Sunday, December 7, 2014


Has it not crossed Jeffrey Shallit's mind that Phillip Johnson is aware of those examples but doesn't find the case for the intermediate or transitional nature of those fossils very convincing?  -- Jonathan M. on Shallit's claim that Phillip Johnson lied about the fossil record 

Some notions of information are more closely tied to the Shannon notion of information (e.g. Thomas Schneider and Richard Dawkins) and tend not to wander much further. Much more closely related to some notion of meaning/function has been the Kolmogorov notion of information, also known as Kolmogorov complexity.  Complexity and information tend to be closely related notions; both are related to probability, and probability has a combinatorial nature.  The idea of mutual information is closely related to that of conditional probability.  A complex thing generally needs more information to describe it (or information and time, but that will have to wait), while a much simpler thing generally required much less information.  Highly improbable things tend to have many levels of contingencies, and convoluted  histories/developments. Optimal codes match less probable events to more bits of information.  An optimal code will tend to not only maximize the Shannon complexity of the messages but to also make the messages more compressed (maximizing K complexity per message length).

Jeffrey Shallit has become something of a public figure, not for math or computer science per se, but for being a "science defender" of the quixotic "anticreationist" variety. (Note: I have noticed Shallit also uses the term "anticreationist," though not in the ironic sense that I do.)  He has made a name for himself ridiculing ideas and the people who promote them.  He has complained in the past that 'specified complexity' would be better off called 'specified improbability.'  His and Elsberry's long critique of the idea of  'complex specified information' scoffs at the difficulty in reducing CSI neatly to one of the two kinds of information . (Everything else is "creationist information.") Winston Ewert's notion of algorithmic specified complexity is very closely related to Dembski's notions of CSI and is itself a specie of randomness deficiency, combining notions of K-complexity and probability.

Something that is ubiquitous among complexity theorists and attempts to quantify 'biological information' is notably absent from Shallit's writings on the topic of complexity.  It's what Jim Crutchfield calls 'humpology'.  Most of the attempts over the last 20 years have tried to distinguish 'useful' information/complexity from both highly ordered and highly disordered configurations.  Kolmogorov information considers highly disordered configurations to be provide more information, but most complexity theory tries to distinguish between useful information (what engineers typically mean by 'information': semantically "rich" information that carries some deep patterns) and the information maximized by total randomness.  This fundamental concern in complexity theory doesn't seem to concern Shallit at all (which is odd given reasons that we'll get to).

Shallit has dubbed specified complexity (SC) or complex specified information (CSI) 'creationist information' to contrast it with the only sort of information he uses: algorithmic information or "Kolmogorov complexity."  He poses the following questions for "creationists" (i.e. anybody friendly to ID), where K represents K-complexity:
   1)  Can K(xx) > K(x)
   2)  Can K(x1y) > K(x0y)?   (restated as equivalent binary string problem)
   3)  Can K(xz) > K(xyz)?
   4)  Can K(perm(x)) > K(x)?
   5)  Can K(perm(xy)) > K(xy)?  (the perm functions in 4 and 5 are potentially different but they aren't well-defined by Shallit) 
What all Shallit's crowing ends up boiling down to is arguing that since he understands K complexity that all functional information must be like K complexity.  If I have instructions for building an airplane, this is something that is complex and specified and significant and useful in a way that many kilobytes of white noise are not.  If I compress the instructions to 10 kilobytes (really it would be many megabytes of data), it is an incomparably more interesting sequence of bits (with a equally more interesting explanation) than the same number of bits containing a compression of white noise.

Jeffrey Shallit has complained in the past that 'specified complexity' would be better off called 'specified improbability.'  He and Elsberry's long critique of the idea of  'complex specified information' demands.  Winston Ewert's notion of algorithmic specified complexity is very closely related to Dembski's notions of CSI and is itself a specie of randomness deficiency, combining notions of K-complexity and probability, order and chance.

The main thing that is clear from reading Shallit's various writings is that to a man with a hammer, everything looks like a nail, and Kolmogorov complexity is Shallit's hammer.   If the control system I'm building requires 30% more source code to make, I don't want a programmer whose idea of providing more information to simply copy the source files and rearrange letters.  If I have a dictionary, giving me another copy of it isn't telling me much.  Randomly changing letters in that second copy isn't telling me much either, because the 'new information' isn't very useful at all.  If Shallit doesn't think 'useful' is a useful concept, he should try to stop complexity theorists from trying to define it.  Maybe he can filibuster their presentations.

There are two examples that leap out at me where Shallit quibbles over the details-- like he wants to treat anyone who makes a mistake as one of his students giving them an F--and never address whether the example can be corrected/improved.  Even fundamental problems with an example can be corrected (I'm thinking of Lakatos' thoughts on proofs), but these tend to be more "nitpicky."   One example is the critique of the Contact-inspired prime number sequence.  He (with Elsberry) charges that the sequence could be more probably if one posits a post hoc distribution based on the observed frequencies (e.g. more 1s than 0s expected).  That only means that the message has to be longer before we can judge its SC to cross a given universal probability bound.

Another example, is Shallit's kvetching over Barry Arrington's insufficiently random string.  Shallit doesn't claim that it is impossible to come up with a string comparable in both length and compressibility (approximating the incomputable K complexity) to the first 12 lines of Hamlet's soliloquy.  Given that is the case, what useful insight would Shallit offer?  Like his insistence on Kolmogorov complexity, this hair-splitting is either a red herring or profound misconception.

Shallit seems to confuse pedantry for insight.  But then the kind of information that produces design of an algorithm or machine--or possibly a nanomachine made from amino acids--often requires insight, whereas the way that natural selection processes the information from the environment is much more like pedantry.  Natural selection can get very close to something useful and miss it completely.  It shouldn't seem that unusual for

As I recall, in his Dover case deposition, Shallit was careful to exclude evolution from his areas of expertise, but included the following: complexity theory, pseudoscience, 'pseudomathematics.' Complexity theory encompasses so much, there are many experts who are experts in their own definitions of complexity and not in others'.  It is more like an umbrella term for a not-so-unified collection of ideas.  I don't get the impression Shallit could give a good overview of it.  His expertise in "pseudothings" largely consists of his efforts to denigrate intelligent design ideas, and more importantly, to disparage prominent ID people.  But Eric Hehner is a computer scientist who drew ire from Shallit by daring to question foundational ideas of computer science at Shallit's own university.  Shallit attempted to derail Hehner's presentation in order to protect his university from wrong thinking.  Nice.

The behavior would not seem so bizarre if committed by a middle-schooler.  Of course, we have all met adults who haven't moved past the hostile snottiness of 7th grade.  Shallit of course does not recognize his bullying nature as such.  Anyone whose idea of "fun" is to boorishly disrupt a visitor's presentation is a kind of bully.  Or a self-appointed science defender.  But I repeat myself.  Is it more about defending science or about the satisfaction of intimidation (in the name of one's Ultimate Concern)?  Hehner suspects that Shallit has a religious devotion to academic orthodoxy.  I suspect that it is this coupled with a very strong authoritarian streak.  The thought police recognized Hehner's ideas as double plus ungood.  Hehner is neither a theist nor an ID figure.

A great irony is that the Turing proof that Hehner scrutinizes owes a great debt to George Cantor. Cantor and his theories suffered disparagement by Kronecker and other mathematical experts. An renowned expert in mathematics  Kronecker considered himself qualified to dismiss Cantor's pseudomathematics, and protect the university's youth from being corrupted.  The charge that Cantor's work was motivated by religion was probably true but totally irrelevant.  The revolution of set theory (not to mention later computation theory) was neither obvious nor immediate to the mathematical community at large.  (Cantor, himself the object of "mathematical" prejudice, was himself as opposed to "infinitesimals" as some were to his infinities.)

Shallit calls the accomplished Freeman Dyson a “blowhard.” He accuses Dyson of getting "cranky" in his old age, presumably he means being a crackpot.
“The mark of the blowhard is not simply to comment on areas outside his competence, but to do so publicly, with the weight of his reputation behind him, while not doing the appropriate background reading and refusing to seek the opinions of actual experts in the field before publishing. In doing so, the blowhard frequently makes mistakes that would be embarrassing even for those equipped with an undergraduate's knowledge of the area.” [emphasis mine]
In other words, a blowhard is someone who dares to contradict received opinion and question things that an undergraduate would know not to question.  Dyson has the temerity to comment outside his field and intrude where critics are not just unwelcome but unnecessary.  Hehner would be a "blowhard" in the Shallit sense if his expertise were not in computer science.  He is instead labeled "fringe. and compared "to fraudulent archaeologists, circle-squarers, and people who believe 1>2."  In other words, a "denier."  Hehner's response:
Challenging sacred truths can be dangerous to one's reputation and career: the priests who protect their truths will attempt to assassinate your character by writing insulting blogs. That's why I waited until retirement to pursue this topic. Here is the real danger: if challenging basic accepted results becomes too costly (it's not easy to bear the insults)science loses its self-correcting character that distinguishes it from religion. [emphasis mine]
The standard definitions of "blowhard" imply a sense of ego and pomposity: "a person who blusters and boasts in an unpleasant way." There doesn't seem to be anything especially blustery or boastful about Hehner; he doesn't seem to be looking down his nose at anyone.  However, the sort of insults listed here seem to be typical of what appears on Shallit's "Recursivity" blog.  When people complain to Shallit about his unpleasantness, he, like so many self-righteous fundamentalists, seems to assume that they simply can't handle the truth.  It seems ironic that, of all words, Shallit would pick the word "blowhard" for those he considers to be willfully ignorant kooks.

I think Shallit's behavior is representative of the anti-ID crowd of science defenders.  Shallit told Hehner before his visit to Waterloo, that Hehner's "criticism of Cantor's work is trivial, silly, and is likely to be completely ignored for those reasons and others."  He also tells Hehner that he has mistaken Shallit's "amusement" for "ire." Shallit then shows up at Hehner's presentation and tries to filibuster it so that it doesn't even get a proper hearing.  "Likely to be completely ignored" is just not good enough odds.

I was reading Shallit's version of the incident over Debating Design (i.e. the anthology edited by Dembski and Michael Ruse) in his Dover deposition and he was explaining how he never intended to go behind Dembski's back in getting his paper into the anthology.  He seems to be explaining that he had no intention of submitting a paper to Ruse, but only wanted to tell Ruse that he had concerns that if he were to submit to both Dembski and Ruse that Dembski would reject it on a technicality.  What this email to Ruse was meant to accomplish is never explained.  Ruse certainly can't guarantee that Dembski will accept it, and Shallit explains to Ruse that he doesn't want Dembski to know what his and Elsberry's arguments will be before they are published in Quarterly Review of Biology unless it gets accepted in the anthology.  Now Shallit still publicly considers his paper a knockdown defeat of Dembski's ideas.   Why he would be worried at all Dembski might have an answer to some of his criticisms seems weird given his ideas about the "self-correcting nature of science."  Again, "likely to be ignored" is not likely enough.  This unconcern with IDers seems frenzied.

that even while deriding others for publicly reviewing books on topics in which they lack expertise, he himself publicly reviewed at least one book on a topic in which he lacks expertise;

Don't get me wrong. Shallit is "probably a nice guy who is kind to [his] pets" and I'm sure he feels that intimidating people from polluting the intellectual airwaves with unsanctioned nonsense is his version of integrity.  If his mastery of "argumentum ab iniuria" furthers this noble goal of intellectual purity, well, he's not a man who will suffer fools.  Are all Darwin defenders as boorish as Shallit?  Of course not.  But think of the leading figures in ID opposition: Myers, Coyne, Elsberry, Dawkins, Dennett, Padian, Pennock.  He fits right in there with the rest of those surly culture warriors. Their attitudes are worse than those of the poker-playing biologists in Randy Olson's Flock of Dodos.

I think his fellow blogger and evolutionary culture warrior Takis explains the attitude well:

No comments:

Post a Comment