Author: John Coffey
Date: 08:45:30 07/26/00
Go up one level in this thread
On July 26, 2000 at 00:37:27, KarinsDad wrote: >On July 25, 2000 at 18:21:34, John Coffey wrote: > >>>Time limits, node limits, and even ply limits all have the same inherent type of >>>problem. They pick the best move found SO FAR, regardless of whether the search >>got interrupted or not. >> >> >>Your example seems to illistrate why node counting works better - because some >>quiescient >>searching will occur. Maybe 512 wouldn't look at 3 ply but that is one of the >>lower levels. maybe bump it up 7 levels to 65536 nodes and you are looking at >>3 ply with some search extensions. (Assuming that your node count for each >>ply level is correct.) >> >> >> >>>And in order to gradiate the scale, you do not want to limit the search and HOPE >>>for a tactical or positional mistake that a given rated player can take >>advantage of in order to play at a similar rating as the computer is set at. >>>Instead, you want the program to play tactically, positionally, and >>strategically inferior to it's normal powerful strength. >> >> >>It is hard to see why not. At my 512 node example the program is going to be >>pretty stupid. At 1024 nodes it is going to be smarter but still stupid. At >>2048 nodes it won't be as dumb etc. >> >>I can prove that this system works because I worked my way up the rating >>scale by first practicing against stand alone chess computers (4 mhz) at the >>low level until I could beat it. Having surmounted that task, I tried the next >>level up and repeated the process until I could finally beat the top level. In >>each case the program just had more time. At 2 minutes it made fewer mistakes >>than it did at 1 second. Every time I went up a level I was facing a just >>slightly tougher opponent. Granted it didn't drop pieces at 1 second per move >>but if I looked deep enough I could see something it didn't. >> >>On my PIII 450 I would have to start with at most .01 seconds per move to get >>the same effect. >> >>John Coffey > >I'm not saying that your method would not result in weaker play by the computer. >It will. I'm saying that your method will not emulate well weak human play. I >think your method would emulate it better than changing max ply (since you would >have a finer granularity), but I think that my method is an improvement over >yours like yours in an improvement over dropping max ply. I also think that >there is not much difference between your method and maximizing the time >(assuming you could get to millisecond or lower time frames). > >KarinsDad :) My concern is that attempts to choose from a random list of moves will produce the same kind of effect that we see with some programs - where they will play really well on some moves but do something completely stupid on another. I have had games with Fritz (on a lower rating) where it clearly could have won a king and pawn endgame but started making random king moves and lost. It was much too obvious that the computer was throwing the game. It seemed bad to struggle so hard against the computer the whole game only to have the computer completely botch it. Now maybe the idea of playing well on some moves and bad on others is how 1200 players actually play. I prefer to have an consistant standard that I am playing against - so even if the program doesn't make one move blunders (unless we set the nodes really low) I am going to get a more consistent feel throughout the whole game. I feel that this is a better training method for me trying to improve my game. It is hard to define how many bad blunders a 1200 should make. I have played 1200's in tournament who drop pieces. On the other hand, I just lost to an 1100 in a tournament who didn't seem to make any mistakes the whole game. (I am an A player.) Yes, I wish for the ability to set the think time to fractions of seconds per move. John Coffey
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.