Author: Ed Schröder
Date: 02:22:32 10/06/99
Go up one level in this thread
>Subject: Re: Crafty and single-computer winboard matches >Posted by Peter Kappler on October 06, 1999 at 03:33:47: >>I am willing to play 100 40/40 auto232 games. >> >>PC-1: Rebel Century PB=ON >>PC-2: Rebel Century PB=OFF >> >>Indentical hardware of course. >> >>My bet... >> >>Expected match score: 70-30 >> >>Now how many ELO is that when that score pops up? >> >>Ed > > >A 70% win expectancy is around 140 ELO, so I guess you are suggesting that >this >is how much stronger Rebel would be if it could ponder every move correctly. >But this seems like a different discussion - I don't quite see how it >relates to >the question I asked Bob. > >The point I was trying to make is that it shouldn't be that hard to modify >Crafty's time management code to handle the PB=OFF scenario. Bob says it >would >take a lot of tuning to get this working well in Crafty, which seems odd to >me, >but I will take his word for it. > >Perhaps it is just too late - am I missing the point of your post? > >--Peter Yes you missed my point. Here is my complete view... Divide the comp-comp issue concerning playing strength into 3 categories: #1. comp-comp matches on 1 Pc. (winboard etc) (no permanent brain) #2. comp-comp matches on 2 Pc's (auto232) #3. comp-comp manual play (WCCM, manual tournaments some people do) I claim it will be not unlikely to see total different results for each category. Example: Program_X, Program_Y and Program_Z. XYZ participate in all 3 categories. In the end it wouldn't surprise me if you get the following picture: Category-1: 1. Program_X 2500 2. Program_Y 2450 3. Program_Z 2400 Category-2: 1. Program_Z 2500 2. Program_Y 2450 3. Program_X 2400 Category-3: 1. Program_Y 2500 2. Program_Z 2450 3. Program_X 2400 Please note that I am not saying Winboard matches are worthless but that they should be judged within its limitations. The very same is true for auto232 matches. They gave you an indication who is best in THAT PARTICULAR category. No more no less. The current discussion is about the value of the permanent brain and IMO its value is highly underestimated and is the BIG issue playing matches on one PC. Program_X may gain 50 elo because of the Permanent Brain, Program_Z maybe gains 100 elo or maybe more. I have started these 100 40/40 games already spread out over 3 autoplayer pairs. I expect to have the results within 3-4 days. To proof my point I will do a second experiment that (hopefully!) will give some more light on this case as I have gotten curious myself and want to know. The second scenario: 100 auto232 games. PC-1: Rebel Century PB=ON / TC 30 seconds PC-2: Rebel Century PB=OFF / TC 60 seconds My bet: PC-1 is at least equal. Ed Schroder
This page took 0.03 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.