Author: Hristo
Date: 12:02:53 07/01/01
Go up one level in this thread
On July 01, 2001 at 12:12:40, Otello Gnaramori wrote: >On July 01, 2001 at 08:53:52, Hristo wrote: > >>However, CTiger played this game very well ... infact after only 20 moves the >>game was over! ;-) However, COMPS are not GMs. Not by a long shot! > >Ok, let's say they are IM. No, they are not!!! This is not the point though! Pure strength in competitition is not what makes human player stronger. Although being in competitions is the main way we attempt to mesure things ... the existance of tournaments is mainly needed so people can make money! You must undestand that at this point (making money) anything goes. Well pretty much! :-) The interesting thing is that good players can summarise their expiriances and give you simple directions or approaches to get better and to understand the game better. COMPS, on the other hand (yeah I know I have five fingers), achieve some mind blowing depths of calculation ... pure and simple is that COMPS go deeper all(!?) the time than most (all) people do, while evaluating a position. Most of the time what COMPS see is good enough to win, because humans make errors (a lot more than we have known before:-) and these errors are primarily tactical. So COMPS punish the human kind. IM and GM do not differ only by the depth to which they calculate!!!!! The basic understanding of the game and different patterns makes them different! (not just ply 12 vs ply 13 kind of thing) Sometimes an IM can go much deeper (calculating) than a GM. The GM will simply reject a given play-path, because "I don't like it." without any long-lines to prove it! What good is it for me then to call a COMP an IM or a GM ... when in fact I know that the _thing_ that makes the computer strong is unattainable by me. 20-30 plys deep is something I will never be able to do on my own during a life game, for every move ... heck 10-15 plys is beyond my scope ... perhaps 5-7 plys ... This is my limit. IM or GM can not change my limit. Thay can make me a stronger player never the less, because they can make me use my 5-7 plys barier better. A COMP can not do that. A COMP will reason me down to: COMP: "What?! You don't see it?! It is clear win!" ME: "No, I don't ... It's too far!" COMP: "You must be blind! ... Look .... (ply 12)" ME: "Ohhh...yes... What about this ...?" COMP: "This is even simpler! Can't you see! Just do this ... <another long line> .... (ply 17 with extensions)" ME: "Jeeez .... how do you do that?!" COMP: "I use an Plutonium CPU. With photon based memory storage. (ver. 2.0.1)" ME: "Ohhh ... goshhh ... where do I get these things?!" COMP: "In a hardware store near you." ME: "OK, then. I go to Idaho now and start growing potatoes." ;-) However, some times, COMPS are un-apt and the pure and simple problem of horizon blindness becomes apparent. This is where humans, most of the time, have their chances. > >> Just >>performance in competitions is not my way of measuring the COMPS. > >What is your way ? Just curious about.. Play COMP-COMP games till you are blue in the face .... then look at the games if you have the time or the tool! ;-) This is the only way it should be done! Humans shouldn't be used as practice-targets .... However it comes down to money. So the COMPS, somehow, get into normal tournaments to gain the recognition. So eventualy the given program will sell and the author will make money! That's fine ... but at some point all top level programs will cover the "COMP_GM_" norm including crafty. (Lots of politics come to mind) ... so, at some point, the only way to promote your program will be to win against the rest of the COMPS. Here comes the good quetstion! How do you measure which COMP is stronger?! ... What is the strongest COMPS (computer + program) available today? Anyone?! See if anyone answers this! How do they validate it?! You'll understand how futile is the question about COMPS being GMs or IMs. best regards to you. hristo BTW. my email is not correct!!!! the correct one is: donquixote@pacbell.net > >>It is given >>that in a tournament most COMPS can cover a GM norm. This is not enough, >>for me, to consider that COMP=GM ... >>For one the GM has a human aspect to it >>and COMP does not! ;-) >> > >But you forget the Turing Test...
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.