Author: Christophe Theron
Date: 11:09:15 08/25/01
Go up one level in this thread
On August 25, 2001 at 02:49:00, José Carlos wrote:
>On August 24, 2001 at 13:01:10, Christophe Theron wrote:
>
>>On August 24, 2001 at 12:39:35, José Carlos wrote:
>>
>>>On August 24, 2001 at 12:33:26, Christophe Theron wrote:
>>>
>>>>On August 24, 2001 at 11:41:47, Harald Faber wrote:
>>>>
>>>>>With these losses it is justified that Tiger didn't become single world
>>>>>champion. What is your explanation for this result? Did you play GT in
>>>>>aggressive mode? What made Tiger lose especially the last game vs. an amateur?
>>>>
>>>>
>>>>
>>>>I can answer your question if you answer mine:
>>>>
>>>>When you flip a coin, why is it head sometimes, and some other time it is tail?
>>>>
>>>>
>>>> Christophe
>>>
>>> Are you saying that luck is the only reason? I should consider then
>>>participating with my engine... Sometimes it would be head, and sometimes
>>>tail...
>>>
>>> José C.
>>
>>
>>
>>Absolutely.
>
> Be serious. What you (and everyone) do to improve your program is exactly
>answer Harald's question. What was wrong in a lost game. When you know it you
>can fix it. The reason why you have a so strong program is that you're very
>smart finding those hidden reasons, and of course avoiding them to happen again.
But that's not how I work.
I have explained several times now that I never try to fix a lost game. I just
look quickly at the games and I try to do something only when I see a repeated
pattern. Then I work not on a single game, but on a collection of positions
where the pattern exists.
>The result of a chess game is not a coin flip at all. Chess is a game with a
>very small set of rules and with complete information. There's always a reason
>for everything inside a chess board.
That's only the surface.
The real thing behind this is that chess, in our practical world, is a
probability game.
During a game, a player or program has to take thousands of decisions without
being able to see the consequences of them.
Sometimes a decision is right, sometimes it is not.
All you can do is make your best so the decisions you take are more often right
than wrong.
That's why a game of chess is a coin flip. You try to make the coin as biased as
possible, but it's still a coin flip.
>>I also wonder why nobody asked Amir what went wrong with Junior 7 when the SSDF
>>match against Tiger began with 5-1 in favor of Tiger...
>
> That would be a good question if asked for the 6 games out of context. Every
>loss of Junior there is a good test for Amir, to try to avoid whatever happened
>there again.
>
>>It seems that people are now educated enough to understand that anything can
>>happen even in a 10 games match, but not enough to understand that one game
>>means very little. Strange, isn't it?
>
> I don't think Harald was implying that Tiger is weaker than the opponent for
>losing that game. He was asking for the reasons why Tiger lost that game. An
>answe to this question can be very simple or very sutile: a tactical mistake in
>move xx, a hole in the evaluation in move xx, a bug in the search (these are
>simple reasons).
>
>>Or maybe it's because the game was played during the WMCCC? Naturally games
>>played in these magic events do not follow the basic rules of statistics.
>
> You're angry because you lost. You shouldn't, because you have an excellent
>program. And people know.
I'm not angry at all. I'm just saying over and over again what I am saying since
a long time.
I have not changed what I say, but you interpret it differently now.
>People
>>expect that a WMCCC game is 100% accurate, and that the strongest program (even
>>if it is stronger only by 10 elo points) is going to win.
>
> Experts don't expect that.
>
>>To the "what went wrong..." question, I think I can answer now: what's wrong is
>>that people give so much importance to one game.
>
> Harald dindn't say the game was important. He only asked what was wrong.
>
>>Sorry Harald, but your question has been getting on my nerves. I'm still amazed
>>that you asked...
>
> You shouldn't. That was a good question to improve the program.
As I said earlier, that's not how I work.
Christophe
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.