Author: Robert Hyatt
Date: 11:10:09 11/14/00
Go up one level in this thread
On November 14, 2000 at 13:26:31, Bruce Moreland wrote:
>On November 14, 2000 at 13:06:52, Robert Hyatt wrote:
>
>>On November 14, 2000 at 13:01:29, Bruce Moreland wrote:
>>
>>>On November 14, 2000 at 12:03:24, Robert Hyatt wrote:
>>>
>>>>On November 14, 2000 at 11:39:23, Djordje Vidanovic wrote:
>>>>
>>>>>On November 14, 2000 at 10:09:28, Kees van Iersel wrote:
>>>>>
>>>>>>I only want to show that computers can still loose games to persons who are
>>>>>>much weaker. The difference is 761.
>>>>>>How would kramnik perfome against a person with so much difference.
>>>>>>Secondly if a computer would win everything who would be interested in seeing
>>>>>>human versus machine games.
>>>>>>
>>>>>>
>>>>>>[Event "ICC 3 3"]
>>>>>>[Site "Internet Chess Club"]
>>>>>>[Date "2000.11.13"]
>>>>>>[Round "-"]
>>>>>>[White "WICKER-MAN"]
>>>>>>[Black "Rebel Tiger 13.0"]
>>>>>>[Result "1-0"]
>>>>>>[ICCResult "Black checkmated"]
>>>>>>[WhiteElo "1884"]
>>>>>>[BlackElo "2645"]
>>>>>>[Opening "Sicilian: Taimanov variation"]
>>>>>>[ECO "B46"]
>>>>>>[NIC "SI.39"]
>>>>>>[Time "23:21:26"]
>>>>>>[TimeControl "180+3"]
>>>>>>
>>>>>>1. e4 c5 2. Nf3 e6 3. d4 cxd4 4. Nxd4 Nc6 5. Nc3 a6 6. Bc4 Qc7 7. Bb3 b5 8.
>>>>>>a3 Nf6 9. O-O Nxd4 10. Qxd4 Bd6 11. h3 Bh2+ 12. Kh1 Be5 13. Qe3 Bxc3 14.
>>>>>>bxc3 Bb7 15. f3 O-O 16. a4 d5 17. e5 Nd7 18. f4 bxa4 19. Rxa4 Bc6 20. Ra1
>>>>>>Bb5 21. Rf3 a5 22. Ba3 Rfc8 23. Bd6 Qb7 24. f5 a4 25. Ba2 Bc4 26. Bxc4 Rxc4
>>>>>>27. f6 Qb2 28. Qg5 Qxa1+ 29. Kh2 Qh1+ 30. Kxh1 Nxf6 31. exf6 Rg4 32. hxg4 g6
>>>>>>33. Qh6 a3 34. Qg7# {Black checkmated} 1-0
>>>>>
>>>>>
>>>>>Hmm... yet another game that confirms Bob's hypothesis about the best programs
>>>>>not yet being strong enough to challenge GM's. I had a strong feeling of deja
>>>>>vu after going through it. It looked just like the many games I won against
>>>>>the Super Conny, Mach III, Designer 2265, Rex Chess, Genius, Fritz, etc...
>>>>>
>>>>>Don't get me wrong -- I did lose the bulk of those, but the occasional win or
>>>>>two, resembling the one above, would always bring back a dose of healthy
>>>>>skepticism regarding my initial enthusiasm and estimates of the programs:)
>>>>>
>>>>>*** Djordje
>>>>
>>>>
>>>>Another point. It is highly likely that the opponent used a computer here. I
>>>>base this on a couple of things. Near the end, there is a deep mate. He played
>>>>it _perfectly_. Which I don't think an 1800 player could do. If I were
>>>>betting, I would bet that white is a computer.
>>>
>>>Bob, you need to look at the game a little bit, I think. It was a normal
>>>anti-computer attack, and anyone could have played the game perfectly from move
>>>26 on. This was a 3 3 and white wa a human. He lost the other 7 games.
>>
>>My problem hits two ideas: 1. No tactical mistakes for 20 consecutive
>>moves; 2. he finds a mate in 7, and follows it perfectly for the next
>>7 moves. In blitz. It _could_ be a good human. But my intuition says
>>"computer" was involved.
>
>Look at the game, he plays like a human.
>
>Moves 1-15 are a sit-and-wait Sicilian. If anything he's getting chased around
>a little. His playing really solid in order to avoid having to deal with wild
>tactics, but he let black crack his pawns.
>
>16. a4 is the first enterprising move of the game, unless you count 1. e4.
>
>16. ... d5 17. e5 is required unless you want to open the game up for the black
>pieces, and it's an anti-computer move because it allows him to think about
>bisecting the board.
>
>17. ... Nd7 18. f4 is necessary to protect the pawn unless you want to play Re1,
>and f4 also continues with the idea of bisecting the board.
>
>18. ... bxa4 19. Rxa4 Bc6 20. Ra1. Maybe he could have put it on d4, which
>might let him think about swinging it faster, but I'm proabably stupid. You
>certainly can't claim that a computer picked 20. Ra1.
But I can. :) I annotated this game with a margin of .01 and Crafty didn't
comment on any move after 18. Which means it agreed 100%. Of course, I can't
figure out how much time per move to spend, so I picked a pretty small number
since this was blitz.
At move 18, Crafty likes Ba3 better, score=-.24 vs -.76 for f4.
After that, they agree.
>
>20. ... Bb5 21. Rf3. Where else does it go? e1? d1? A human is thinking g3
>eventually.
>
>21. ... a5 22. Ba3. Gotta put the bishop somewhere. He can't do f5 immediately
>or he loses e5. The QR isn't going someplace wonderful yet. The KR is on an OK
>square. So Ba3 seems obvious from a human point of view.
>
>22. ... Rfc8 23. Bd6. You can't claim this is a computer move. Put the bishop
>on a good square, annoy the queen.
It isn't a "computer move" (a move that is obviously a computer). But it is
a move that most will probably play. Mine did.
As I said before, based on one game, my conclusion is a suspicion that this
player used a computer. It is possible he did not. One game is not enough
for me to make a serious conclusion. however, 1800 players don't do this
against my program, when I play them as guests or at the local chess club.
It looks "wrong".
>
>23. ... Qb7 24. f5. Everything is solid, why not? He wants to play f6. This
>is thematic with the moves e5 and f4.
>
>24. ... a4 25. Ba2. Where else?
>
>25. ... Bc4 26. Bxc4. Black threatened to win a rook after an exchange, and
>white gets to get rid of a piece that he's not going to use.
>
>26. ... Rc4 27. f6. White has nothing to lose, and it's obviously time for
>this. Maybe black can find an improvement in the past few moves, but that's not
>white's problem, white
>
>The rest of the game isn't worth commenting on. Any player who has seen the Q+P
>mate on g7 could find the rest.
>
>bruce
>
>>It is only intuition of course... and it can definitely be wrong. I merely
>>raised the spectre. I would want to see more games myself, before I would be
>>willing to say positively...
>>
>>
>>
>>>
>>>>If I were to criticize moves, I would pick the following moves for black as
>>>>suspect (from a human perspective): 11. ... Bh2+ seems totally pointless.
>>>>23. ... Qb7 seems to be worse than pointless. This is a queen, not a bishop.
>>>>I also don't like 24. ... a4. Advancing passers is often good, but the further
>>>>they advance, the easier they are to attack and the harder they are to defend.
>>>>White is attacking on the kingside. Black really doesn't need to waste time
>>>>on the queenside just yet, when he has no pieces for defending the kingside.
>>>
>>>I haven't looked at it with a computer, but at the point white plays 24. f4,
>>>he's out for blood, and the problem black has is avoiding getting mated on g7.
>>>It looks like it's pretty hard to stop.
>>>
>>>>However, on the other hand, Tiger was playing a computer in human clothing.
>>>>Nothing good can come of that, and drawing conclusions is harder. I would
>>>>_never_ believe than an 1800 player can beat today's programs. yes, it might
>>>>happen once in every 1000 games. But that is close enough to zero to instantly
>>>>turn on warning lights when I see it. In this case, if you analyze the game
>>>>with another program (I used Crafty) it couldn't find any improvement for white
>>>>from move 18 on. Which is _very_ suspicious...
>>>
>>>White's moves are all either forced or typical human moves.
>>>
>>>bruce
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.