Saturday, August 31, 2013

Random Thoughts on Schneider vs. Dembski: What Kinds of Information are Real?

Still thinking about Tom Schneider's ev simulation and these are the salient points:
  • What engineers mean by the "information" inherent in a complex, interactive machine is something different from Shannon information, something not necessarily synonymous to Kolmogorov information but certainly more closely related to it. 
  • Schneider thinks that Shannon information is the only kind of information there is--the only definition of information worth considering.  The references to Kolmogorov information that Dembski makes are lost on him.  Schneider is obviously very smart, very well versed in the sort of information he is concerned with, but like most that concern themselves with biology, he is oblivious to anything beyond the most mechanistic conceptions of information.
  • Jeffrey Shallit, by contrast, is interested in Kolmogorov information, but in his world, giving you a copy of a phone book you already possess has given you more information (a few bits more, to be precise) and a lot more information if the second book has a lot of the letters and numbers scrambled randomly.  Schneider would say that a completely randomized phone book would have zero information. There is something completely missing from both these conceptions.  
  • Hubert Yockey would have said that they were both confused.  Schneider would say that Yockey and Shallit were confused. Go figure.  About the only one who comes close to recognizing the fundamental problem is Gregory Chaitin, though he believes in Darwinism as well.  The only thing that they have in common is that they all seem to have very different justifications for dismissing Dembski. (I'm not certain that Chaitin dismisses Dembski, but he finds the informal arguments for evolution persuasive even though he thinks there is no real mathematical theory explaining evolution.)   
  • Schneider doesn't like that Dembski says that Schneider has claimed to produce what Dembski calls complex specified information (CSI) because he doesn't like that term. It does seem (need references) that Schneider has claimed to have literally "created information" and seems to have this in mind as support for the Cambrian Explosion not being really that strange.   
  • What Schneider has in mind by creating information is that natural selection could allow a co-evolution between varying binding sites and the binding molecule, so that both are refined.  To the degree that the emerging pattern is specific, I'm not sure that it's complex; to the degree that it is complex.   There is not much that is holistic about the problem--each nucleotide site can be refined more or less independently of the others--which makes it a very different problem than adapting an enzyme with highly cooperative sections.   
Generally, it seems that Schneider's implicit argument is:

  • Leading ID advocates claim that the Cambrian explosion is not explained by Darwinian models because in a evolutionarily short amount of time, the majority of body plan innovations, with their novel and unique architectures and developmental programs, all arose.  The body plans indication that enormous amounts of prescriptive information for development and hundreds of supporting protein specifications were developed for each arising phylum.
  • The ev simulation shows the specificity of binding sites being converged on to produce specific patterns that are informational and might be considered CSI. (Schneider disdains the concept of CSI so much that he doesn't thoughtfully consider how it might relate--or not--to Shannon information, so it's curious.  He is apparently incensed at Dembski for putting words in his mouth and then seems to argue that this is actually what he means by "creat[ing] information.")
  • Specificity of site recognition is an emergent kind of information, and phylum body plans, also being information, can presumably emerge as well.

     ____________________________________________________
         Ergo, the Cambrian Explosion is not hard to account for at all.

Now if we could just make a genetic algorithm to generate the specs for novel, working airplanes.... or for novel, working inventions in general...  As Ray Kurzweil might say, the singularity is near...

No comments:

Post a Comment