Not logged inRybka Chess Community Forum
Up Topic The Rybka Lounge / Computer Chess / Lyudmil Tsvetkov vs. Kappatoo: Anti-computer chess
1 2 Previous Next  
- - By Kappatoo (*****) [de] Date 2017-12-03 21:55 Upvotes 2
Lyudmil and I had a disagreement over whether it's possible to reliably steer an engine (armed with a simple opening book) into a closed position that it doesn't play well. (Details here: http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?tid=32234)
We thus started a game - the idea was to play 15 or so moves in which he tries to reach a position that is either much better for him or so closed that engines will play it poorly. For reasons of space, I'm moving the game here.

1. c4 Nf6 2. Nc3 d5 3. cxd5 Nxd5 4. g3 e5 5. Bg2 Nb6 6. d3 Be7 7. Nf3 Nc6
Parent - - By Lyudmil Tsvetkov (**) [bg] Date 2017-12-04 00:15
Excellent, except that you added 'armed with a simple opening book', when we never actually talked about books, and 'much better' position, when I am perfectly fine with 'sufficiently better'.

8. 0-0
Parent - By Kappatoo (*****) [de] Date 2017-12-04 00:32
Sorry, I didn't mean to distort our previous discussion. However, I had said the following two things in the other thread before the game started:

> In most cases, it's sufficient to exchange one pair of pawns to avoid getting into a closed position. How are you going to avoid that? Isn't a pretty simple opening book going
> to take care of this?


> How about a test? Let's play 15 or so moves here on the forum. Your goal is to get a position that is either clearly better for you or so closed that the engine will play it poorly.

Parent - - By Kappatoo (*****) [de] Date 2017-12-04 00:35
1. c4 Nf6 2. Nc3 d5 3. cxd5 Nxd5 4. g3 e5 5. Bg2 Nb6 6. d3 Be7 7. Nf3 Nc6 8. 0-0 0-0
Parent - - By Lyudmil Tsvetkov (**) [bg] Date 2017-12-04 03:28
I guess it is obvious clearly better is much less emphatic than much better and more in line with sufficiently better.

9.Be3
Parent - - By Kappatoo (*****) [de] Date 2017-12-04 09:21
Sufficient for what?
I find it hard to see your point here. This was about whether you manage to keep the position closed, which worked for exactly 3 moves. Now we're in an asymmetric, open position that would be a nightmare to play against an engine and in which you just keep following mainline theory. So are you going to make a few more theory moves and then declare victory in an equal position on the basis of an evaluation that only you can see?
 
1. c4 Nf6 2. Nc3 d5 3. cxd5 Nxd5 4. g3 e5 5. Bg2 Nb6 6. d3 Be7 7. Nf3 Nc6 8. O-O O-O 9. Be3 Be6
Parent - - By Ozymandias (****) [es] Date 2017-12-04 11:17

>are you going to make a few more theory moves and then declare victory in an equal position on the basis of an evaluation that only you can see?


That would be a repetition of his line of reasoning here and here:

>I am following the game simultaneously with watching SF scores, and mostly they seem 100cps convertible positions.


>that is prone to errors too, of course, everyone and everything prunes some moves and evaluates at times below par, so mistakes are unavoidable.


>but what I am seeing is mostly white wins.


........

>regarding the Marschall, I can not share anything, as there are endless lines to analyse, but one thing is certain: there is not a single sound black gambit, and the Marschall is a gambit.


>after Be3, black has no satisfactory reply, one way or another.


>interesting, how those players manage to draw it?


Let's hope it's not the case.
Parent - - By Lyudmil Tsvetkov (**) [bg] Date 2017-12-04 11:36

>That would be a repetition of his line of reasoning here and here:
>Let's hope it's not the case.


Sorry, I am tired enough to go checking what you have posted.
Generally, you are well-meaning, so I hope this is the case again.

I hope you don't prefer simple ways of reasoning like: white is better, black is better, unclear, with compensation.
That has no meaning at all. It is either draw or win for one side, so my attempts to bring more clarity should only be lauded.
Parent - - By Ozymandias (****) [es] Date 2017-12-04 13:57

>I hope you don't prefer simple ways of reasoning like: white is better, black is better, unclear, with compensation.
>That has no meaning at all. It is either draw or win for one side, so my attempts to bring more clarity should only be lauded.


The conventions for chess annotation are, more than anything, a compromise. We all would love for a way to tell if a position is won or drawn, but outside 7-men positions, all we can do is postulate. It's in that grey area, where the method used to ascertain a position, varies. Some prefer human assessment, others veer towards statistical analysis and finally there's those who prefer engine evaluations. For the longest time, human assessment was the only way to go, hence the compromise whenever a mate sequence, or a few drawn patterns, didn't appear on the board.

My preference would be stats>evals>assessment. But the problem is that, for the two first methods, the conditions are essential, in order to assign the correct level of certainty. I'm still waiting to know what conditions were met, before you made that particular statement. Lacking that information, I can't adhere to the attempt. Likewise, once you guys are finished with the issue at hand, I will take a look and pronounce myself accordingly.
Parent - - By Lyudmil Tsvetkov (**) [bg] Date 2017-12-04 18:56 Upvotes 1

>The conventions for chess annotation are, more than anything, a compromise. We all would love for a way to tell if a position is won or drawn, but outside 7-men positions, all we can do is postulate. It's in that grey area, >where the method used to ascertain a position, varies. Some prefer human assessment, others veer towards statistical analysis and finally there's those who prefer engine evaluations. For the longest time, human >assessment was the only way to go, hence the compromise whenever a mate sequence, or a few drawn patterns, didn't appear on the board.


>My preference would be stats>evals>assessment. But the problem is that, for the two first methods, the conditions are essential, in order to assign the correct level of certainty. I'm still waiting to know what conditions >were met, before you made that particular statement. Lacking that information, I can't adhere to the attempt. Likewise, once you guys are finished with the issue at hand, I will take a look and pronounce myself accordingly.


I don't know if people found this post of mine in The Secret of Chess thread, it is too long and unorderly, so I am reposting it here:

I am so happy, the first review for 'The Secret of Chess' has appeared,
and by none other than Australian GM and book reviewer David Smerdon:
http://davidsmerdon.com/?p=1970

Thank you, David!

The review is also available on the chess.com blog of the grandmaster:
https://www.chess.com/blog/smurfo/the-secret-of-chess
Parent - By Ozymandias (****) [es] Date 2017-12-04 20:32
Quoting my post, doesn't make yours more of an actual reply.
Parent - - By Lyudmil Tsvetkov (**) [bg] Date 2017-12-04 11:27 Edited 2017-12-04 11:31

>Sufficient for what?
>I find it hard to see your point here. This was about whether you manage to keep the position closed, which worked for exactly 3 moves. Now we're in an asymmetric, open position that would be a nightmare to play >against an engine and in which you just keep following mainline theory. So are you going to make a few more theory moves and then declare victory in an equal position on the basis of an evaluation that only you can see?


10. Rc1

White has clear advantage, way abobe the usual first move white advantage, so you should already admit you were wrong.
Black has huge problems along the semi-open c file, and I am not certain how you are going to solve those.
Evaluation is already somewhere around 35-40cps, so very close to white win, but maybe still not having crossed the threshold.

I don't know why you make me play a game you are unhappy with.
I can stop at any time, just tell me. But again, 2...d5 was subpar, and then you made one more slight inaccuracy.

Please, read what you are posting, because it does not make any sense, the bet was either-or, and not just to get a closed position.
When one says something, one should remember his words.

You can always open the position, for example 1.c4 b5, is that what you want?
Parent - - By Kappatoo (*****) [de] Date 2017-12-04 12:48

> Evaluation is already somewhere around 35-40cps, so very close to white win


According to what exactly? Your secret evaluation formula?
I give you 10:1 odds for drawing this position against a strong engine.

1. c4 Nf6 2. Nc3 d5 3. cxd5 Nxd5 4. g3 e5 5. Bg2 Nb6 6. d3 Be7 7. Nf3 Nc6 8. O-O O-O 9. Be3 Be6 10. Rc1 f5
Parent - - By Lyudmil Tsvetkov (**) [bg] Date 2017-12-04 19:26

>According to what exactly? Your secret evaluation formula?
>I give you 10:1 odds for drawing this position against a strong engine.


Yes, according to my secret evaluation formula.

I will beat you that game.

11.a3
Parent - - By Kappatoo (*****) [de] Date 2017-12-04 19:49
You will beat me? Does that mean you want to play a game against me? I thought the idea was that you play against an engine. But sure, anytime. But I want to play against you, not against your engine.
Seriously - do you have any interest to demonstrate that your formula works, or to back up your talk with anything else?

1. c4 Nf6 2. Nc3 d5 3. cxd5 Nxd5 4. g3 e5 5. Bg2 Nb6 6. d3 Be7 7. Nf3 Nc6 8. O-O O-O 9. Be3 Be6 10. Rc1 f5 11. a3 Kh8
Parent - - By Lyudmil Tsvetkov (**) [bg] Date 2017-12-04 22:11

>You will beat me? Does that mean you want to play a game against me? I thought the idea was that you play against an engine. But sure, anytime. But I want to play against you, not against your engine.
>Seriously - do you have any interest to demonstrate that your formula works, or to back up your talk with anything else?


You already started playing 'human + engine' from the very first moves.
But as of late, what you started making even more often, is mistakes.

12.b4
Parent - - By Kappatoo (*****) [de] Date 2017-12-04 22:29
Any answers to my questions?
As I'm sure you are aware, we are still in book.

Seriously, I have the impression you are just trolling. You keep claiming that you see things no-one else can see, and back it up with ridiculously simplistic views about chess.
It's plain to see for everyone that we are in a standard, equal position that no human being could hold against an engine. So prove me wrong on this - I am open to your suggestions how you would like to test your claims. If you are not willing to back up your claims with anything, you're definitely just trolling.
Parent - - By Lyudmil Tsvetkov (**) [bg] Date 2017-12-04 23:21

>Seriously, I have the impression you are just trolling. You keep claiming that you see things no-one else can see, and back it up with ridiculously simplistic views about chess.
>It's plain to see for everyone that we are in a standard, equal position that no human being could hold against an engine. So prove me wrong on this - I am open to your suggestions how you would like to test your claims. >If you are not willing to back up your claims with anything, you're definitely just trolling.


My claim is you are already almost lost.
It's your move.
If you want, you might as well just resign.
Parent - - By Kappatoo (*****) [de] Date 2017-12-04 23:29
Please. The idea was to play around 15 moves, remember that? So what's your point, do you want to play a whole game on the forum now? Or do you expect to checkmate me before move 15?

What I'm saying is: If you think you're almost winning, back it up. I'm waiting for you to suggest conditions. Again: We are supposed to play until move 15 or so. Would you like to change that?
Parent - - By Lyudmil Tsvetkov (**) [bg] Date 2017-12-05 09:49

>Please. The idea was to play around 15 moves, remember that? So what's your point, do you want to play a whole game on the forum now? Or do you expect to checkmate me before move 15?


>What I'm saying is: If you think you're almost winning, back it up. I'm waiting for you to suggest conditions. Again: We are supposed to play until move 15 or so. Would you like to change that?


Only way for me to prove it is by continuing the game.
Whether you utilised a book or engine help, black made a couple of inaccuracies(at least d5, f5 and Kh8 are such), so white already has clear advantage.
Was not this the point: to show clear advantage? Now, you have it. I am certain top engines with longer thinking time on fast hardware will prefer white by a margin.

If we play some more moves, yes, until 15, the advantage will increase further, because small advantages only tend to increase.
Then you will get a more convincing score from the engines.

As said, openings books are not perfect, neither top engines are.
You are happy we did not get a closed position, because I am stronger there.

I have played my move.
Parent - - By Kappatoo (*****) [de] Date 2017-12-05 12:53
I regret that I assumed you were serious about this.
The point was that you claimed to be able to reliably steer the engine into positions in which humans have a chance against it, and I wanted to test this. You utterly failed in this respect, opening the position at move 3. It doesn't help your cause that you now claim that a highly theoretical, equal position is almost winning for you.
Even if the position was amazingly good for you, my point has always been that you lose this position against the engine without a chance. So, again: What are your conditions?
If you want, we can play to move 15 and then have an engine give an assessment of the position, even thought that would mean to completely change the topic. But again, what are your conditions for that? Latest Stockfish version? How deep? And what is the score you would require to win? Something like +1.00?

1. c4 Nf6 2. Nc3 d5 3. cxd5 Nxd5 4. g3 e5 5. Bg2 Nb6 6. d3 Be7 7. Nf3 Nc6 8. O-O O-O 9. Be3 Be6 10. Rc1 f5 11. a3 Kh8 12. b4 a6
Parent - - By rocket (***) [se] Date 2017-12-05 12:59
Here are some ways to force a closed position even against opening book

fujichess
2009.04.24
Master Om, Rybka 3 1-cpu
Engine Room
1-0
Rated game, 1m + 0s

Om Small Book 1.0.ctg, 256 MB 1. Nf3 2 1... Nf6 B/0 0 2. Nc3 1 2... d5 B/0 0 3. d3 1 3... c5 -0.18/9 1 4. e4 (Bf4) 1 4... d4 -0.28/9 1 5. Ne2 1 5... Nc6 -0.22/10 0 6. h3 (c3) 1 6... h6 -0.31/8 1 7. Bd2 (c3) 1 7... Qb6 -0.34/9 2 8. Qc1 (Rb1) 1 8... e6 -0.44/9 1 9. g4 (g3) 1 9... Bd6 -0.55/9 2 10. Bg2 1 10... O-O -0.65/9 0 11. Bxh6 (0-0) 1 11... gxh6 -0.81/7 1 12. Qxh6 1 12... Qa5 -0.77/9 0 13. c3 1 13... Be7 -0.86/10 0 14. Ng5 (e5) 1 14... dxc3 -1.40/8 1 15. O-O (Nxc3) 1 15... cxb2 -0.82/7 8 16. Ng3 (Rab1) 0 16... Qc7 #11/7 2 17. Nh5 (Rab1) 1 17... Qh2 #6/3 0 18. Kxh2 1 18... Bd6 #5/3 0 19. Kh1 (Kg1) 1 19... bxa1=Q #2/3 0 20. Qg7# 2

Bravochess
2009.04.02
Bh7, Rybka 1.2 OM
Engine Room
1-0
Rated game, 1m + 0s

Om Book 1.0.ctg, 512 MB 1. e4 1 1... e6 B/0 0 2. d3 2 2... d5 B/0 0 3. Nc3 1 3... c5 B/0 0 4. Bd2 1 4... Nc6 -0.03/10 2 5. Nf3 1 5... Nf6 -0.03/11 1 6. h3 (Be2) 1 6... d4 -0.12/11 2 7. Ne2 (Nb5) 1 7... Bd6 -0.25/10 2 8. Qc1 (c3) 1 8... O-O -0. 37/8 0 9. Bf4 (g3) 1 9... e5 -0.35/10 3 10. Bd2 1 10... Nh5 -0.29/10 0 11. g4 1 11... Nf6 -0.29/11 0 12. Ng3 (Bg2) 1 12... h6 -0.27/10 2 13. Bxh6 (Bg2) 1 13... gxh6 -0. 79/7 1 14. Qxh6 1 14... Be7 -0.50/8 0 15. Ng5 (Nf5) 1 15... Qa5 166.80/5 0 16. Kd1 2 16... Bd7 166.80/4 0 17. Nh5 2 17... Bxg4 #4/3 0 18. hxg4 2 18... Qd2 #3/3 0 19. Kxd2 1 19... Nxe4 #2/3 0 20. dxe4 1 20... Bxg5 #2/3 0 21. Qxg5 1 21... Kh7 #1/ 3 0 22. Nf6# (Ng3#) 4

A8: pgn4web v2.65 debug info  B8: show this position FEN string  C8: show this game PGN source data  D8: show full PGN source data  E8: search help  F8: shortcut keys help  G8: shortcut squares help  H8: pgn4web help
A7: pgn4web website  B7: undo last chessboard position update  C7: redo last undo  D7: toggle highlight last move  E7: flip board  F7: toggle show comments in game text  G7: toggle autoplay next game  H7: toggle enabling shortcut keys
A6: pause live broadcast automatic refresh  B6: restart live broadcast automatic refresh  C6: search previous finished game  D6: search previous unfinished game  E6: search next unfinished game  F6: search next finished game  G6  H6: force games refresh during live broadcast
A5: repeat last search backward  B5: search prompt  C5: repeat last search  D5: search previous win result  E5: search next win result  F5  G5: toggle auto-scroll moves text to current move  H5: scroll moves text to current move
A4: search previous event  B4: search previous round of same event  C4: search previous game of same black player  D4: search previous game of same white player  E4: search next game of same white player  F4: search next game of same black player  G4: search next round of same event  H4: search next event
A3: load first game  B3: jump to previous games decile  C3: load previous game  D3: load random game  E3: load random game at random position  F3: load next game  G3: jump to next games decile  H3: load last game
A2: stop autoplay  B2: toggle autoplay  C2: autoplay 1 second  D2: autoplay 2 seconds  E2: autoplay 5 seconds  F2: autoplay custom delay  G2: replay up to 6 previous half-moves, then autoplay forward  H2: replay the previous half-move, then autoplay forward
A1: go to game start  B1: go to previous comment or variation  C1: move 6 half-moves backward  D1: move backward  E1: move forward  F1: move 6 half-moves forward  G1: go to next comment or variation  H1: go to game end
<<    <    +    >    >>
Bravochess
2009.04.02
Bh7, Rybka 3 Dynamic
Engine Room
1-0
Rated game, 1m + 0s

Om Book 1.0.ctg, 512 MB 1. e4 1 1... e6 B/0 0 2. d3 1 2... c5 B/0 0 3. Nc3 1 3... d5 B/0 0 4. Bd2 1 4... Nc6 -0.28/9 2 5. Nf3 1 5... Nf6 -0.43/9 0 6. h3 (e5) 1 6... d4 -0.44/9 1 7. Ne2 (Nb5) 2 7... Bd6 -0.33/9 4 8. Qc1 (c3) 1 8... O-O -0.79/8 1 9. Bf4 (c3) 1 9... Qc7 -0.84/8 2 10. Bd2 (Bxd6) 2 10... h6 -1.03/8 1 11. Bf4 (g4) 3 11... e5 -1.12/8 2 12. Bd2 (Bxh6) 2 12... c4 -1.10/8 1 13. Ng3 1 13... Be6 -1.01/9 0 14. Be2 1 14... Nb4 -0.88/8 0 15. O-O 2 15... cxd3 -0.89/11 0 16. Bxh6 (Bxd3) 6 16... dxe2 -5.02/7 1 17. Re1 (Bxg7) 2 17... Nxc2 -5.56/7 1 18. Qd2 (Qg5) 4 18... gxh6 -6.29/6 1 19. Qxh6 (Rec1) 1 19... Qe7 -6.30/6 1 20. Ng5 (Nxe2) 1 20... Nxe1 -0.07/ 5 7 21. Nh5 (Rxe1) 5 21... Nf3 #4/3 0 22. gxf3 1 22... e1=R #3/3 0 23. Rxe1 2 23... Rae8 #2/3 0 24. Qg7# 1 .
Parent - - By Kappatoo (*****) [de] Date 2017-12-05 13:11
That's why I said a simple opening book can take care of this.
My challenge stands: Let me emulate such an opening book and anyone can try to reach a position against me after 15 moves that isn't hopeless against an engine. One can do much better than my opponent in this thread (for instance, by playing the Reti or something like this), but I'm still not sure it's doable. But I would like to test it.

(By the way, a simple improvement in all of the games you give below would have been to exchange on e4, instead of playing d4 - just as an illustration.)
Parent - - By rocket (***) [se] Date 2017-12-05 13:32
Okey. There are a few openings against 1. e4 in which I can force matters. If you play the Sicilian defence, I respond 2 c4:cool:

If you play 1.e5, I play 2 c4:cool:
Parent - - By Kappatoo (*****) [de] Date 2017-12-05 17:12
Wanna try it? 15 moves in a game from 1. e4 e5 2. c4, and you have to keep the position closed and manageable against an engine?
Parent - - By rocket (***) [se] Date 2017-12-05 20:19
I don't need to. I can show you my draw against Stockfish 5 as BLACK. It is much  weaker than version 8 but I am sure version 8 is equally clueless against my home-cooked opening:

White: Stockfish 5

Black: Rocket

Time control 5+3 ponder on

Result: 1/2-1/2

1.e4 Nc6 2. Nf3 d6 3. d4 a6 4. be2 Nf6 5. Nc3 (last book move) e6 6. 0-0 Be7 7. Bf4 Nh5 8. Be3 Nf6 9. a3 0-0 10. Re1 e5 11. d5 Nb8 12 Nd2 c6 13 Nc4 c5 14 a4 b6 15 Qd3 Nbd7 16 Bg5 Bb7 17. b3 h6 18. Bd2 Re8 19 Bc1 Nf8 20. Bb2 Ng6 21. g3 Qc7 22. Ne3 Bf8 23 Qd1 Rac8 24 Bd3 Rcd8 25 Qe2 Ra8 26 Rab1 Qd7 27 f3 Qc7 28 Qf2 Qd7 29 Ncd1 Rad8 30 Qf1 Ra8 31 Nf2 Qc7 32 Rbd1 Qc7 33 Rd2 Qc7 34 c3 Qd7 35 Red1 Qc7 36 Rb1 Qd7 37 Qe2 Qc7 38 Rf1 Qd7 39 Rdd1 Qc7 40. Rfe1 Qd7 41 c4 a5 42 Qd2 Qc7 43 Bc3 Qd7 44 Bf1 Qc7 45 Bg2 Qc7 46 Re2 Qc7 47 Rde1 Qd7 48 Bb2 Qc7 49 Nfd1 Qd7 50 Nc3 Qc8 51 Nb5 Qd8 52 Nf5 Ba6 53 Qc2 Qb8 54 Bc3 Qd8 55 Qb1 Qb8 56 Qd3 Qd8 57 Ra1 Qb8 58 Ne3 Qd8 59 Qd2 Bc8 60 Rd1 Bd7 61 Rde1 Qc8 62 Nf5 Qb8 63 Ra1 Bc8 64 Bh3 Kh8 65 Qc1 Kg8 66 Bg2 Ba6 67 Ra2 Bc8 68 Rac2 Ba6 69 Ne3 Bc8 70 Qf1 Ba6 71 Ra2 Bb7 72 Rf2 Bc8 73 Rae2 Bd7 74 Bh3 Bc8 75 Nf5 Ba6 76 Qa1 Bb7 77 Rf1
Ba6 78 Ne3 Bbb7 79 Qb2 Ba6 80 Ref2 Bb7 81 Qd2 Ba6 82 Ra1 Bb7 83 Re1 Ba6 84 Bg2 Bb7 85 Nf5 Ba6 86 Qc2 Bb7 87 Rff1 Ba6 88 Re2 Bb7 89 Qd1 Ba6 90 Qd2 Bb7 91 Bbb2 Ba6

Draw by 50 move rule
Parent - By Lyudmil Tsvetkov (**) [bg] Date 2017-12-05 23:09

>I don't need to. I can show you my draw against Stockfish 5 as BLACK. It is much  weaker than version 8 but I am sure version 8 is equally clueless against my home-cooked opening:


Drawing is easy, I have more than 70% draws against Stockfish, show us some wins.
Parent - By Lyudmil Tsvetkov (**) [bg] Date 2017-12-05 19:24

>Okey. There are a few openings against 1. e4 in which I can force matters. If you play the Sicilian defence, I respond 2 c4:cool:


>If you play 1.e5, I play 2 c4


That is indeed like that. You might be able to get a closed position, but this way of playing is far from perfect and would give black full equality.
Parent - - By rocket (***) [se] Date 2017-12-05 13:40
I guarantee you that I can draw any top engine as white with or without anti-computer chess. And this will always be the case.
Parent - - By mist (**) [gb] Date 2017-12-05 15:26
I wonder what happened to the chess worlds jester (Fool)  a.r.b. ?!!!!! :twisted::roll:
Parent - - By rocket (***) [se] Date 2017-12-05 16:43
He showed how easy it is to beat todays engines with anti-computer chess with the black pieces. There still hasn't been much progress there.
Parent - By Lyudmil Tsvetkov (**) [bg] Date 2017-12-05 19:27

>He showed how easy it is to beat todays engines with anti-computer chess with the black pieces. There still hasn't been much progress there.


I am sorry but his games are so so full of inaccuracies that I wonder how is it possible to win like that.
He seems also to be playing LTC instead of faster games.
Parent - - By Kappatoo (*****) [de] Date 2017-12-05 17:11
That's impressive.
Parent - - By mist (**) [gb] Date 2017-12-05 17:35

>He showed how easy it is to beat todays engines with anti-computer chess


A long time ago...on his hardware, no one else could make the software engines make the same moves his did!? Anyway, enough about that chess world troll.

If you are saying you are able to beat chess engines I assume you MUST have a very high chess ranking?  Because even the best G.M. would have trouble against any of the top 3 engines.

Unless the engine is only 'thinking' for a couple of sec's?
Parent - - By Kappatoo (*****) [de] Date 2017-12-05 17:38
You mean me? I'm a better player than most on this forum, but way too weak to beat an engine, even if it only thinks for a few seconds.

Edit: I guess you meant to reply to Rocket.
Parent - By Scott (****) Date 2017-12-05 23:21

> You mean me? I'm a better player than most on this forum, but way too weak to beat an engine, even if it only thinks for a few seconds.


We agree with you of course, also, M. Carlson could say the same thing with the exception of changing "most" to "all".
Parent - By mist (**) [gb] Date 2017-12-06 09:06

>Edit: I guess you meant to reply to Rocket.


Yes!:smile:
Parent - By Lyudmil Tsvetkov (**) [bg] Date 2017-12-05 19:22

>One can do much better than my opponent in this thread (for instance, by playing the Reti or something like this), but I'm still not sure it's doable. But I would like to test it.


Was not this like a Reti?
Parent - - By Lyudmil Tsvetkov (**) [bg] Date 2017-12-05 19:20

>Here are some ways to force a closed position even against opening book


This was long long ago.
He wants current-day examples.
Parent - By rocket (***) [se] Date 2017-12-05 20:07
Stockfish 6 is "only" 89 elo weaker tham version 8. I bet you those old school kings indian tactics still work
Parent - - By Lyudmil Tsvetkov (**) [bg] Date 2017-12-05 19:17

>I regret that I assumed you were serious about this.
>The point was that you claimed to be able to reliably steer the engine into positions in which humans have a chance against it, and I wanted to test this. You utterly failed in this respect, opening the position at move 3. It >doesn't help your cause that you now claim that a highly theoretical, equal position is almost winning for you.
>Even if the position was amazingly good for you, my point has always been that you lose this position against the engine without a chance. So, again: What are your conditions?
>If you want, we can play to move 15 and then have an engine give an assessment of the position, even thought that would mean to completely change the topic. But again, what are your conditions for that? Latest >Stockfish version? How deep? And what is the score you would require to win? Something like +1.00?


Ok, I was somewhat busy.
I don't know what you are talking about, I am not even reading it, the bet was obvious, but you will not acknowledge it.
+40cps is a winning score in the middlegame, so you migth want to draw your own conclusions.

13.Na4
Parent - - By Kappatoo (*****) [de] Date 2017-12-05 19:38

> the bet was obvious, but you will not acknowledge it.


Please explain. What was the bet, what was it obvious from?
Let's fix clear conditions, now.
Parent - - By Lyudmil Tsvetkov (**) [bg] Date 2017-12-05 23:16

>Please explain. What was the bet, what was it obvious from?
>Let's fix clear conditions, now.


I am not going to read your posts for you.
Don't be so formal, just play some moves to see what score we will reach.

First move advantage, according to current understanding, is some 15cps in chess, so, if the score shown by the top engines after a longer search is higher than that, that will mean I have been able to outplay both the book and the black engine.
If not, then 2...d5 has been a good move, and it is possible to avoid closing the position, while at the same time getting an acceptable play.
Parent - - By Kappatoo (*****) [de] Date 2017-12-06 11:51
I prefer clear rules. Otherwise you'll just run and declare victory without any basis. So, again, latest Stockfish, depth around 40, eval 1.00? Is 1.00 too much for you? (Weren't you saying you're almost winning?)
Parent - - By Lyudmil Tsvetkov (**) [bg] Date 2017-12-06 12:55

>I prefer clear rules. Otherwise you'll just run and declare victory without any basis. So, again, latest Stockfish, depth around 40, eval 1.00? Is 1.00 too much for you? (Weren't you saying you're almost winning?)


That is right, 45cps is most probably the correct winning margin in the middlegame, so over it would be a win.
I am currently assessing the position as +35cps for white, so very close to winning, but still not. I am waiting for you to make one more slight inaccuracy. :)
Initially, we started with 15cps first move white advantage, so I actually should have outplayed the engine/book even in this very open position.
Parent - - By Kappatoo (*****) [de] Date 2017-12-06 13:35
Is there ever any basis for anything you're saying? Centipawns according to what, which engine? And why 0.45? And where does your 0.35 assessment of the current position come from? Finally, how are we going to measure the evaluation - are you going to do that, based on your imagined formula?
Parent - - By Ozymandias (****) [es] Date 2017-12-06 21:28

>why 0.45?


Indeed, I've seen persistent, higher evals in the middle game, that went nowhere. And I'm talking cross-referenced ones above depth 30.

As for calling a 0.35 advantage as "close to winning", it doesn't offer more clarity in the assessment of a position, than the old nomenclature.
Parent - - By Lyudmil Tsvetkov (**) [bg] Date 2017-12-07 10:05

>Indeed, I've seen persistent, higher evals in the middle game, that went nowhere. And I'm talking cross-referenced ones above depth 30.


>As for calling a 0.35 advantage as "close to winning", it doesn't offer more clarity in the assessment of a position, than the old nomenclature.


I was referring to the general rule, and not some exceptions, like for example what you have seen.
I am the person in the world, who has closely followed top engine evals the most, I might have done so in some 30 000 or more games,
following every single move on the screen with the respective engine eval, and then seeing the outcome.
I guess the second-best person in that field would have somewhere 1/5 of my experience, because it is very specific.
Parent - By Ozymandias (****) [es] Date 2017-12-07 22:52

>I was referring to the general rule, and not some exceptions, like for example what you have seen.


Establishing general rules that may clarify the situation, for positions in the "grey area", will always have to deal with exceptions, the problem is how abundant they are. In this particular case, the longer the TC and the stronger the engine, the more you'll see that clear advantages, go nowhere. This is mainly, both because chess is a draw, and also because engines are becoming stronger, at the cost of risking less (most risks aren't sound).
I don't know what the score should be, for a given evaluation to be considered a win, at a given phase of the game. Even saying how long each phase lasts, is problematic. You have experience watching, I do playing, and what I know is that I don't stop looking for an alternative move, when I have one that gives me 0.45, unless it's forced.
Parent - - By Lyudmil Tsvetkov (**) [bg] Date 2017-12-07 10:00

>Is there ever any basis for anything you're saying? Centipawns according to what, which engine? And why 0.45? And where does your 0.35 assessment of the current position come from? Finally, how are we going to >measure the evaluation - are you going to do that, based on your imagined formula?


My observations show that, below 45 cps in the mg, engines are usually able to save the draw; above that margin, the score usually only increases,
and ends in a win.
Again, that is the usual behaviour, maybe in some 80% of cases.
Up Topic The Rybka Lounge / Computer Chess / Lyudmil Tsvetkov vs. Kappatoo: Anti-computer chess
1 2 Previous Next  

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill