Author: Enrique Irazoqui
Date: 06:28:24 08/28/00
Go up one level in this thread
On August 28, 2000 at 07:10:27, Ed Schröder wrote:
>On August 28, 2000 at 06:28:08, Uri Blass wrote:
>
>>On August 28, 2000 at 06:02:23, Marcus Kaestner wrote:
>>
>>>>>>>>Which program is best in playing human players? Shredder 5.0
>>>>>>
>>>>>>no, no, no. much better is rebel.
>>>>
>>>>Oh? How was this determined? I agree Rebel is great at playing against human
>>>>opponents, but 'much better'? I recall that Shredder's results in the Israeli
>>>>leagues were not too bad.
>>>
>>>have you ever seen rebel was beaten in a human-comp comparison by another
>>>program in the last years?
>>>neither in israel, nor at the aegon tournament.
>>
>>The aegon tournament was a long time ago so I cannot count it today.
>>
>>Rebel did the best result in Israel against humans but you can blame the
>>operator of shredder for this fact.
>>
>>If you compare Junior's results in dortmund with other results you can see that
>>Junior did the best results so the picture is not clear.
>>
>>Rebel may be the best against humans but I am not sure about it.
>>
>>The meaning of being the best against humans is also not clear because it is
>>possible that one program is best in winning weak players when another program
>>is best in playing against strong players.
>>
>>>have you played yourself many programs?
>>>i can say, that i sometimes can survive against tiger, seldom shredder and
>>>fritz, but i have not a single draw against rebel. only losses!
>>
>>It proves that Rebel is the best against you but it does not prove that Rebel is
>>the best against humans.
>>
>>Uri
>
>I think you are missing an important point. Have you ever seen Rebel
>strangled in the last 25-30 GM games as we have seen in the dutch
>championship and at Dortmund?
>
>Ed
It sure got strangled. Look at 21... e5?? and what happens after.
[Event "Monthly GM Challenge"]
[Site "?"]
[Date "2000.??.??"]
[Round "7"]
[White "Scherbakov, R."]
[Black "Comp Rebel Century"]
[Result "1-0"]
[ECO "D17"]
[WhiteElo "2540"]
[BlackElo "2500"]
[Annotator "Scherbakov,R"]
[PlyCount "111"]
[EventDate "2000.??.??"]
1. d4 {My preparation for the game was not quite usual - I refreshed in memory
all possible gambits which are known to be dubious, including 1. d4 d5 2. c4
e5?! and 1. d4 d5 2. c4 e6 3. Nc3 c5 4. cd5 cd4?! Of course, Chigorin Defence
which Rebel played in the previous game was not forgotten...} 1... d5 2. c4 c6
3. Nf3 Nf6 4. Nc3 dxc4 {This time Rebel prefers solid approach.} 5. a4 Bf5 6.
Ne5 (6. Nh4 {After which I played a couple of times, White sometimes should
move forward all kingside pawns to fight for advantage - too risky against
Machine.}) 6... e6 (6... Nbd7 {
I was hoping for Morozevich's to show one interesting idea. Next time...}) 7.
f3 Bb4 8. Bg5 {I had prepared the Bishop's sortie a couple of months ago
exactly for the game against Mikhail Kobalia in Russian Cup Final. The main
idea was to confuse my opponent with rare variation but nevertheless Black
should know what to do to obtain a good play. Of course, it's impossible to
take the Machine by surprise but I had already no good alternative...} (8. e4 {
In the main line} 8... Bxe4 9. fxe4 Nxe4 10. Bd2 Qxd4 11. Nxe4 Qxe4+ 12. Qe2
Bxd2+ 13. Kxd2 Qd5+ 14. Kc2 {White probably has small advantage but it's too
complicated to play against Machine. There are a lot of tactics, white King is
not safe so I had no intention to go here.}) 8... h6 9. Bh4 c5 (9... b5 $5 {
I expected} 10. e4 Bh7 11. Be2 Qb6 12. O-O $44 {with good compensation for the
pawn as was in R. Scherbakov - M.Kobalia, Ekaterinburg 1999. I was not much
afraid to sacrifice a pawn in this situation. Position is very complicated but
in my opinion strategy prevails tactics here. Of course there are a lot of
tactics but first of all both sides should play by plans rather than by
concrete operations. I saw a couple of Rebel's games and have a strong feeling
he much more likes strategically clear situations.}) 10. dxc5 Qa5 (10... Qxd1+
{The alternatives} 11. Kxd1 {with e2-e4 soon}) (10... Qd5 $5 {
and lead to the endings with small advantage for White.}) 11. Qd4 Nc6 (11...
Bxc5 {Another possibility is} 12. Qxc4 O-O 13. e4 Bh7 14. Nd3 Be7 {
, the game Yusupov - Timman, Reykjavik 1988 continued with} 15. Qb5 Qc7 16. Bg3
Qc8 $11 {and Black has obtained good play thanks to the idea Nc6-d4. White has
not finished the development yet and his Queen is too advanced.}) 12. Nxc6 bxc6
13. e4 (13. Bf2 $6 {
looked too dangerous as Black will have important resource e6-e5 in the future.
}) 13... Bxc5 (13... Bg6 {Black has usually played after which White has the
possibility to defend the pawn with} 14. Bf2 {Black will capture pawn c5 soon
with Nd7 so the question is how effectively White can exploit passive position
of black Bishop g6. I suppose White can hope for a small advantage.}) 14. Qxc4
Bg6 15. Qa6 {Otherwise White has problems to finish the development.} 15...
Qxa6 (15... Qc7 {After I was going to play} 16. Bg3 e5 17. Qc4 $5 {
pushing the Bishop from active position:} 17... Bd6 (17... Bb6 {
is worse because of} 18. a5 $5 Bxa5 19. Qc5 Bxc3+ 20. Qxc3 Nd7 21. Ra6 Rc8 22.
Qa3 Ra8 23. Bc4 $36 {with strong initiative.}) 18. Bf2 Rb8 19. Qa2 {
with next Bc4 with advantage.}) (15... Qb6 {is also in White's favour after}
16. Qxb6 axb6 17. Bf2 $14) (15... Qb4 $6 {I was slightly worried about but
objectively this should give a big advantage for White. I could probably play}
16. Qxc6+ Ke7 17. Qb5 $5 (17. Qc7+ {
should also be good for White but position is not so clear after} 17... Kf8 18.
Rd1 Kg8) 17... Rab8 (17... Qd4 $2 18. Qb7+) 18. Qxb4 Rxb4 19. Ra2 {
and White has good chances to extinguish Black's initiative little by little.})
16. Bxa6 Rb8 17. Bxf6 {This decision was not easy but perhaps it's forced.
Till this moment Rebel played almost instantly while I have spent
approximately 1.20 but now he surprisingly started to think which made me
panicked for a while. Is he thinking about 17. ..Rb2?! Fortunately it's just
bad for Black.} (17. O-O-O $6 {After Queens swapping I felt much better but
now realized that I cannot defend pawn b2 comfortably with because of} 17...
Nd5 $1 18. Be1 (18. exd5 $6 {in case of White could only get problems after}
18... Be3+ 19. Rd2 Bxd2+ 20. Kxd2 Rxb2+ 21. Ke3 exd5 {with next Kd7 and Re8.})
18... Ne3 (18... Be3+ $5 {the manoeuvre} 19. Kb1 Bf4 $5 {looks promising}) 19.
Rd2 {and now immediate} 19... f5 $5 {deserves attention with excellent play.
The Knight can be supported with f5-f4 if required. Black can also double his
Rooks along b-file.}) (17. Nd1 {I thought about but it looked too
sophisticated. The main strategic idea to exploit position of Bishop g6 could
hardly be realized here as it's impossible to prevent from f7-f5. Besides, the
lack of development would cause troubles for White. Surprisingly it happened
to be the game Adianto - Kramnik, London (rapid) 1994 which continued with}
17... Nd7 18. Rc1 Bb4+ 19. Kf2 Nc5 20. Be2 Nxa4 21. Rxc6 Bc5+ 22. Ne3 O-O {
with clear advantage for Black.}) 17... gxf6 18. O-O-O Ke7 19. Kc2 (19. Bd3 $6
{I have spent a couple of minutes for} 19... Rb4 20. Bc2 $6 {but it's just
pointless as the Bishop cannot defend both f5 and b3 square anyway.
Furthermore this waste of time could let Black to take the initiative with}
20... Bd4 $36) 19... Rhd8 $6 {
The first and very nice surprise. I could only dream to swap a pair of Rooks.}
(19... Rb4 20. b3 Rhb8 {with next f6-f5 was quite acceptable for Black.} (20...
f5)) 20. Rxd8 Rxd8 21. Bd3 {
The main idea of the whole line - to keep the Bishop on g6 out of play.} 21...
e5 $4 {Unbelievable!!! Obviously Rebel did not consider seriously White's next
move after which Black is practically a piece down.} (21... f5 $5 {It was bette
r to make almost any other move (or even don't move on the whole!). Still was
not so bad as after direct} 22. exf5 Bxf5 23. Bxf5 exf5 {
Black has active pieces and good play on the kingside.}) 22. g4 $1 {The Bishop
on g6 is now a "big pawn". The attempt to escape with f6-f5 is pointless.}
22... h5 23. h3 h4 $2 {This makes White's life easier. It was much better to
keep the Rook on h8 or to move the King to g5 - White had to keep one of piece
(R on h1 or N on e2) in defence so it was more difficult to break the
queenside.} 24. Rd1 {I was thinking about other possible plans, for example
a4-a5 then Ra1-a4-c4 looked promising but the intuition prompted me it would
be not so big task to swap the Rooks on d-file.} 24... Rd4 {Yes! Instead of
this pointless move Black should keep the Rook somewhere on b8 making more
difficult White's task on the queenside. I was not much worried about possible
a7-a5 as after b3, Bc4 Ne2-c1-d3, Rb1, Kc3 White will break with b3-b4 anyway.}
25. b3 Rd6 (25... Rd8 {It was a good time to go back with}) 26. Ne2 (26. Ba6 $1
{Immediate was more precisely but after Black's last move I was sure the Rook
will not leave the d-file.}) 26... Be3 27. Ba6 Rxd1 28. Kxd1 $18 {
Position is obviously winning for White. Black can only stay and waiting.}
28... Kd7 29. Kc2 Kc7 30. b4 Bg5 31. Bc4 Kd6 32. Kd3 Kd7 33. Nc3 Bh6 34. Nb1 {
My first intention was to break with b4-b5 after Bb3 and Kc4 which was
probably also enough but I decided to try another idea first. Besides, I had a
lack of time and did not want to change the pawn structure before the time
control.} 34... Bf8 35. Kc3 Kd6 36. Nd2 Bh6 37. Kd3 Kc7 38. Bb3 Bf8 39. Kc4 Bg7
40. Kd3 Kd7 41. Nc4 Bf8 42. Kc3 Bh6 43. a5 $1 {The most clear way to win.}
43... Kc7 44. Ba4 Bf4 (44... a6 {After White win easily:} 45. Nb2 {then Nd3, Bb
3-c4 (K should stay on b7), Nc5 (forcing Bc5 bc), K goes to e3, then f3-f4,
e4-e5 and so on.}) 45. a6 $1 Bg3 46. Kd2 (46. Na5 $2 {Not because of} 46... c5
{although it should be winning as well.}) 46... Bf2 47. Na5 c5 (47... Kb6 {
This is forced as in case of} 48. Bxc6 $1 Kxa6 49. Nc4 {
black King could suddenly find himself in the mating net - b4-b5 is inevitable.
}) 48. b5 Bg1 49. Kd3 {Absolutely unnecessary move which allows Black to open
the diagonal with c5-c4. White can win without a King but I did not realized
it yet.} 49... Kb6 50. Nc4+ Kc7 51. Bb3 (51. b6+ $1 axb6 52. Bb5 $1 {could fini
sh the game with nice picture as taken from draughts: all white forces are on
the light squares and there is no defence against Queen promotion:} 52... Bd4
53. Nd6 Kb8 54. Bc4 $1 {
with next Nb5. To my excuse I can say it was deep night already.}) 51... Kb8
52. b6 {At this moment I realized my omission but decided to stop thinking
about the way to return and just win the game - there was not much time left.}
52... axb6 53. Nd6 Ka7 54. Bc4 b5 55. Bxb5 c4+ 56. Kxc4 {It was the only way
to stop a-pawn but now White can collect all kingside pawns starting, for
example, from Ne8 then Nf6-d7, Kd5, Ne5 etc. so Black resigned.} 1-0
Another example of anticomputer chess:
[Event "SuperGM"]
[Site "Dortmund GER"]
[Date "2000.07.12"]
[Round "5"]
[White "Kramnik, V."]
[Black "DEEP JUNIOR 6"]
[Result "1-0"]
[ECO "D00"]
[WhiteElo "2770"]
[PlyCount "65"]
[EventDate "2000.07.07"]
[Source "Mark Crowther"]
[SourceDate "2000.07.17"]
1. d4 d5 2. e3 Nf6 3. Bd3 e6 4. f4 Be7 5. Nf3 c5 6. c3 O-O 7. Nbd2 Ng4 8. Qe2
c4 9. Bc2 f5 10. Rg1 Nc6 11. h3 Nf6 12. g4 Ne4 13. Qg2 g6 14. Qh2 Kh8 15. h4
Nxd2 16. Bxd2 fxg4 17. Ng5 Qe8 18. h5 gxh5 19. Rxg4 Rf6 20. Rh4 Rh6 21. O-O-O
a5 22. Rh1 b5 23. Bd1 Ra7 24. Bxh5 Qf8 25. e4 Bd8 26. f5 b4 27. Bg6 Rxh4 28.
Qxh4 bxc3 29. bxc3 Bf6 30. Qxh7+ Rxh7 31. Rxh7+ Kg8 32. Bf7+ Qxf7 33. Rxf7 1-0
I am just trying this Dortmund game Kramnik-Junior. After already 42 minutes on
a PIII-600E, Rebel still picks the terrible 8... c4, on its way to a strangled
and lost position. All programs are quite lousy against anti-computer
strategies, as a few test positions from this kind of games proved many times.
I don't think Uri missed the point. We have no proof of any sort of superiority
of a given program against human players.
Aegon: overall, Rebel scored best; in the last 2 years of Aegon, Kallisto and
Nimzo achieved the highest TPRs. Which proves nothing. Few games, erratic TPRs.
Israel: Rebel got the highest TPR, but without the last game lost by Shredder
due to operator's incompetence, Shredder has by far the highest TPR. Which
proves nothing again: few games.
The very highest TPR ever achieved by a micro corresponds to Junior in Dortmund:
2704 ELO, first GM norm ever. Few games, other programs didn't participate...
Etc., etc., etc.
As for subjective appreciations, everybody has one. Anand prefers Hiarcs,
Kasparov prefers Junior, others swear by Fritz, or Rebel, Genius, you name it.
Enrique
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.