Author: Uri Blass
Date: 15:48:09 10/21/02
Go up one level in this thread
On October 21, 2002 at 17:57:58, Uri Blass wrote: >On October 21, 2002 at 17:03:59, Johan Hutting wrote: > >>On October 21, 2002 at 11:34:33, Uri Blass wrote: >> >>>>No clue, if I see larsenVB and tscp playing in 3rd division I wonder what Celes >>>>does in 4th :) >>> >>>Simple answer >>>see >>>http://wbec-ridderkerk.nl/html/his3rdedition.html >>> >>>Celes simply did not score well enough in the previous 4th division. >>> >>>The level to get to the 3th division is clearly better than tscp. >>>The level of tscp is not enough even to promote to the 4th division. >>> >>>You may ask Leo to use the last Celes but I doubt if it is going to be strong >>>enough to promote. >> >>Leo already has a copy of the latest delphi version. >>> >>>For the low divisions the level that is needed to promote is higher than the >>>level of the higher division. >> >>So I've noticed with engines like list, tao and knightdreamer around. >> >>>It is not going to be the case in the high division and part of the programs >>>that pronmoted to the first division went back to the second division. >>> >>>Abrok,Biglion,Knightdreamer,List,Rdchess,Tao and maybe gnuchess(if Leo finds it >>>stable) may fight for promotion. >>> >>>It seems that Celes needs to be better than one of them in order to promote. >>> >>>7th place may be also enough for pronmotion if Celes can beat Soldat. >> >>I already pointed out the insanity of this on the winboard forum. However, it's >>Leo's tournament and he's free to make up his rules. And well, maybe Celes is >>lucky :) >> >>>I admit that I did not analyze the games but it seemed to me that their problem >>>were after the opening and the d8->d5->d6 and few moves later->d8 line for black >>>does not seem to be a good line when I look at it. >> >>That's standard procedure in that line... idem dito in the Qa5 line, after c6 >>the queen usually returns to d8 or c7. >> >>>>>My latest movei(not public) has book and pondering >>>>>but my tests are almost always with no book when >>>>>I always change the >>>>>first move so I can get different games. >>>> >>>>hmmm, you'll still get alot of repetitions that way. >>> >>>If you play more than 40 games then I agree but if you play 40 games you get no >>>repetition when you always change the first move(the only way to get repetition >>>is changing the order of moves but it does not happen often when the first move >>>is 1.a4 or 1.f3(it can happen in 1.e4 e6 2.d4 to 1.d4 e6 2.e4 but I do not see >>>logical change from 1.a3 e5 2.e4 to 1.e4 e5 2.a3 >> >>You're serious about testing movei by opening with moves like 1.a3 and 1.h3 vs >>other engines? > >Yes > >It even could beat Ruffian with 1.f3 at 7 minutes per game. >The total result was 37.5-2.5 for Ruffian(you can find the game in the winboard >forum when I posted the only 2 games that movei won) > >One of them was with 1.d4 f5 and one of them was with 1.f3 > >I am interesting in the strength of the engine and not in building opening book >and I can compare results of different versions > >Uri Sorry for the previous empty post 2 notes 1)The games when movei beated Ruffian are in the following post: http://www.f11.parsimony.net/forum16635/messages/36718.htm 2)Ruffian is using average time of 6 seconds per move at 2 hours/1200 moves thanks to it's stupid time management but inspite of this fact I believe that it probably can beat most of the engines at that time control even if they assume 2 hours for game. I am even not sure if latest movei is good enough not to lose against Ruffian in these conditions. I think that using 6 seconds per move does not give better quality than using 7 miutes per game so the quality of the moves of Ruffian at 7 minutes per game is good enough for a lot of engines and I think that Ruffian at 7 minutes per game can play at similiar level to the public Ant at 2 hours per game (If the public Ant use time in a logical way in 2 hours/1200 moves then it is possible to check it). 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.