Author: Christophe Theron
Date: 07:56:19 05/21/00
Go up one level in this thread
On May 21, 2000 at 03:51:03, Mogens Larsen wrote:
>On May 20, 2000 at 23:44:52, Christophe Theron wrote:
>
>>I wouldn't either.
>>
>>The computers do not have the BOOKS when they participate. They just have
>>memorized opening theory in their BRAIN. That's exactly what a human player is
>>allowed to do.
>
>There's a small but significant difference, but it's impossible to make the
>concept completely identical due to the obvious differences between computers
>and humans. The same goes for the eternal question of "fairness".
>
>>But if you want to get an idea of progress in software, just take the best PC
>>program of 10 or 5 years ago and let it play against the current best PC
>>programs.
>>
>>You'll realize how much software has improved.
>
>I'm sure they have improved a lot, but that doesn't necessarily tell you about
>the innovative progress made in the last 5 years or so. How long did it take for
>you to close the gap to the leading chess programs when you went pro?
I went pro once I have closed the gap, so I'm not sure I can answer your
question.
Anyway, don't forget I'm working on Chess Tiger since 1982...
>Crafty is also closing the gap faster without changing the hardware a lot, even
>though Dr. Hyatt isn't an amateur, the program itself is experimental by nature
>as I understand it. More experimental than its competition I would imagine.
I tend to think it's the opposite.
Last time I read Crafty's sources, it was 3 or 4 years ago (version 9.x), I was
very disappointed because I have seen only very academical, classical,
techniques implemented.
In comparison, my program was using a lot more of different and risky pruning
techniques, for example.
I do not know if it is still the case, but I think that to be on top you have to
be extremely creative and invent original techniques.
I personally spend 80% of my time on Tiger trying new ideas and techniques. And
of course, I have to reject most of them. But a few of them remain.
>The number of strong or semistrong WinBoard engines also tells me that it's
>easier now than it ever was to create a competitive chess program. I'm not
>trying to belittle the work effort of programmers, just giving you my impression
>so far.
It is indeed much easier. You just connect on the net, download Crafty and
several other free chess programs with their source code, and work from that.
Great, but the drawback of this is that programmers do not have to reinvent
everything. That makes them more lazy, and I'm sure a lot of creativity is lost.
New chess programmers are so busy implementing alphabeta, null move, extensions,
hash tables and bitboards, that they probably have no time left to invent any
other new techniques. Because once they have implemented the above known
recipes, they have a program that is between 2300-2500 elo. Is there any
motivation left to spend 2 additional years or hard work to invent something
else that will bring only 50 additional elo points? At that point it is more
tempting to write a flashy graphical interface, for example. Or a good Winboard
interface, or...
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.