Author: Rolf Tueschen
Date: 12:38:48 06/12/05
Go up one level in this thread
On June 12, 2005 at 14:31:29, Robert Hyatt wrote: >On June 12, 2005 at 07:12:35, Rolf Tueschen wrote: > >>On June 11, 2005 at 15:51:27, Robert Hyatt wrote: >> >>>Actually the rating of Chess 4.x ought to be reasonable, because it was _not_ >>>based on playing other programs. It was a pure USCF rating earned by playing in >>>tournament games, just like we did with Cray Blitz, and like Hsu, et al did with >>>Deep Thought and Berliner with HiTech. >> >>This is wrong science applied. >> >>1. the expression "pure USCF" is misleading because it's about the performance >>of an alien among human chessplayers; aliens dont have "pure" USCF ratings and >>this is also why machines were banned from such tournament play > >Not exactly, anymore than the Cray Blitz rating was similarly skewed. It played >regularly in USM chess events, and it played in every Mississippi state >championship event starting around 1975 or so. The players knew it well, and >were not "shocked" to be paired against a computer they had never played >before... Of course it's so, because the assumption that "knowing" a machine is the same as "knowing" the play of a human player is false. > > >> >>2. the fault is in the assumption that playing against a machine is the same in >>terms of chess as if you would play a human player - which is apparently not so > > >No it isn't. Cray Blitz played well back then, but it won lots of games based >on the idea of not making simple errors, and hanging on until the human did make >"that one mistake" that would lose the game... But humans figured that out >pretty well, and tried to play against the machine in ways they knew it would >probably make errors. Back then, deep tactical threats were not picked up very >well, and good humans played wild gambits, letting the computers eat pawns until >they choked on them. So the humans were prepared to play... I shared the experience much later when I won against a program on hardware in 1996. It's an outdated experience but in general you are right because it's still true that you can win by pushing the decisive moment beyond the possible depth of the machine. > > > > > > >> >>3. the game of chess played by human chessplayers is based on experience of >>human chess; a good chessplayer can play against everything a human could offer; >>machine chess is something different and without experience it's beyond any >>rating attempts > >But that is assuming facts that are not correct. "without experience" in the >Case of Cray Blitz is wrong, because it was a regular player in Mississippi >chess events... Here you make the erroneous setting equal of experience through replay and experience through training on the object. Along the old saying that you yourself must play to become a better player. > > > > >> >>4. if you play machines you must use other experiences; you should play against >>the machine, and not against the chess as if played by a human >> >>[The same fault is in the statistic if seldomly a strong woman player appears in >>tournaments; we all know about the sub-conscious of male chessplayers; now >>playing a woman is a differet experience; this is also why such a strong female >>player like Judith Polgar for example has a highly exaggerated score, simply >>because males didn't play really dirty on her, which would be judged as unfair >>sports BTW; to get the difference just take the famous Kasparov vs DB2m we >>already discussed so oftenly; Kasparov was not prepared to meet the team playing >>dirty; say what you want about contracts, the dirtyness is lying in some trivial >>details, you know well from your own life experience; even we two are a good >>example for the existence of such existent details: we can talk in a friendly >>manner like this but let us begin a debate about death verdicts and such some, >>we could quickly end in some virtual verbal massacre - dont you think so? BTW >>best regards to you from Germany]
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.