Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Horizon Effect: Worse for Humans?

Author: Robert Hyatt

Date: 08:14:50 10/24/02

Go up one level in this thread


On October 24, 2002 at 07:37:34, Marc van Hal wrote:

>On October 23, 2002 at 20:33:36, Robert Hyatt wrote:
>
>>On October 23, 2002 at 15:30:58, Marc van Hal wrote:
>>
>>>On October 23, 2002 at 11:14:47, Robert Hyatt wrote:
>>>
>>>>On October 23, 2002 at 08:09:30, Otello Gnaramori wrote:
>>>>
>>>>>On October 22, 2002 at 17:32:52, Robert Hyatt wrote:
>>>>>
>>>>>>>
>>>>>>I doubt Fritz saw more.  It was just lucky that the sacrifice didn't work.
>>>>>
>>>>>I was convinced that in this game the "lucky factor" don't exist...
>>>>>
>>>>>w.b.r.
>>>>>Otello
>>>>
>>>>
>>>>I don't know what you mean, exactly.
>>>>
>>>>My point was this:
>>>>
>>>>If Nxf7 works, deep fritz had absolutely no idea that it did, because it
>>>>couldn't see the
>>>>consequences of the move.
>>>>
>>>>if Nxf7 doesn't work, the same statement is still true.
>>>>
>>>>So the issue becomes one of luck.  Kramnik played the move.  Fritz had no idea
>>>>whether
>>>>it was good or not.  Luckily for Fritz, it turned out to be bad.
>>>>
>>>>A different position.  Fritz (or any program) could have been playing against
>>>>Shirov a
>>>>few years ago where he made his famous bishop sacrifice in an endgame.  The
>>>>programs
>>>>all think the sacrifice is unsound.  They are all wrong.  So in that case, since
>>>>they blundered
>>>>into a position where a sound sacrifice worked, but they had no idea about the
>>>>sac, they
>>>>were unlucky.  In the Kramnik game, Fritz blundered into a position where the
>>>>sacrifice
>>>>was played and it was lucky that it was unsound.
>>>>
>>>>that was the idea...
>>>>
>>>>Luck is _always_ a part of the game...
>>>
>>>Steinitz and Dr Lasker strongly would disagree with you
>>>(Acording to them nothing was as bad as a bad sacrefice.)
>>>Though Tal and Shirov would agree a litle.
>>>You also have the bad knight sacrefice from Fischer against Donner.
>>>Then you have to be carefull because when Donner asked why he made this
>>>sacrefice.
>>>Fischer said because I despise you.
>>>Maybe this was what Kramnik ment and only didn't say it because he knew Fritz
>>>wouldn't understand it ouch.
>>>Still I think the horizon problem is a computer problem and not a human problem.
>>>Marc
>>
>>
>>There are two kinds of "horizon" problems.
>>
>>In one, you simply don't see deeply enough, and so you enter into a line that
>>gets you killed.
>>
>>In the other, you play delaying moves untkil you get tired of doing so, and
>>then you stop analyzing down that path.  This is how the infamous "horizon
>>effect" hurts programs still today...
>>
>>The latter is really not a human issue at all.  But the first one is...
>
>Ok ehm your right
>But the first line did not aplie here
>(This aplies when you get  over played which was not the case here.)
>But when the sacrefice was corect and Fritz didn,t find it than it was
>mostlikely because of the infamous horizon problem.
>Marc


Agreed.  _many_ real sacrifices, that are perfectly sound, can't be found by
programs.  They
are simply too deep.



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.