Author: John Merlino
Date: 15:31:12 08/01/02
Go up one level in this thread
On August 01, 2002 at 18:18:43, Tony Werten wrote:
>On August 01, 2002 at 18:13:14, John Merlino wrote:
>
>>On August 01, 2002 at 18:12:17, Tony Werten wrote:
>>
>>>On August 01, 2002 at 16:18:17, John Merlino wrote:
>>>
>>>>On August 01, 2002 at 16:13:18, Gilles Garrigues wrote:
>>>>
>>>>>Take BigBase 2002 (more than 2,000,000 games, compiled by ChessBase GmbH). Use
>>>>>ChessBase 8 to open it and click on "statistics".
>>>>>
>>>>>You end up with NEGATIVE average Elos for Black and White. The progress of the
>>>>>stats is OK until about game 1,700,000 where a sharp drop occurs (as if one or
>>>>>several games had huge negative Elos).
>>>>>
>>>>>Has someone a good explanation for that ?
>>>>>
>>>>>--
>>>>>Gilles Garrigues, Dunkerque, France
>>>>
>>>>Signed vs. Unsigned integer. The highest bit of the integer is 1 when:
>>>>
>>>>1,700,000 (games) * 2369 (average ELO) ~= 0xF000000
>>>>
>>>>jm
>>>
>>>Let me guess. This won't happen (anymore) in Chessmaster ?
>>>
>>>Tony
>>
>>What do you mean? We don't have an average ELO stat for a database in
>>Chessmaster.
>
>Ah, sorry. I saw the numbers and thought "he has seen this bug before". I
>guessed it was in your own program.
>
>Tony
It's a very common bug, and it has occurred in various guises in Chessmaster
over the years. It's basically just short-sightedness on the part of the
programmer ("nobody will EVER have a 2-million game database!").
jm
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.