Wesley R. Elsberry
Posts: 4991
Joined: May 2002
|
| Quote (fnxtr @ May 06 2014,10:45) | | Quote (midwifetoad @ May 06 2014,08:40) | Winston Ewert is almost, but not quite, as clever as evolution.
| Quote | | The evolutionary model, Avida, is best known for evolving the EQU function. In the supplementary materials for the 2003 Nature paper, the authors presented the shortest known program to compute EQU taking 19 instructions. They note that it hasn't been proven that it was the shortest program. In fact it is not, and I present a program that computes EQU only using 18 instructions. |
http://www.uncommondescent.com/compute....uctions
Lenski:
| Quote | | The number of instructions required for EQU ranged from 17 to 43, with a median of 28 instructions. Notice that one evolved type apparently needed only 17 instructions to perform EQU, whereas our shortest hand-written program used 19 instructions. |
https://www.msu.edu/~pennoc....lex.pdf |
Speaking of "pathetic level of detail"... |
'johnnyb' proudly linked to his BSG paper on irreducible complexity and computer science.
'johnnyb' there makes an attempt to cast IC in terms of Universal Turing machine functionality. He comes up with five 'principles' that he believes establish IC as a computational reality. The one that interested me was the fifth:
| Quote |
5.
Because the chaotic portions of the Universal machine are being used, the solution cannot have been arrived at incrementally because it violates the definition of chaotic behavior, which does not display smooth and predictable outcome changes when the initial values of the tape are changed. Therefore, incremental searches will not make the searches find a solution any faster except perhaps on trivial problems.
|
That's a simple category error. The phrase 'johnnyb' was apparently looking for instead of 'incremental' was 'deterministic'. Fortunately, nobody has tried to shoehorn evolutionary processes or evolutionary computation into determinism (except Sal Cordova, who also was wrong). The remainder of the essay is an exercise in deriving conclusions from false premises, including the section on Avida that 'johnnyb' is apparently particularly proud of, but shouldn't be.
--------------
"You can't teach an old dogma new tricks." - Dorothy Parker
|
