, I continued reading.

Next, the author cites a well-respected book on epigenetics, this time one that I’ve read, and tries to draw computational consequences from the epigenetic understanding that evolution occurs on multiple channels, not just the DNA sequence. Again, I’m not very impressed by his argument.

Finally, the author moves on to his key point. He claims that “super-recursive” algorithms are more powerful than Turing machines and therefore: 1) Turing machines (and traditional algorithms) are not powerful enough to emulate biological evolution, and 2) super-recursive algorithms are sufficiently different that they can emulate biological evolution.

I won’t address the first of these two points. While I do believe that Turing machines are powerful enough to represent all elements of biological evolution, I readily concede that this will not be proven until someone builds an effective model. The second point is more provocative. Roughly speaking, a super-recursive algorithm is one that modifies its behavior based on the results of previous runs of the program. Granted, this is beyond the classical definition of a pure Turing machine, but it is well within the capabilities of every physical computer built since the 1940s. So, at best, this article argues that the pure model described by Alan Turing 80 years ago is not quite powerful enough to describe evolutionary processes.

Maybe I’m just too far outside the philosophic community fascinated by this, but this article just does not resonate with me at all.