{"id":799,"date":"2009-08-20T22:08:35","date_gmt":"2009-08-20T22:08:35","guid":{"rendered":"http:\/\/scientopia.org\/blogs\/goodmath\/2009\/08\/20\/quick-critique-dembski-and-marks-in-ieee-journal\/"},"modified":"2009-08-20T22:08:35","modified_gmt":"2009-08-20T22:08:35","slug":"quick-critique-dembski-and-marks-in-ieee-journal","status":"publish","type":"post","link":"http:\/\/www.goodmath.org\/blog\/2009\/08\/20\/quick-critique-dembski-and-marks-in-ieee-journal\/","title":{"rendered":"Quick Critique: Dembski and Marks in IEEE Journal"},"content":{"rendered":"

As lots of you have heard, William Dembski and Robert Marks just had a
\npaper published in an IEEE journal<\/a>. In the last couple of days, I’ve received about 30
\ncopies of the paper in my email with requests to analyze it.<\/p>\n

My biggest criticism of the paper is how utterly dull<\/em> it is. It’s obvious
\nhow they got it published – they removed anything that’s really interesting from it. It’s
\na rehash of the stuff they’ve written before, stripped of any content that directly hints
\nat the anti-evolution part of their claims – which leaves a not-particularly-interesting
\npaper on search algorithms.<\/p>\n

I’m not going to rehash my primary criticisms of Dembski’s approach here – I’ve done it lots of times before, most recently in this post<\/a>, which critiques a very<\/em> closely related paper by D&M. In fact, this paper
\nis really just an edited version of the one I critiqued in that post: it’s that paper with all of the intelligent-design speak removed. <\/p>\n

<\/p>\n

What this paper does is take the basic idea of the NFL theorems, and builds on it in a
\nway that allows them to quantify success. If you recall, NFL says that averaged over
\nall possible search landscapes<\/em>, the performance of all search algorithms are
\nequivalent – and therefore, no search algorithm is generally better than pure random
\nsearch. But searches obviously work – we use search algorithms every day, and they
\nclearly work a whole lot better than random guessing.<\/p>\n

Search algorithms work because they’re not intended to succeed in all
\npossible landscapes<\/em> – they’re designed to work in specific kinds of landscapes.
\nA successful search algorithm is designed to operate in a
\nlandscape with a particular kind of structure. The performance of a particular search is
\ntied to how well suited the search algorithm is to the landscape it’s searching. A hill-climbing search-algorithm will work well in a continuous landscape with well-defined hills. A partitioning landscape will work well in a landscape that can be evenly partitioned in a well-defined way.<\/p>\n

So how can you quantify<\/em> the part of a search algorithm that allows it to
\nperform well in a particular kind of landscape? In terms of information theory, you can
\nlook at a search algorithm and how it’s shaped for its search space, and describe how
\nmuch<\/em> information is contained in the search algorithm about the structure of the
\nspace it’s going to search.<\/p>\n

What D&M do in this paper is work out a formalism for doing that – for quantifying the amount of information encoded in a search algorithm, and then show how it applies
\nto a series of different kinds of search algorithms.<\/p>\n

None of this is surprising. Anyone who’s studied Kolmogorov-Chaitin
\ninformation theory will respond to this with “Yeah, so?”. After all, K-C theory describes
\nthe information content of stuff in terms of the length of programs – the search programs
\nclearly contain information – they’re programs<\/em>. And they clearly only search certain spaces, so some portion of the information in the program is an algorithm that
\nexploits the structure of the search space. This much is obvious. Being able to quantify
\nhow much information is encoded in the search? If you could do it well,
\nit would be moderately interesting.<\/p>\n

It’s a nice idea. So what’s wrong with the paper?<\/p>\n

Personally, I come from a background where I look at these things in K-C terms. And
\none of the major results in K-C theory is that you can’t<\/em> really quantify
\ninformation very well. How much information is in a particular search? Damned hard to say
\nin a meaningful way. But D&M don’t bother to address that. They just use a very naive
\nquantification – the same logorithmic one that Dembski always uses. I dislike it
\nintensely, because it essentially asserts that any<\/em> two strings with equal
\nnumbers of bits have the same amount of information – which is just wrong<\/em>.
\n(“00000000” and “01001101” have different amounts of information, according to information theory.) The reason that Dembski can quantify things specifically is really because
\nhe’s handwaved that important distinction away – and so he can assert that a particular
\nsearch encodes a certain number of bits of information. What he actually means is
\nthat a given search algorithm can be said to contain something equivalent to a string whose length is N bits – but he doesn’t say anything about what the actual true
\ninformation content of that string is.<\/p>\n

Second, he rehashes some old, sloppy examples. As has been discussed numerous times,
\nDembski likes to harp on the “Methinks it is like a weasel” example used by Dawkins. But
\nhe misrepresents the example. In Dawkins’ presentation of that example, he used a program
\nthat searched for a string based on its closeness to the target string. It did
\nnot<\/em> lock positions in the string. But Demsbki has always presented it as if it
\nlocks. (In a locking version of this, when a character position is identified as correct,
\nit’s removed from the set of possible mutations – it’s locked, so that it will never
\nchange from correct to incorrect. In Dawkins’ real example, a correct character position
\nwould sometimes change to incorrect – if a given mutation round, one of the candidate
\nstrings added two new correct character positions, but changed one formerly correct one to
\nsomething incorrect, that candidate would survive.) In this paper, one of the examples
\nthat Dembski uses is the Weasel with locking. (And he does it without any reference to
\nDawkins.) He uses it as an example of partitioned search. He’s absolutely correct that the
\nlocking version of that is partitioned search. But frankly, it’s a lousy<\/em> example
\nof partitioned search. The classic partitioned search algorithm is binary search. For more
\ncomplicated examples, people generally use a graph search. So why did he use the locking
\nweasel? I’m attributing motive here without proof, but based on his track record, he wants
\nto continue harping on the weasel, pretending that Dawkins’ example used locking, and he
\nwants to be able to say that his use of it was approved of by a set of peer reviewers.
\nIf<\/em> he claims that in the future, it will be a lie.<\/p>\n

As for intelligent design? There’s really nothing<\/em> in this paper about it. I’m Dembski will run around bragging about how he got an ID paper published in a peer-reviewed journal. But arguing that this is an ID paper is really dishonest, because all ID-related content was stripped out. In other places, Dembski has used these quantification-based
\narguments to claim that evolution can’t possible work. But this paper contains none of
\nthat. It’s just a fairly drab paper about how to quantify the amount of information
\nin a search algorithm.<\/p>\n

A final couple of snarky notes:<\/p>\n