Author: Roberto Waldteufel
Date: 16:23:16 05/24/00
Go up one level in this thread
On May 24, 2000 at 12:49:45, Bruce Moreland wrote: >On May 24, 2000 at 10:28:57, Oliver Roese wrote: > >>Hi all! >> >>This question is about pondering during the opponents time... >>Crafty does the following: >>It predicts the oppononts move, assuming "optimal" play and then starts to >>work until the opponent moves. >>If it predicts the opponents move correctly it has a great edge, otherwise >>only some hashtableentries. >>If it wouldnt predict the opponents move it would gain a small contribution to >>_every_ move. >>Obviously the better it predicts the opponents move, the better is the first >>method. >>From my experience as a mere chessplayer i would say the following: >>-Predicting the opponents move is very difficult even in games of the >>highest value (disregarding trivial cases and extraordinary circumstances). >>-Intuitively i would judge a small contribution to every move as more >>worthfully than an extremly big one that occurs seldomly >>To say it exaggerated: If you have 20 moves to made and distribute >>your resources evenly, you may have a chance. If you invest all in the first >>move, making the other 19 moves very bad, you are dead for sure. >>In more general terms: >>The relative benefit of predicted moves decreases rapidly with increasing >>searchdepth, i think. >> >>Maybe one could use a hybrid approach? >>What is the reason to having this in crafty? >>Thanks in advance for any input and giving me some of your time. >> >>Oliver Roese > >Crafty, and in fact almost every chess program, uses an all or nothing approach >that works very well some of the time. Against other computers, there are times >that it works fantastically, since you can predict several moves in a row and >use essentially no time, while your opponent does all the work. > >This approach also puts pressure on human opponents some of the time, and >against inexperienced human opponents it can be unnerving. > >So all in all I'd say that it works pretty well. > >Devising an alternative is difficult. You could spend 3% of your time on each >move, but at the end you've saved only 3%, best case. Or you could pick out a >few moves to explore in more detail. If you pick the wrong moves, you are in >the same place you are with the current system, and if you pick the right ones, >you've saved maybe 30%. > >The version of mine that I took to the Hong Kong WCCC just thought for the >opponent when it was the opponent's turn to move, so it sat their filling hash >table. When the opponent moved, my program would get a little kick through the >first few plies, but nothing particularly significant. > >I think the current mechanism is good. > >bruce Hi Bruce, My program does precisely what you describe - it ponders by simply searching as if it were to play for the oponent. Have you noticed a significant improvement by focusing only on the predicted move? My reasoning in adopting this approach was that the predicted move would be the root PV move in our method, which takes the lion's share of the search time anyway, so the time investment in the other moves is only a modest, and some advantage is achieved whatever the oponent plays. Maybe I should switch to Crafty's method? Best wishes, Roberto
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.