Author: Dagh Nielsen
Date: 14:11:34 11/09/05
Go up one level in this thread
Yes, openings are an important point. Some absolute world class player posting on TCCMB said that in his experience, the majority of top level CC games were decided between move 6 and 14. The problem for engines is that it is impossible to make a general opening book of the same level as dedicated analysis on a concrete position. Once the human knows what opening system they have entered by means of exchanging the first few moves, he can analyse this system to a far greater extent than the engine's bookmaker who had to prepare a complete repertoire of opening systems. The other advantage of the human + engine constellation is that humans are extremely good at pruning a tree of analysis. After some analysis, humans can make judgement calls like "move A is definately inferior to move B" and then don't have to analyse move A anymore. Likewise, humans have very good intuition about which lines need more analysis before a verdict can be reached, something that is very difficult to program into an engine. The result is that engines will waste a lot of calculation power every time they go a ply deeper. Engine analysis assisted by humans is safer and can reach deeper depths where necessary. There is also the question of correctly assessing endgames. In the end it comes down to the question: Is human input and guidance beneficial or detrimental? I firmly believe that chess is of a nature where human input is very valuable, and hence humans + computer will always (or at least for a very long time) be stronger than either alone. Other calculations would not benefit from human input, for example the task of factoring a number composite of two primes. Here human input during calculations would probably be detrimental. Regards, Dagh Nielsen On November 09, 2005 at 09:31:34, Uri Blass wrote: >On November 09, 2005 at 08:26:36, Dagh Nielsen wrote: > >>I'm FIDE 2165 or something, and I bet I can beat Hydra in a correspondence chess >>match :-) > >Maybe and it is not clear. > >> >>Human + computer is far superior to either alone, as was also demonstrated in >>the freestyle chessbase tournament where Hydra (not complete version admittedly) >>participated and got kicked out. > >not enough games to know. >It may be result of bad luck in the choice of the openings. > >I did not analyze hydra's games but unlike games of humans luck in the choice of >the opening is a significant factor in games of chess programs. > >strong humans usually do not have x% probability of choosing 1.e4 c5 and y% >probability of chooisng 1.e4 e5 and z% probablity of choosing 1.e4 c6,... > >The result is that strong humans are more stable than chess programs > >Note that I also do not know if hydra(not complete version) was stronger than >computer opponents at the relevant time control. > >Uri
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.