Author: J. Wesley Cleveland
Date: 11:07:45 07/30/99
Go up one level in this thread
On July 30, 1999 at 10:45:30, Shep wrote: >On July 30, 1999 at 09:05:43, Robert Hyatt wrote: > >>I don't really know... my 'feeling' is that as we get more we will hit them >>more. IE it is pretty hard to get to 3 pieces from 10 pieces, because that >>requires a lot of deep trades/captures... I suspect it is easier to get to >>5 from 12, than to 3 from 10, for example. So getting to 6 from 13 ought to >>be even easier, and you might be right. > >[snip] > >>So you might be right. 12-14 seems common in blitz. In a standard game, I >>suspect that this might stretch out a bit more...: > >Thanks for the interesting info! > >The point I am trying to make is that if this is a typical phenomenon - i.e. if >an N-man TB is not just hit a lot from 3+N but maybe from 2*N or higher -, we >may be "a little" closer to perfect play because we will certainly not need the >infamous "32-man TB". > >If for example a 15-man TB is already hit a lot from positions with 22-28 >pieces, we are many orders of magnitude better off. > >Now 15-man may still be out of reach for a looong time, but maybe 10-man is >possible within the next 50 years; that would make middlegames really tough for >humans if the machine starts hitting the TB's in situations with 18-25 pieces... I have been thinking about something similar. In 8-10 years, when we should be able to have 6 man TBs, billion entry hash tables, and the processor speed to fill them, computers will begin to play endings perfectly. At that point, we might want to change the evaluation function to virtually a null function, and brute force it.
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.