Author: Barry Baker
Date: 07:19:59 04/19/04
Go up one level in this thread
On April 18, 2004 at 22:36:34, Mike Byrne wrote: >On April 18, 2004 at 20:36:56, Steve Maughan wrote: > >>I'm looking to implement the new ELO rating support in UCI 2. In the past I >>haven't given much thought as to how one could make an engine predictably >>weaker. Does anyone have any suggestions? After the little thought I have >>given the subject here's what I've come up with. >> >>1. Search to a shallower depth / shorter time. This seems somewhat >>unsatisfactory as it will ignore the other setting. >> >>2. Add random elements to the evaluation term - the bigger the term the weaker >>the play. >> >>3. Round off at the root. So to weaken the engine slightly you would have the >>root round off to 0.1 pawns, a further weakening would be in units of 0.5 pawns >>etc. This means that if a value came back as +1.23 the engine would regard it >>as +1.20. I like this idea since it is relatively simple to implement :-). >>However I'm not sure how effective it would be at actually weakening the engine >>(it may speed the engine up since there may be more cuttoffs!?!). >> >>4. Within the tree you could ignore some moves e.g. all losing capture after x >>depth. This is also interesting. It may closely match how humans think and >>err. Of course it is more complex to implement. >> >>Any other ideas. This could be an interesting area for discussion as it has >>virtually no commercial value (IMO nobody is going to buy an engine that is >>weak). I'd be interested to know of other idea. >> >>Regards, >> >>Steve Maughan > >The two simplest methods are to reduce depth and to provide offbeat score values >for pieces. I use a combination of both in dumming down Crafty in the modified >Se version. In additon, I implemeneted a delay move feature so it would would >not move so quickly while serching to a SD of 3. There are versions that have >tens of thousands of games played already with a rating near 1300/1400 for some >of the weaker personalities. Those operatotors can tell you more than me how >effective the dumb down Crafty's are. The fact that it has already played in >excess of 10,000 games on ICC tells you much people like it. > >With a search depth of 1 with full extensions, Crafty will play near 1500 in >blitz on ICC. To get it weaker, you have to reduce extensions and/or provide >offbeat piece values and/or weaken the evaluation function. I do the first two, >never touched the evaluation function to weaken it. As Mike indicated, the weakened Crafty's are very popular on ICC. I operate KingPawn, which uses Crafty v19.11 SE modified to play with a 1400-1500 rating on ICC. KingPawn has played almost 50,000 games in less than three months, averaging approximately 600 games per day. All games are standard time control and virtually all are against humans. KingPawn doesn't play GM level chess but it's a great sparring partner for many ICC members, particularly those with standard ratings of 1100-1500. ICC members have an average standard rating of 1561. KingPawn's play may not always be "human-like" but, then again, it's available to play 24/7 (83% uptime), its level of play is reasonably constant (albeit weak by computer standards), it never cheats, it's never rude, and it never disconnects in a losing position. Barry
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.