Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Quality of Chessbase full analysis feature using Shredder 7.04

Author: Mike Hood

Date: 09:58:54 12/21/03

Go up one level in this thread


On December 21, 2003 at 01:02:02, Mark Young wrote:

>On December 21, 2003 at 00:03:28, Robert Hyatt wrote:
>
>>On December 20, 2003 at 22:45:43, Mark Young wrote:
>>
>>>On December 20, 2003 at 22:21:22, margolies,marc wrote:
>>>
>>>>
>>>>I think this poster committed an ergo propter hoc fallacy.
>>>>Why should it be surprising that Kasparov's analysis comports with Shredder's
>>>>analysis? We should actually be surprised if the opposite were true--
>>>
>>>You need to read my statement again. I said Shredder found shots that Kasparov
>>>missed. And the computer seems to be correct. This is what I found surprising
>>>and made me take notice, but the computer is far from perfect.
>>
>>The one thing they all miss is some human-like commentary.  IE programs
>>that try to insert text messages into the commentary look like a room full
>>of monkeys with a typewriter.  They produce something that looks sensible
>>on a quick look, but when you read it carefully, it is just "static".
>>
>>Computers are very good at giving scores and variations.  But there ability
>>to say "of course not Nxf3?? as that hangs the queen to this five-move threat:"
>>and the like...
>>
>>IE I have seem some really comical comments from Chessmaster, when it attempts
>>to do just this...
>
>Chessbase is not like Chessmaster when giving full analysis. I know what you
>mean about chessmaster, thats why I never really looked at this feature in
>chessbase. Chessbase is much more in the style of chess Informant with not much
>text at all. Very easy for us "old school" chess players to read and understand
>as written text. Chessbase stays mostly with what computers are good at as you
>pointed out above, but it will show you "of course not Nxf3?? and why.  I will
>give an example below.
>
>[Event "16"]
>[Site "London m4  ;HCL 18"]
>[Date "1834.??.??"]
>[Round "62"]
>[White "Macdonnell A"]
>[Black "De Labourdonnais L"]
>[Result "0-1"]
>[ECO "B32/06"]
>[Annotator "Shredder 7.04 (180s)"]
>[PlyCount "74"]
>[EventDate "1834.??.??"]
>
>
>[d]3b1r1k/3P2pp/8/p7/2Q1p3/8/PP1p1pPP/2R1qR1K w - - 0 34
>
>34.Qc5
>[34.Rfxe1?? is not to be advocated because of the following mate in 2 fxe1Q+
>35.Rxe1 dxe1Q+ 36.Qf1 Qxf1#]
>
>Chessbase analysis of  move 34 Macdonnel - De Labourdonnais 1834. Chessbase is
>saying it agrees with 34. Qc5 as the best move and give a simple "trap" to be
>avoided.
>
>This is just a quick example to address the point you made above.

You're right, Mark, Fritz and his brothers do give good human-sounding
commentary in short games. The problem is when the games are longer. Then the
comments begin to sound repetitive.



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.