Author: Harald Faber
Date: 03:47:39 04/28/99
Go up one level in this thread
On April 27, 1999 at 08:46:17, blass uri wrote: >>>2200-2300 correspondence rating players have good pracrical chances to win but >>>the computer also has a good chance to win. >> >>May depend on the tie control. >> >>>Remember that the cc players sleep,eat when the computer can concentrate in the >>>position for many hours without getting tired. >>> >>>The computer may find a good move that the cc player missed because (s)he did >>>not give the computer enough hours to analyze the position. >>>Uri >> >>That would be playing comp-comp. I assume the cc player developping ideas and >>plans on his own and only CHECKING them with a program for tactical holes. >>In this issue humans ARE better than programs. > >There can be a tactical hole that the programs need hours to see and the human >may miss it if (s)he prefers to check with the computer many positions and >cannot concentrate with it only at the root of the tree. > >Junior4.6 won a correspondence game in the 1/2 final of the correspondence >championship of Israel because of a tactical mistake of the opponent that >computers needs a long time to avoid. > >The opponent told me that he used a computer to help him. > >The opponent is not a weak correspondence player. >He is playing in the final of a previous correspondence championship of Israel > >Uri I guess he didn't use the program in an ideal way. To check with computers I meant to check your favourite lines. Because of horizon and other effects programs don't find (or need a very long time) deeper tactical solutions. But if you play [through] your favourite and main lines things suddenly may change. Especially Genius often needs only one ply to find the idea. That means he may be at depth 11/23 and sees nothing, if you make the best move, suddenly and MUCH earlier than in depth 11/23 you see fail low/fail high in Genius analysis. These things are very dangerous!
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.