Author: Albert Silver
Date: 16:02:55 11/27/01
Go up one level in this thread
On November 27, 2001 at 18:25:27, ALI MIRAFZALI wrote:
>On November 27, 2001 at 18:20:35, Kurt Utzinger wrote:
>
>>You are right: you can test 100 actual programs and none of them will play
>>12...Qxh3. But it's a great difference between a set up position and a real
>>game. I assume such faults may be the result of a hashtable collision.
>>Kind regards
>>Kurt
>I do not think it is the result of a hash table collision.I did not set up the
>position when analyzing with Zarkov 2.6 .It was in continous analysis mode.
You'd be surprised. I saw a report in CB Magazine about a year or two ago, when
several games were played between some German GMs and IMs and Fritz 5 or 6 (I
don't remember which). The operator was none other than Mathias Feist or
Wuellenweber (again I don't remember which, though I believe Wuellenweber). At
the critical moment Fritz uncorked an absolutely unbelievable blunder,
squandering away an easy win, to let the player off the hook. The engine could
never be made to play the move at any depth, yet it unquestionably did so. This
was in the hands of CB's own people, and reported by them. I could try to dig it
up, but believe it was during one of the Frankfurt events.
Albert
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.