Author: Roy Eassa
Date: 14:30:06 08/27/01
Go up one level in this thread
On August 27, 2001 at 13:54:23, robert flesher wrote: >I find this VERY VERY hard to believe! Chess tiger 12.0 solves this very easy >and never reproduces these bad moves. Try turning off use of table bases. Tiger >12.0 knows that white is winning almost instantly after the rooks come off the >board. Please re-read the original post. My _human_ opponent (on ICC), not Tiger, played the actual weak move(s). However, Tiger's analyses and evaluations were quite suspect in the early moves. (In the next several moves, white's evaluation indeed rose steadily.) My post gave very specific evaluations, PASTED from Gambit Tiger 2 running on my 1.2 GHz Athlon. I also looked at the same positions with Chess Tiger 14 and it gave very similar results. BTW, I was NOT using tablebases. I did not spend enough time with either Tiger to let it see significantly beyond 15 ply deep. I have no doubt that any decent program can EVENTUALLY calculate that white is winning in the no-rooks position. My complaint was that the two Tigers (and also Fritz and Junior, but Shredder was a little better) did not UNDERSTAND the endgame relatively early on in the evaluation, and that Tiger actually DID indicate in its PV that black should recapture with the rook (although when that position became the "current" one, it did select the correct recapture). This was the first position I'd ever encountered where I, a human rated under 2100 ICC, understood at a glance something that took my fast computer running several of the best chess apps, many seconds to get. I honestly didn't realize until then that there were such "obvious" gaps in the programs' "knowledge". [Certainly no offense to the authors of these great programs: each app would probably win a 100-game match with me with approximately 100-0!]
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.