< Earlier Kibitzing · PAGE 11 OF 11 ·
|Jun-07-15|| ||jith1207: I think some people might be brilliant or have their own resources (read as computers with necessary analysis programs) so I understand they might not like seeing some pages of kibitzes "hijacked" for pure analysis. I don't have patience or interest to go through them all the time, but a year back when I stumbled upon this very game, I went through <Aylerkupp> analysis for few days as much as I could and that created more interest in chess in my life and have made me a regular visitor and contributor to this site. |
I'm not a chess player but lover and follower of the game and I don't have time and resources to do an analysis like that after my job. I think what people forget is this site is visited by people of all kinds and every kind of kibitz (except troll) helps in some way to some one and even if it doesn't it helps to archive for future. That's what kibitzing is meant for. I understand it might not be pretty to have extensive analysis of every game in every other page but this game deserves what has been done here.
I very much appreciate, <Aylerkupp>. As much as people have rights to complain and not want analysis here, people also have rights to post what they think beneficial or necessary or just fun., except trolling of course.
In the end, it takes only few seconds to skip the kibitzes we don't want to read but it takes hours to carefully post what you love about a particular game or player or statistic or page here.
|Jun-08-15|| ||An Englishman: Good Evening: I wish to concur with <jith1207> and express my gratitude to <Aylerkupp> for taking the time to write all of those posts, even if a computer might have assisted a little on the analysis of the game. |
And let's give credit to Messrs. Hamppe & Meitner, who for 18 moves were as good as the greatest who ever played the game, giving us a puzzle that might remain not-quite-solved.
|Jun-08-15|| ||sofouuk: <AylerKupp>had completely forgotten i'd even seen the game before today until i looked at the comments and wondered what the heck you were talking about :) my off the cuff conclusion 'clearly advantage white' was an eyeballing 'i'd take white in correspondence' rather than 'white has a forced win', but i won't change it - there's no way black is winning, possibly there might be some kind of dynamic blockade which prevents white from making progress but pretty difficult to prove anything in such a messy position|
|Jun-08-15|| ||sofouuk: <AylerKupp>if you could post your 'best play by black' line I'd be interested in trying to crack it, btw - right now I'm not sure what line to try and improve for white|
|Jun-08-15|| ||kevin86: White must submit to a perpetual check or be mate-if he moves to b7 or a4.|
|Jun-08-15|| ||AylerKupp: <An Englishman> even if a computer might have assisted a little on the analysis of the game.>|
I would say that the computer assisted more than "a little", it (or, more accurately, the engines) performed the entire analysis! As I indicated above, I'm not that good of a player so any significant effort that I put into analysis is probably mostly wasted, since it is certain to be wrong. As Clint Eastwood said in "Magnum Force", https://www.youtube.com/watch?v=_Vr....
My contribution (if you can call it that), is to give my opinions on the computer lines since, even though I know my limitations, I am not necessarily willing to accept them. :-) And, after reviewing the engines' lines, I looked for possible improvements for both sides and guided the engines by doing some forward sliding on what I thought were promising alternatives.
|Jun-08-15|| ||AylerKupp: <sofouuk> That's OK, a lot of people often wonder what I am talking about. I even wonder myself sometimes. And remember that the analysis was the result of <chessgames.com>'s asking whether anybody had any comment on your analysis using Rybka and, of course, I couldn't refuse an offer to "help". My conclusions were even more ambivalent than yours, I simply didn't know what to conclude since the 3 engines I used were not unanimous in their conclusions, so I simply passed it back to the team. Perhaps if <chessgames.com> had made this game the subject of their first Thematic Challenge we might have reached a definitive conclusion.|
And let's not forget that I did this analysis more than 3 years ago, and that is almost an eternity in computer time. Effort seems to have stopped in developing Spike 1.4 and it is no longer nearly as highly rated as it used to be (#7 in CCRL's Feb-2012 rating list and #22 in their recent May-2015 rating list, along with a corresponding decrease in its rating, 3127 to 3033) so its evaluations should not be considered as accurate as those of more recent engines. And many new releases of both Houdini and Stockfish, of increasing strength, have been made since I posted those analyses.
But, to get back to your original question, after your (and Rybka's) 11.Kb5 instead of Hamppe's 11.Kb4, the best lines for both sides, per the engines, are those that I posted (the Principal Variation), all starting with 11...Ne7. But today's versions of the engines (Houdini 4 instead of 1.5a, Stockfish 6 instead of 2.22) would likely give different (and probably more accurate) results, and I would use Komodo 9 instead of Spike 1.4. That's one (among many) problem with engines, new versions get released all the time, and they (along with their natural determinism) will give different results each time you run an analysis. So what might have been the best line for Black more than 3 years ago, might not be indicated as the best line for Black today. Just for the heck of it, I'll re-run the analysis starting at 11.Kb5 and see what these new versions of the engines say.
But, of course, White might have better places to deviate from the game than 11.Kb5, so the conclusion, if any, following what Black's best line would be after 11.Kb5 might be different after a different White move.
|Jun-09-15|| ||AylerKupp: <The Immortal Draw> (part 1 of 4)|
I did (well, the engines did) the initial analysis after <sofouuk>'s suggested 11.Kb5 instead of Hamppe's 11.Kb4 in Mar-2012. Since that time one of the engines I used, Spike 1.4, has not been improved and both its ranking and rating have dropped considerably, to the point where it is not considered a top engine. The other 2 engines I used, Houdini and Stockfish, have been improved since then, with multiple releases since Mar-2012. Since <sofouuk> asked me for my best lines for Black, I decided that it might be worthwhile to update the analyses with the latest released versions of Houdini and Stockfish, and using the latest Komodo instead of Spike. So here are the updated results of the analyses.
Position after 11.Kb5:
click for larger view
<Houdini 1.5a vs. Houdini 4>. Both versions of Houdini selected the same best move for Black, 11...Ne7, and arrived at similar evaluations, [+1.01] at d=27 for Houdini 1.5a vs. [+1.09] at d=29 for Houdini 4. Here is Houdini 4's best line (Principal Variation):
11...Ne7 12.c4 d4 13.Kxc5 a5 14.Qa4+ Kd8 15.Qxa5 Rxa5+ 16.Kb4 Nc6+ 17.Kb3 b5 18.d3 bxc4+ 19.dxc4 Ra8 20.Bd2 Re8 21.Ne2 e4 22.Be1 Be6 23.Nf4 Ne5 24.Bb4 Rb8 25.Nxe6+ Rxe6 26.c5 d3 27.g3 f5 28.Bg2 Kd7 29.Raf1 g6 30.Rd1 Ke7
Houdini 4 does not deviate from Houdini 1.5a until 17...b5 when Houdini 1.5a selected 17...Be6 as Black's best move. 17...b5 seems better as it saves a tempo and opens lines quicker, although Black's bishop still winds up on e6 anyway. This is probably its best square, although 19...Bf4 cuts off one of the White king's retreat squares and facilitates ...Ke7 and ...Rb8(+), bringing Black's last piece into action. So perhaps it should be considered. But most important, Houdini 4 still considers it necessary to give up White's queen by 15.Qxa5, otherwise 15...b6#.
With 19...Ra8 and 20...Re8 Black changes plans from a direct attack against White's king to consolidation and an attempt to exploit Black's center pawn majority and connected passed pawns. But this approach, although perhaps safer and sounder, seems too slow, although White does not seem to be able to take advantage of it.
This is the final position of Houdini 4's PV:
click for larger view
White has a 2B vs. N+P advantage but Black's advanced and connected passed pawns sure look dangerous, and White's passed a-pawn will not be ready to advance for quite a while. I would think that Black should be able to hold.
Restarting the analysis from the final position Houdini 4 evaluates the resulting position at only [+0.65], d=28 after 31.Rhe1 Nc6 32.Kc4 Ne5+ 33.Kc3 Nc6 34.Re3 h5 35.b3 Kf6 36.Rf1 Kg5 37.h4+ Kh6 38.Rd1 Re7 39.Bh3 Kg7 40.Bf1 Kf6 41.Bg2 Re6 42.Rf1 Kg7 43.Rf2 Re7 44.Rd2 Kf6 45.Re1 Ree8 46.Kc4 Ne5+ 47.Kd4 Nc6+ 48.Kc3 Re7 49.Bh3 Ne5 50.Rdd1 Nc6 51.Rd2 Ne5 so, as far as Houdini 4 is concerned, it looks like Black is close to equalizing.
|Jun-09-15|| ||AylerKupp: <The Immortal Draw> (part 2 of 4)|
<Stockfish 1.91 vs. Stockfish 6>: Both versions of Stockfish also selected the same best move for Black, 11...Ne7, and also arrived at similar evaluations, [+1.09] at d=27 for Stockfish 1.91 and [+1.13] at d=40 for Stockfish 6. Here is Stockfish 6's PV:
11...Ne7 12.c4 d4 13.Kxc5 a5 14.Qa4+ Kd8 15.Qxa5 Rxa5+ 16.Kb4 Nc6+ 17.Kb3 b5 18.d3 bxc4+ 19.dxc4 e4 20.Bf4 Bf5 21.Ne2 Kd7 22.Bd2 Ra6 23.Rd1 Be6 24.Bb4 d3 25.Ng3 f5 26.Bxd3 exd3 27.Rxd3+ Kc8 28.Re1 Rd8 29.Rxd8+ Nxd8 30.Ne2 Bd7 31.Nf4 Ne6 32.Nd5 Kb7 33.Re5 c6 34.Ne7 f4 35.Nf5 Ra8 36.Kc2 g6 37.Nd6+ Kc7 38.Ne4 Nd4+ 39.Kd3
Stockfish 6 does not deviate from Stockfish 1.91 until 18.d3 when Stockfish 1.91 selected 18.Nh3. And both Stockfish versions selected Houdini 4's 17...b5. Of course, by now we know that White has to give up its queen by 15.Qxa5 in order to avoid mate. And, like Houdini 4, Stockfish evaluates that it is in Black's best interest to consolidate and establish an imposing pawn center rather than continue an all-out attack against White's king.
In fact, Stockfish 6 apparently evaluates Black's pawn center to be so threatening that it decides to give back some of its material advantage in order to destroy it by 26.Bxd3. And as a side benefit, also activates its White pieces and gains the initiative. So it seems like a sound evaluation to me and something that a (good) human player would likely do, particularly since White still retains an extra pawn.
This is the final position of Stockfish 6's PV:
click for larger view
White has the extra pawn and the more active pieces, and Black's k-side majority is not ready to advance. So Stockfish 6's evaluation seems reasonable, but the BOC might result in the game ending in a draw in spite of White's advantage.
But Stockfish disagrees. Restarting the analysis from the final position Stockfish 6 evaluates the resulting position at [+2.73], d=37 after 39...Ne6 40.Kc3 h6 41.b3 Ng5 42.Bd6+ Kc8 43.Nxg5 hxg5 44.Rxg5 Bf5 45.a4 f3 46.gxf3 Ra7 47.Rg1 Rd7 48.c5 Ra7 49.Re1 Rf7 50.b4 Bd7 51.Re3 Kb7 52.h4 Rh7 53.Bg3 Ka6 54.Kd4 Rf7 55.Kc4 Rh7 56.Kc3 Bf5 57.Re8 Rd7 58.Bd6 Rh7 59.Ra8+ Kb7
click for larger view
Three pawns down I don't think that Black can survive this, BOC notwithstanding, although White may lose its h-pawn. I don't know why Stockfish considered it necessary to give up 2 pawns by 41...Ng5 and 45...f3 since the incarceration of White's rook is over after 46.gxf3, but apparently Stockfish evaluates Black's position as being more desperate than I thought.
|Jun-09-15|| ||AylerKupp: <The Immortal Draw> (part 3 of 4)|
<Komodo 9.02> Komodo also evaluated 11...Ne7 to be Black's best move and evaluated the position at [+1.09], d=35, after 11...Ne7 12.c4 d4 13.Kxc5 a5 14.Qa4+ Kd8 15.Qxa5 Rxa5+ 16.Kb4 Nc6+ 17.Kb3 b5 18.d3 bxc4+ 19.dxc4 e4 20.Bd2 Ra8 21.Ne2 Bd7 22.Be1 Kc8 23.Nf4 Kb7 24.h3 f5 25.a4 d3 26.Kc3 g5 27.Nxd3 exd3 28.Bxd3 Ne5 29.Bc2 Bc6 30.Rf1 Be4 31.Bxe4+ fxe4 32.Rf5 Rhe8 33.Rxg5 Rad8 34.b3 h6 35.Rh5 Rd6 36.Bf2 Rd3+ 37.Kb4
Yes, of course it also found it necessary to give up White's queen to avoid mate and evaluated 17...b5 as Black's best move. And Komodo saves a tempo from Stockfish 6's 20.Bf4 by 20.Bd2, since Stockfish later plays 22.Bd2. Komodo also tries a different approach to activating Black's king and connecting its rooks by 21...Bd7, 22...Kc8, and 23...Kb7. And its 26...g5 is certainly the most aggressive attempt so far to activate Black's k-side and center pawn majority, although it only forces White to do what I think is the best thing to do, return some of its material in order to destroy Black's passed pawn center by 27.Nxd3. And I think that 35.Rg7 keeping the rook more active was better than 35.Rh5 since Black can protect its Ph6 without difficulty.
This is the final position of Komodo 9.02's PV:
click for larger view
I think that White stands better than Komodo's evaluation would indicate. White is up two pawns, its king is safe, and it has a B vs. N open board advantage. Black's only trump card is its advanced passed e-pawn, but it can also become weak.
But I am apparently wrong. After either 37...e3 or 37...Nc6+ 38.Ka3 e3
Komodo evaluates that it is necessary for White to give up its bishop for Black's advanced e-pawn either immediately by 38.Bxe3 or after it queens by 45.Bxe1. Restarting the analysis from the final position Komodo evaluates the resulting position at d=32 as follows:
1. [-1.94]: 37...e3 38.Bh4(A) Nc6+ 39.Ka3 Nd4(B) 40.Rb1 e2 41.Rb2 Re4 42.g4 Rxh3 43.Be1 Rd3(C) 44.Ka2 Rd1 45.Bc3 Nf3 46.Rxh6 Nd2 47.Bxd2 Rxd2 48.Rxd2 e1Q 49.Rhh2 Rxg4 50.Rhg2 Re4(D) 51.Rb2 Rd4 52.Rge2 Qa5 53.Rbc2 Rd3 54.Rb2 Rd1 55.Rb1 Rd7 56.Rbb2 Rd3 57.Rg2 Qc3 58.Rg5 Qf6 59.Rb5+ Kc8 60.a5(E) Qd6(F)
click for larger view
(A) At lower plies Komodo played 38.Bxe3 immediately.
(B) This position certainly doesn't look very good for White!
(C) 43...Rxh5 44.gxh5 Nf3 might be more straightforward. But "straightforward" means nothing to a computer.
(D) Apparently Komodo evaluates possible mating attacks more than "straightforward" simplification. But White is close to building a fortress, a concept that chess engines have difficulty with.
(E) White is apparently trying to win!
(F) I am not sure that Komodo can figure out a way for Black to win in spite of its great material advantage.
2. [-1.86]: 37...Nc6+ 38.Ka3 e3 39.Be1 e2 40.Ra2 Rd1 41.Bf2(A) Re3 42.a5 Rf1 43.a6+ Ka8 44.Rxh6 Rxf2 45.Rxe2 Rfxe2 46.Rxc6 Rxg2 47.Rxc7 Rg6 48.Kb4 Rxa6 49.Kb5 Ra7 50.Rc8+ Kb7 51.Rh8 Rxb3+ 52.Kc5 Ra5+ 53.Kd4 Rf3 54.Rh7+ Kc6 55.Rh6+ Kd7 56.Rh7+ Ke6 57.Rh6+ Rf6 58.Rh4 Kd7(B) 59.Rg4 Rd6+ 60.Ke4 Rc5 61.Kf3 Rd3+ 62.Kg2 Re5
click for larger view
(A) Apparently Komodo is more than willing to give up White's bishop for Black's e-pawn but Komodo's Black is not accommodating.
(B) After 58...Ra3 followed by 59...Rff3 White's h-pawn will fall and, if it comes to that, Black's king is still in the square to prevent White's c-pawn from queening.
|Jun-09-15|| ||AylerKupp: <The Immortal Draw> (part 4 of 4) - Conclusions|
The most recent released versions of Houdini and Stockfish generated lines and evaluations that were very similar to those generated by their earlier versions. And neither could seem to make White's small advantage a winning one.
Komodo, on the other hand, played the Black side more aggressively and efficiently than either Houdini and Stockfish, first forcing White to give up most of its material advantage (just like the other two engines did) and then obtaining a material advantage of its own after either 37...e3 or 37...Nc6+. But Komodo doesn't seem to play these endings very well in spite of its great material advantage, and in the first case (37...e3) maybe not even a winning one, although I think that Komodo should be able to win the second case (37...Nc6+) although it seems to make it more difficult than it needs to be.
Of course, like the earlier analyses, this is just a sequence of moves that COULD happen, not necessarily what WOULD happen. So, even though some variations resulted in a likely win for White, this is not conclusive. Many deviations are possible and they could change the assessment of the PVs.
And I'm reasonably sure that neither Hamppe nor Meitner conceived of the complicated game that would have resulted if they had not gone for the draw! So, <sofouuk>, if you want to try improvements for either side, I wish you luck.
|Jun-09-15|| ||sofouuk: I think it's clear that this is not really about Hamppe or Meitner anymore :) and 9...Qa4 was clearly the best move in the position, in the sense that it gave white the greatest probability of losing the game|
11...Ne7 is certainly better than the alternatives and then it's essentially forced until 17.Kb3. the immediate b5 should be the best reply bcz black knows for sure he's going to play that, in contrast to other moves, so the serious analysis starts here
click for larger view
certainly black's game plan has changed to hoping that the pawn center compensates for the material disadvantage, but as intuition and <AylerKupp>'s lines show, white won't be able to break the center up without sacrificing a piece, and black has to prevent white being able to do that with advantage. if white never manages to challenge the center his game will never get going, bcz the qside pawns are easily contained, and there is even the danger of being steam rollered through the middle - the central pawns are stronger than they might at first look
it's extremely difficult to analyse such positions bcz both sides are shadow boxing and there are so many plausible moves at each turn - in short it's the kind of position where good correspondence players can really shine :)
I'm going to state with confidence that 18.Nh3 is the best move, not just bcz it develops a piece but also there are variations where the ability to immediately jump to g5 or f4 restricts black's options. d3 looks slow and gives black the game he wants
... to be continued, I guess
|Jun-10-15|| ||sofouuk: well after playing around with variations most of the day i'm going to admit that i give up :) the evals seem to be biased towards white due to the material advantage, but in fact it's extremely difficult for white to develop and coordinate - black really does have a lot of play and white is always a tempo or two too slow (this is a bit surprising because in the starting position black's development is hardly better than white's). |
as one example of what rybka came up with (<AylerKupp> has posted similar lines already, or possibly even an identical one ...) after 18.Nh3 Ra8! (black is waiting for d3 and knows the rook must move after Bd2) 19.d3 bxc4 20.dxc4 Rb8+ 21.Kc2 Bf5+ 22.Kd1 Rb3!
click for larger view
it's hard to believe at first that this leisurely rook maneuvre is the best plan in the position, but now white is so tied down that rybka just shoves a3-a4 with Ra1-a3 follow, getting the black rook off b3 at the cost of doubled isolated pawns on an open file - it's hard to see white winning after that
this could mean that 18.d3 is better than Nh3, but in any case i'll withdraw the claim that 'white is clearly better' - in fact, nothing is clear bcz it's so hard for white to coordinate effectively and the two central passed pawns are worth more than a minor piece (if we value them at e.g. 3.5 points white would still have a material advantage, but only if he can successfully deploy that material)
in which case i'll give white credit for playing Kb4 over Kb5 in the first place - Kb5 would've significantly increased the chance of losing while perhaps only slightly increasing the chance of winning, and the quick fire draw by repetition is a far more attractive finish to boot
|Aug-29-15|| ||rwbean: Great analysis by <sofouuk> and <AylerKupp> ... I suggest just running a tournament with the latest versions of Stockfish, Komodo and Houdini after 11. ♔b5 ♘e7 as this seems to be the most interesting position?|
|Aug-31-15|| ||rwbean: After 11. ♔b5 ♘e7 12.c4 d4 13.♔xc5 a5 14.♕a4+ I started a 20 game tournament:|
Stockfish 6 vs Komodo 8 using Scid vs Mac on a Macbook Pro, 1 GB each of hash, 30 seconds each for moves ... White won 17 games, 3 were draws. A lot of the games looked like easy "fortress" draws for Black to me (e.g. White pawns on a3, b2, d2, White rook on a1, White bishop on c1, Black bishop on b3, Black rook on a4, Black pawns on d3, e4) but I must have been mistaken... unless I should run it all again with a longer time control?
|Sep-05-15|| ||rwbean: I finished a 10 game tournament at 5 minutes per move with the same programs. Unfortunately I've noticed that Stockfish 6 is crashing a bit.|
This time (making it a draw if the evaluation is fairly balanced and Stockfish crashed) it got 5 wins for White and 5 draws. There are lots of positions which look like easy fortress draws but the programs don't seem to understand too well.
|Jun-24-16|| ||rwbean: How about some other people run some tournaments after these two positions: 11. ♔b5 ♘e7 12.c4 d4 13.♔xc5 a5 14.♕a4+ |
(a) 14... ♔d8 15. ♕xa5 Rxa5+ 16. ♔b4 ♘c6+ 17. ♔b3
(b) 14... ♔f8 15. ♕xa5 Rxa5+ 16. ♔b4 ♘c6+ 17. ♔b3
It can split up quite a lot after that, there are many reasonable moves for Black in either position. I'm interested to know how well programs understand "fortress" draws.
|Feb-13-17|| ||Straclonoor: <rwbean: How about some other people run some tournaments after these two positions: 11. ♔b5 ♘e7 12.c4 d4 13.♔xc5 a5 14.♕a4+
(a) 14... ♔d8 15. ♕xa5 Rxa5+ 16. ♔b4 ♘c6+ 17. ♔b3 ... I'm interested to know how well programs understand "fortress" draws>
Below is the line. Final position of the line don't look like fortress
Stockfish 110217 64 BMI2 1.30 (depth 40) 11.Kb5 Ne7 12.c4 d4 13.Kxc5 a5 14.Qa4+ Kd8 15.Qxa5 Rxa5+ 16.Kb4 Nc6+ 17.Kb3 b5 18.d3 bxc4+ 19.dxc4 Ra6 20.Bd2 e4 21.Ka2 Ne5 22.Re1 Bb7 23.c5 Ra8 24.Bf4 Ng6|
|Mar-04-17|| ||rwbean: I put it through Stockfish as well again with at least 8 hours per move ... got|
14 ... ♔d8 after 49 ply
17... b5 after 48 ply
18. d3 after 48 ply
18... bxc4+ after 49 ply
19. dxc4 is forced
19... e4 after 47 ply (although it does like 19... ♖a8 too)
20. ♗d2 after 46 ply
20... ♖a6 after 46 ply
21. c5 after 47 ply
21... ♗e6+ after 49 ply
22. ♗c4 after 52 ply
evaluation is +1.55 to +1.65
11. ♔b5 ♘e7 12.c4 d4 13.♔xc5 a5 14.♕a4+ ♔d8 15. ♕xa5 Rxa5+ 16. ♔b4 ♘c6+ 17. ♔b3
b5 18. d3 bxc4+ 19. dxc4 e4 20. ♗d2 ♖a6
21. c5 ♗e6+ 22. ♗c4
and the PV is
22... ♔c8 23. ♗xe6+ fxe6 24. ♖e1 e3 25. ♗xe3 fxe3 26. ♖xe3
White is a pawn up and has a passed a pawn versus Black's passed e pawn, and has two pawn islands versus Black's three. There doesn't seem to be a way for White to set up a fortress. White is just clearly better.
|Oct-15-17|| ||Lambda: My Stockfish likes 9. d4 exd4 10. Bxa6 bxa6 11. a3 and it thinks white is about +2! Taking the pawn on g2 apparently doesn't help. |
click for larger view
|Feb-08-18|| ||kishore4u: What is this??|
|Feb-08-18|| ||ughaibu: <What is this??>|
A question mark.
|Feb-09-18|| ||zborris8: zing!|
|Feb-09-18|| ||Howard: Larry Evans ran at least 3-4 questions about this game in his CL&R column back from about 1975-77, as I recall.|
In fact...I just came across one of those questions last night when perusing some back issues.
|Jun-03-18|| ||The Kings Domain: :-) Fun game.|
< Earlier Kibitzing · PAGE 11 OF 11 ·