Author: Robert Hyatt
Date: 08:26:17 02/09/00
Go up one level in this thread
On February 09, 2000 at 05:06:29, Colin Frayn wrote: >>>Does anyone seriously believe what he said about computers not playing better >>>with increased ply, or at least asymptotically slowing down? I have 600 games >>>against Crafty which prove the opposite. >> >>I didn't say that. I said "there is plenty of evidence that says that an >>additional ply is _always_ good." Either you misunderstood, or I did a >>horrible job of answering that question. > >Sorry - no I worded that badly. I _know_ that you don't believe it, it's just >that you hinted that others _do_. Others _do_ believe it. You read that here all the time. They always offer playing program X vs program X with 1 more ply, and show a curve that indicates 'diminishing returns'. Monty/I (and Ernst) tested this differently and we decided that this doesn't seem to be happening, at least to depths we can reasonably reach. I'd like to have DB's hardware for a couple of months and take my 15 ply searches out to 20+ to see if things change out there. > >>>Thanks for the advice, Prof. Hyatt. That world title's mine next time :) >>>*grin* >> >>Exactly how are you going to arrange for that "good luck" factor you need so >>badly? :) > >Luck? Bah! I just need a 256 processor supercomputer. Not a problem :) >By my meagre mathematical calculations and a little experimenting, I reckon that >should give me the extra 700 ELO-ish needed to be in with a chance of a win :) > You will be very surprised. :) I played many events with just such a handicap, and lost a game here and there. Some due to bugs... some due to bad luck in choosing openings, some due to bad luck in general (IE we follow a path that _both_ programs decide is winning for me, only to reach a position a few moves later where there is a tactical shot neither saw that turns the game around.) Luck is present. I had my fair share (or more than my fair share) over the past two weekends... >If anyone wishes to donate me said supercomputer then feel free to email me on >the above address. > >Cheers, >Colin > >P.S. Never underestimate beginner programs - no matter how bad they are now :) I _never_ do. :)
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.