Author: Don Dailey
Date: 09:23:22 02/25/98
Go up one level in this thread
On February 25, 1998 at 12:09:24, Ulrich Tuerke wrote: >On February 25, 1998 at 11:29:06, Don Dailey wrote: > >>>I have wondered why programs evaluations are measured in pawns intead of >>>probablilities of winning. Has no one done this? Has anyone ever taken >>>the evaluations in Informant as probabilities of winning and regressed >>>them against explanatory variables such as material, space, etc. to fit >>>this function. >>> >>>George Essig >> >>I'm actually working on this! This is how I think of >>evaluation and it would be natural to convert the program >>to this system. However I'm not sure it's any more useful >>than simply finding the right function to convert a score >>to a probability. But most of it's usefulnes is just >>thinking in these terms (whether you actually implement >>it or not.) For instance, I believe having an advanced >>passed pawn should not affect your probability of winning >>too much if you are already a piece up, but should have >>more impact on the score if you are down a piece. A simple >>linear bonus for this passed pawn might not be quite right. >> >>In general, I believe many positional terms should change >>in value when material is not close to zero. Another way >>of viewing this is to say "don't be as eager to hunt pawns >>if you are already have extra material." It's the same >>concept. > >IMHO, this is a very interesting idea. However, I am afraid that a >reasonable implementation of such behaviour is very tricky. So, as as >consequence a program having large material advantage would be inclined >to give back material for - may be temporary - positional advantage. I >think that the program shouldn't try to further increase its material >advantage on cost of large positional decreases, but on the other hand, >material should not be given away unforced. If you succeed and it works >fine, then the playing style of the program will become far more >human-like and aesthetically. >Good success, Uli > >> >> >>- Don Yes, I agree it will be somewhat tricky. I think you have to be fairly conservative. If your program already has the ability to sacrafice a pawn for advantage, then you probably can implement this idea in conservative way with good effect. The trick is to not go overboard. I don't expect it to give up a knight to prevent the opponent from getting a passed pawn unless there are terms to show it's unstoppable. But a pawn? If you are 2 or 3 pawns ahead and your evaluation already is good enough to see it's dangerous, a little bit of this idea can go a long way. By the way, I suspect Mchess (and maybe many other programs) already do similiar things. More than other programs I've seen, Mchess seems to have a sense of this kind of danger and seems to be more willing to make these judgements. - Don
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.