Rob
Posts: 154
Joined: July 2007
|
| Quote (Hermagoras @ July 20 2007,18:51) | Dembski's latest: | Quote | | Here’s a fun interview with my friend and colleague Robert Marks. I hope you catch from the interview the amibitiousness of the lab and how it promises to put people like Christoph Adami and Rob Pennock out of business (compare www.evolutionaryinformatics.org with devolab.cse.msu.edu). |
Let the comparing (DING!) begin! |
Been there, done that.
Here is the work of the Evolutionary Informatics Lab in 300 words or less:
Dembski and Marks came up with a concept they call "active information", which is a property of searches. By definition, a search has active info if it performs better than a baseline random search.
D & M show several ways that a search can be better than random, and they conclude the following: | Quote | | If any search algorithm is to perform better than random search, active information must be resident in it. |
Plugging their definition of active info into this quote, what they're saying is: | Quote | | If any search algorithm is to perform better than random search, it must perform better than random search. |
Not terribly insightful. But that is seriously the upshot of their whole project. As Dave Barry says, I am not making this up.
In comparison, the Devolab at MSU looks like it produces more than tautologies.
--------------------------------------------
Further comments on D & M's project:
D & M draw some unwarranted conclusions from their semantic game-playing. For instance, they opine that "attempts to characterize evolutionary algorithms as 'creators of novel information' are inappropriate." I don't know who D & M are quoting here, but the quoted person certainly isn't referring to D & M's "active information". So it's just another shameless equivocation on the word "information".
They also think that if a study involves a particular search, then it should say up front how much active info is in the search, and that a failure to do so amounts to "smuggling" or "sneaking" information in. But to say that a search has active info is merely to say that it performs better than random. Is anyone trying to hide the fact that genetic algorithms, etc., perform better than random searches for virtually all useful problems? Ah, but in an alternate universe where natural laws aren't well-behaved, genetic algorithms would be no better than random search. So it all boils down to the age-old question of why this universe has well-behaved laws. Unless D&M have a convincing answer to that philosophical question, their conclusions have no merit.
--------------
-- Rob, the fartist formerly known as 2ndclass
|
