Author: Vincent Diepeveen
Date: 17:39:40 07/30/02
Go up one level in this thread
On July 30, 2002 at 15:52:18, Ren Wu wrote:
A story regarding playing strength says really *nothing*.
DIEP was the only active playing chess program on icc that topped
all bullet rating lists. Crafty already couldn't play any human more
so strong it was at bullet.
It all says nothing. How much work you put in a game says nothing whether
the strongest program beats strong players.
Let me give you 2 examples:
- chess 40 moves average branching factor, strongest programs definitely
GM strength but real soon programs were at a decent level.
- international checkers, branching factor 10 or something (don't
blame me when it's 9 or 12). Of course with the help of a strong draughts
player i make a draughts program and directly get second. No Truus is
not actively playing anymore here. Napoleon destroys truus at tests
here at home.
When i show up, i completely outsearch opponents, even havin gmore knowledge
in evaluation (of course not so well tuned, which is the only major problem,
as we lack time to tune it all). In endgame i sometimes outsearched opponents
by 20 ply.
It took 4 years without modifying a byte to the search of napoleon before
the strongest programs, with help of some EGTBs, came close to that
search depth.
Nevertheless, some work real hard on their programs, and not a single
draughts program, despite finding tactical shots of world champions in
litterary microseconds, is capable of beating real strong draughts players.
There have been of course very little official matches, but if i see
Marcel Monteba and several others easily draw or win from the thing, i
definitely know that chessprograms are STRONGER than draughts programs.
This is real weird.
Of course a big explanation lies in the fact that with a few weeks of
programming and a analytical strong draughts player to the right of me
i am directly in the 'world top' if there is a world top anyway here.
But the obvious thing is of course that the game is much harder to play
than a general chess game, simply because in chess the CENTER is so
hard.
However mating the king is pretty hard.
In Chinese Chess mating the king is pretty easy compared to chess.
Try to mate a king in chess with just a rook and a knight without help
from your own king!
In Chinese chess this should be no problem!
The game starts in complete chaos there. So that's great for programs.
The more chaos the better. Of course both chess and chinese chess face
extensive openingsbooks. In fact nowadays the DIEP book is 750000 positions,
all given in by hand by arturo ochoa. Automatic PGN collections DO NOT
WORK anymore already for a few years in chess! If you show up at a world
championship with an automatic generated and/or tuned book, then you
are history! You need extensive manual labour here!
Saying the best players lose from a program is obviously no good commercial
for the strength of the professional players in that game, but even if
the first statement is true here, it is completely independant from how
much effort has been put in a game.
Branching factor of chess is worse than draughts. In fact any draughts program
is searching tactical far beyond the capabilities of any human. Yet human
beats the programs easily. Only those who try a tactical game against it,
they of course get annihilated.
The center is not so important in draughts. It is important in chess however,
so if a player has no good judgement there, he's already soon history.
In chinese chess the king stands on a small square, mating it is of course
a peanut for programs. Just one shot at it and there you go. Enter a rook
and a canon at the same time and it's game over.
I dare to state that tactics in chinese chess are more important than
in chess!
Best regards,
Vincent
>On July 30, 2002 at 09:57:03, Vincent Diepeveen wrote:
>
>>On July 30, 2002 at 05:57:38, Omid David wrote:
>>
>>Competition is strong there, it's not like the tictactoe section
>>such as the amazones where everyone who has a bugfree program
>>can win easily the gold medal.
>>
>>I was actually watching a few games there from Tony against one
>>of the stronger programs. Chinese chess is running a lot behind
>>on how advanced chessprograms are nowadays.
>
>Obvious this is not true. But it is understandable here. The center stage has
>been in mainland china for last few years, and it is hard for westerner to know
>what is going on there. It is all in chinese. Take a look at this site
>
>www.movesky.net
>
>it does have a english version, but unfortunetly very limited, chinese versin
>have a lot more news and events there.
>
>This is the premier site for advanced chinese chess players, include many of the
>professional masters and grandmasters.
>
>My chinese chess program, mrsj, has been playing in that site for some time. It
>has set a few records there, and it won twice for its LeiSheng Tournament, and
>is the current defending LeiSheng.
>
>That tournament works like this. Only the top 32 players in the server can
>partispate in a 7 round two-games swiss tournament, the first place finisher
>then challenge the defending champion. The time control is 30 minutes free time,
>then 1 minutes per move.
>
>From the tournaments, and all other games my program played there, All experts
>agreed that mrsj have the proformance of China's group B level. In China, the
>natinal individual tournament has been divide to two groups, group A usually
>about 24 people, and was the strongest, group B, again from 24-48 people. Giving
>the dominence of China at Chinese chess, one can safely guess that my program,
>running on a standard pc, was at around world 100th, if not 50th, place. I
>don't think that chess program is much advance than that.
>
>Computer chinese chess is not behind than what computer chess has, not even a
>inch.
>
>Ren.
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.