Wesley R. Elsberry
Posts: 4991 Joined: May 2002
|
Quote (Patrick @ Feb. 06 2015,15:37) | Okay, this got me to peek over at UD for the first time in months. The first sentence of Ewert's abstract:
Quote | According to conservation of information theorems, performance of an arbitrarily chosen search, on average, does no better than blind search.
|
This is a typical IDCist misstatement of the No Free Lunch theorems. A more accurate summary is "No algorithm outperforms blind search when averaged over all possible search landscapes."
On a particular landscape, of course, some algorithms far outperform blind search. Leaving aside the issues with modeling evolution as a search, fitness landscapes like those we find in our environment are quite amenable to search by known evolutionary mechanisms.
When he starts off with an error of that magnitude, I don't have a lot of confidence in what's coming next. |
Nothing new under the sun, apparently.
-------------- "You can't teach an old dogma new tricks." - Dorothy Parker
|