Author: Albert Silver
Date: 08:12:33 08/07/04
Go up one level in this thread
On August 07, 2004 at 09:53:25, Rob Basham wrote:
>On August 07, 2004 at 08:51:51, Thomas Lagershausen wrote:
>
>>[D]3r2rk/3n1pp1/2p1b2p/3q3P/pp1PNQ2/2P2P2/PP6/KB4RR w - - 0 29
>>
>>In this speedchessgame in round two of the fide-wcc 2004 the IM Neelotpal
>>(2457)found with 29.Rg1-g6 !! with the thread to sacrifice the rook on h6 the
>>strongest move to show that white is not(!) worse in this position.
>>
>>I bet that every computer of the world wouldn´t have found this in a compareable
>>time.
>>
>>So this is a lesson in tactics where computers can learn form human players.
>>
>>Do you agree?
>>
>>TL
>
>This move...Rg6 is too "risky" for computers to find...
>Even with rooks at g1 and g2, the best move is Nf6!
>
>If this GM won by playing Rg6, I think he (or she) was just having a lucky
>day.....:-)
>
>However, I agree with you, it is a great surprise move, but white's evaluation
>drops emediately!
>
>But like many of the Paul Morphy "great moves"....most chess computers just do
>not find them
>(because they are too risky)!
>
>Rob
I'm a bit confused by the post. How is the move risky? This is a clear winning
tactical blow. After this play, white is simply winning. Risk implies
uncertainty, and here there is no uncertainty.
As to showing that computers can learn tactics from humans.... well, it's nice
that we have a position where the human vision led to the correct answer
quicker, but this isn't new really. There are lots of examples of tactics such
as some of the Nolot positions, where the programs were only able to find the
same winning sequence in at least 10 times the time the human player did.
However, this makes a good test position for the bots in any case.
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.