On Mar 6, 8:43 am, mikeh...@[EMAIL PROTECTED]
wrote:
> On Mar 6, 6:35 pm, mark <mschr...@[EMAIL PROTECTED]
> wrote:
>
> > Kramnik played a weak program on a slow computer while getting help
from the computer and he still lost. That was in 2006 and the computer's
rating only go up with time.
>
> the computer was not slow, it was an octal. also realize that fast
> hardware help blitz ratings far more than ratings at slow time
> controls (the reasons of which should be obvious).
The computer was not an octal. The computer was a Dual Intel Core 2
Duo 5160. See www.chessbase.com/newsdetail.asp?newsid=3504
The rule is double the speed the program improves 75 points in games
of 40 moves/120 minutes. For blitz games doubling the speed increases
much more than 75 points.
> > Which wiki article on shogi? Do you mean the Wikipedia encyclopedia
article on shogi? They talk about the game between Watanabe and Bonanza.
Watanabe did not rate the computer highly.
>
> yes, the notable quotes are:
>
> "Watanabe commented after the game that he should have lost if Bonanza
> had played a defensive move before entering the endgame."
Watanabe never said that. I read the translated comments of Watanabe.
Watanabe said "The variations are long, but a strong player will spot
this repeat attack theme immediately. In this position, almost
anything is good as long as White doesn't give Black a knight. The
computer, on the other hand, has to consider all the possibilities
over a wide range, which might make it difficult for it to focus on
the key point and probe it deeply. In this respect, human experience
still seems to be superior."
Other people said that means if the computer made different moves the
computer would have won. Hey, I could say the same thing. If I had
made different moves I also would have won. But the moves were way too
hard for the computer to find. Watanabe was never in any danger from
the computer.
> "This game can be looked at as a clear example that computer Shogi,
> although still not perfect, has reached top pro level."
That's not a quote from Watanabe. Watanabe said, "they are getting to
be a match for professionals" not top pro level.
The wikipedia article made another mistake. The article says the
computer has an "awesome endgame ability". Watanabe said the opposite
"I learned from this game is that endgame technique apart from tsumi
is still a challenge for the computer."
The translation was by Richard Sams. It can be found on The Shogi
Discussion List (SHOGI-L).
See
http://lists.topica.com/lists/shogi/read/message.html?mid=812678696&sort=d&start=1961
and
http://lists.topica.com/lists/shogi/read/message.html?mid=812678824&sort=d&start=1961
This wikipedia paragraph in the shogi article exaggerates the strength
of computer shogi.
The Bonanza Watanabe game was a mismatch. Before you play a top
professional you should be qualified by winning a game against a weak
professional. No computer has ever won against any professional. The
best way to see how strong the computer is by playing in human
tournaments. Until the computer beats a professional we can not say
the computer has reached professional level.
> > That's counterintuitive. Why doesn't the 17 ply set a trap that the 14
ply won't see because of the horizon effect? The 17 ply should win.
>
> no, it's not counterintuitive at all. the reason is simple: xiangqi
> is simply not a deep/complex enough game such that 17-ply NC can win
> against its 14-ply counterpart, provided a good opening is chosen (for
> black, as it's the 14-ply side). i realize this claim may be
> controversial and may not sit well w/ the xiangqi community
> (particularly w/ xiangqi grandmasters), but i'm extremely confident
> it's the truth. in other words, the horizon effect is far less
> pronounced in xiangqi than in chess. no such deep trap exists in the
> game (contrived examples aside).
>
> consider the following facts:
>
> 1. xiangqi has far fewer playable openings than chess.
> 2. xiangqi has fewer attacking pieces than in chess (attacking here
> means pieces that can cross the river), 11 vs 15.
> 3. the advantage of the having the first move in xiangqi is weaker
> than in chess. my analysis has shown red's opening advantage is
> ~1/6th of a soldier. in chess, white's advantage is ~1/3th of a pawn.
> 4. it takes far more material deficit to "ensure" a win in xiangqi
> than in chess. going by neuchess eval, a score > ~ +6 needs to be
> obtained before victory is ensured. even if we're conservative, that
> means u need an advantage equivalent to approximately a horse (usually
> valued at +4). in chess, +2 (i.e. 2 pawns) is usually enough.
> sometimes 1 pawn is enough.
>
> factoring in all of the above together, and it's pretty clear that
> xiangqi is a far less complex game than chess. far less complex means
> it's gonna be much easier to draw even if your opponent is neuchess.
> or god.
You say as the games get simpler it easier for the world champion to
draw. Take Othello, it is simpler than xiang qi. In 1999 a computer
beat the world champion 6 to 0. Here we see a simpler game but the
world champion cannot draw the computer. More complex games like xiang
qi are even harder to draw the computer.
What about the draw odds handicap match played by Rybka? Rybka played
an 8 game match with GM Joel Benjamin at Draw Odds. Every draw for
Benjamin counts as a point. For GM Benjamin no difference between a
win and a draw. All he needed was 5 draws to win the match. And
Benjamin has white in all games. Rybka won 6 games and drew 2 games so
Rybka won. See www.rybkachess.com/index.php?auswahl=Events
2 examples where it is hard to draw with a superior opponent.
|
