Joined: Jan. 2007
|Quote (Richardthughes @ Nov. 16 2007,12:20)|
|They're still banging on about "active information".|
I would suggest that the "active information", ie the selection process resides in the environment itself and so poses no real problem for evolution.
I think you are correct - "active information" is supposed to tell us how much better any search does than a search over all possible fitness landscapes, which is equivalent to asking "how much worse does this search do over a pathologically difficult search space" (since almost all search spaces are pathologically difficult). Of course no one cares how well searches do over all these insane spaces; we only care about how well searches do over physically realistic search spaces - i.e. spaces like our own where life is generally good.
As far as I can tell, Dembski's "active information" critique of genetic algorithms is the equivalent of complaining because RM+NS doesn't produce life in a vacuum.