Author: KarinsDad
Date: 12:44:19 01/18/99
Go up one level in this thread
On January 18, 1999 at 08:05:58, Fernando Villegas wrote: >On January 18, 1999 at 02:36:21, Laurence Chen wrote: > >> > >Smart example, Laurence.- >F Smart? More like inane. Not to criticize Laurence for having a different opinion, but huh? Paint and chess programs? A chess program has the capability of being more than just a strong engine (i.e it can be every color in the rainbow, and infrared, and ultraviolet, based on the settings). When a program allows a 2 move hanging of a piece and does an obscure 8 move mate both within the same game, then the program is FLAWED with regards to the level settings (or whatever terminology is used to indicate a weaker set of play for a given program). > >There has been some posting criticizing the strength of chess engines inability >>to play at the amateur ELO ratings, and let me ask you this, can you turn red >>paint into blue paint? Really, if you go to a paint store and buy red paint, you >>cannot expect the red paint become blue no matter how much dilluting you do to >>it. Red is red, and if you dillute too much it will disappear all together. The >>same principle applies to chess engines, if you go and buy a chess engine which >>plays at a GM strength, it does not matter how much tinkering you do to the >>program, it will either continue to play strong or too stupid to play chess like >>the paint scenario. If you want a chess engine to play at amateur strength than >>perhaps the freeware/shareware version would offer such programs, or get some of >>the old chess engines such as Psion Chess, Sargon II, Sargon III, Chessmaster >>2000. Either that or start applying better studying techniques to improve your >>game. Going back and playing "old" versions is no better than playing the current versions at lower settings. They will still make random mistakes at lower settings as opposed to mistakes, based on their inherent MinMax search engine design. They base the choice of move not on a limited understanding of the game, but by randomly responding with moves that resulted in a lower score. >>Also, can you ask a GM to play at a 1400 ELO? See the dilemma that chess >>programmers face, they don't know how a chess amateur thinks and how a chess >>amateur analyzes chess positions, so no matter what they do, they will never >>produce a chess engine which will play like a human amateur BTW, most chess programmers do not know how GMs think either (in fact, GMs have a hard time with how they think since a lot of it is done subconsciously). The resultant strength of their programs is based on the sum of a variety of heuristics designed to eliminate mistakes. This is a LOT stronger than the actual chess understanding of the programmers and in fact, often surprises the programmers with unexpected great moves. When Joel Benjamin was helping out the Deep Blue team, he did not impart GM level wisdom to the programmers. Rather, he assisted with adding elements to the evaluation such as open files, pawn islands, and similar concepts (as well as improving the opening book). These concepts are ones easily understood (but not necessarily easily implemented OTB) by B class players. When you add just a little chess knowledge to a search engine that can already do 200 million positions a second, you create a program that is not only great, but almost GM human-like in it's playing ability. The bottom line is that the commercial manufacturers have not taken the time out to analyze how lower rated players think (by analyzing their games and the resultant mistakes) and hence have not YET implemented a more sophisticated lower level of play in their programs (which BTW repeatedly have been advertised as having this feature). >>, and amateur chess >>analysis don't sell in chess publications. I don't understand what amateur chess in publications has to do with the issue. There are a lot of amateur chess games, some with annotations as to mistakes (a valuable learning tool for the amateur) in Chess Life and every state newsletter in the US. Most amateurs can learn more from an analysis of amateur level games than they can from an analysis of GM level games. Also, if a GM cannot play at a 1400 ELO, how can you expect the majority of chess literature to be aimed at the 1400 player (even though the majority of players are within 200 points of that level)? A GM may have a tough time remembering which concepts are basic to a 1400 player and which are advanced. Do 1400 players know about pinned pieces, passed pawns, isolated pawns, pawn islands, king hacks, deflection, destruction? Gee I cannot remember, so I guess I'll have to give simple examples for all of these. It's much easier to write a book aimed at people of the same calibre as yourself (try explaining a C Switch statement to my wife and you'll understand the difference). KarinsDad
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.