|
< Earlier Kibitzing · PAGE 4 OF 4 ·
Later Kibitzing> |
| Jan-07-10 | | zanshin: This is an interesting game to compare human versus engine analysis. I have read a few reviews and can certainly understand <psmith>'s concern that <With this you reject years of GM opinion of the game.> <GM Keene on 18.Nh1: !! A wonderful idea. White has in mind the manoeuvre Nh1-f2-h3-g5, in conjunction with Qh5, as a method of assaulting the position of Black's king. When I first read My System I was so impressed by this game that I deliberately created situations in my next few games where the move Ng3-h1 was possible> 
click for larger viewI think there are two main reasons an engine would not suggest <18.Nh1>: 1) horizon effect, i.e. the advantage is too far down for the engine to evaluate properly; or 2) the advantage is positional and cannot be quantified properly. Sometimes, I have found that when you start the analysis at the point the advantage is apparent, you can slide backward to the move in question. Having evaluated the outcome, the engine can now see the value of the move. I did not find it in this case. Therefore, I am left with the preliminary conclusion that the engine cannot evaluate the positional advantage properly. Sliding forward again from move 18, Rybka reports no significant White advantage all the way to move 26: 
click for larger view [+0.36] d=20 26...Be7 27.Qxd4 Bf6 28.Qc5 b6 29.Qf2 b5 30.Bb3 Re8 31.Ng5 Rxe2 32.Qxe2 Qd6 33.Kf1 Bd5 34.Bxd5 Qxd5 35.Be3 Qd6 36.Qd2 Bxg5 37.fxg5 Qxh2 38.Qc3 Kg8 39.Qf6 Qc7 40.Qxa6 Qd7 41.Qa3 Ne6 (0:07.20) 61379kN Here, <26...Be7> (protecting h4?) seems to be fine, but <20...Bb6?> looks like a mistake that White failed to take advantage of ([+2.66] d=16 27.Qe1 (0:03.22) 21002kN). So I conclude that White gained no long term advantage from <18.Nf1>. If the GMs are correct, then Rybka did not find it. |
|
| Jan-07-10 | | zanshin: Engines are more reliable in tactical analysis, so everyone pretty much agrees that White missed <27.Qe1>. <ray keene: it is inevitable that modern computer anlysis will find corrections to classic games and variations such as this one-here 27 Qe1 seems to be a genuine improvement , but apart from that nimzo did ok in the game > 
click for larger view [+3.13] d=17 27.Qe1 Be4 28.Bb3 Re8 29.Nf2 Nc6 30.dxe4 d3 31.Re3 h5 32.e5 Bxe3 33.Bxe3 Rd8 34.Qc3 d2 35.Bxd2 Qe7 36.Nh3 Nd4 37.Kf2 Nxb3 38.axb3 Kg8 39.Ng5 Qd7 40.Qc4 Qd5 41.Ke3 Qxc4 (0:12.15) 76135kN But since he played <27.Qh4> instead, White's advantage is slight ([+0.65] d=18). All the way to move <28.Re5 (!)> the eval is only +0.24. White takes the lead for good after <31...Bxe8> when Qxe8 was probably better. [+1.26] d=15 31...Qxe8 32.Qxh7 Kf6 33.a4 Bd5 34.Kf1 Ba7 35.h4 Qc6 36.Kg1 Qe8 (0:01.43) 14017kN [+3.64] d=15 31...Bxe8 32.Qe1 h5 33.Qe7 Bf7 34.Qe5 Kh6 35.Qf6 Bxa2 36.Qxb6 Qg7 37.Nf3 Kh7 38.Ne5 Bb3 39.Qxd4 Qf6 40.Qd7 Qg7 41.Qd8 Qg8 (0:03.36) 30556kN Undoubtedly, there were some inaccuracies in this game - which should not detract from its value. However, I agree with <RV> to a certain extent that <18.Nh1> was not the decisive move and really contributed little to the end result. |
|
| Jan-07-10 | | zanshin: One last note:
<!! Who would expect the death-blow to come from this quarter? If Black plays 34..axb5 he is mated as follows: 35 Ne6 h5 36 Qf6+ Kh7 37 Ng5+ Kh6 38 Bb4! In view of this, Rubinstein elects to surrender a piece but that too is obviously without hope.> Rybka evaluates <34.Nf7+> as slightly better (admittedly low ply): 
click for larger view [+5.67] d=15 34.Nf7 (0:03.17) 38115kN
[+4.62] d=14 34.b5 Qg7 35.Qe6 h6 36.Nf7 Kh7 37.bxc6 bxc6 38.Ne5 (0:01.40) 18391kN |
|
| Jan-08-10 | | psmith: <zanshin>
Let us consider again the position at move 26.
click for larger viewRubinstein played 26...Bb6 which loses to 27. Qe1. Better is 26...Be7. You say "(protecting h4?)" but that seems not to be the proper reason for the superiority of 26...Be7 since after 26... Bb6, Nimzo's 27. Qh4 is the weaker move. You give a line from Rybka indicating only a slight White advantage after 26...Be7. Rubinstein (in I think a very human response) clearly chose 26...Bb6 to protect the pawn on d4, which falls with check after 26...Be7 27. Qxd4+. Whether White's pawn advantage is really good enough after that is unclear. But the line from Rybka is by itself not all that convincing. The reason is that there are many alternatives after each move, and as a computer generated line gets longer and longer the likelihood that a move-by-move analysis would diverge from the line generated gets stronger and stronger. For example in the line you give: 26...Be7 27.Qxd4 Bf6 28.Qc5 b6 29.Qf2 b5 30.Bb3 Re8 31.Ng5 Rxe2 32.Qxe2 Qd6 33.Kf1 Bd5 34.Bxd5 Qxd5 35.Be3 Qd6 36.Qd2 Bxg5 37.fxg5 Qxh2 38.Qc3 Kg8 39.Qf6 Qc7 40.Qxa6 Qd7 41.Qa3 Ne6, it seems that White might improve with 31. Rxe8+ Bxe8 32. d4 followed by d5. Black may have ways to hold this but further analysis would be required. A bit further down in this Rybka line, 33. Kf1 is pretty hard to understand. 33. Nf3 might be an improvement. On the other hand, why is 26... Be7 superior to 26...Bb6? Strange as it may seem one should answer that the reason is that 26...Bb6 gives up the defense of e7 (which is why 26...Bb6 27. Qe1 is so powerful -- Black has no good defense to 28. Re7+). After 26...Be7, although the Bishop now occupies e7 rather than defending it, it also blocks the e-file, allowing 27...Re8 (defending e7) in response to 27. Qe1. Even Nimzo's move 27. Qh4 is aimed at exploiting the weakening of e7. But it allows 27... Re8 whereas 27. Qe1 does not. The decisive mistake was then 28...Nf7 whereas Black could have continued to hold with other moves such as 28... Qd7. So the real question is: which square is it more important to defend, d4 (where a pawn can be captured with check) or e7 (where a check with the Rook can win the Queen)? The answer: e7. And this is real insight, gained from computer analysis, not just a questionable line and and an eval. Or so I claim. |
|
| Jan-08-10 | | zanshin: <And this is real insight, gained from computer analysis, not just a questionable line and and an eval. Or so I claim.> <psmith> I think we can all agree that you cannot really understand a game by simply leaving your engine on infinite analysis and examining the results - especially not a complex game like this. Engines can be useful in complementing human analysis, checking for tactics, etc. - as I'm sure you are well aware. And I agree that moves further down a line are going to be less reliable. I read a book where the author claimed that the only move with any reliability is the first one. Before we go too far here, I would like to address one point at a time. The first and most obvious one deals with <18.Nh1>. What is your assessment of this move? Is it the brilliancy claimed by past analysts, or has <RV> dared to say the emperor has no clothes and this move is suspect? Having gone through the game, my sense is that <RV> is on to something (although I would probably have couched my language differently ;-)) |
|
| Jan-08-10 | | psmith: <zanshin>
Well, I think the question is complex. The move was played in a human-human game. In that context concepts apply that don't really have application in a computer chess context. For example, the concept of orignality, the concept of a long-range plan, the concept of chess-cultural historical context, and finally the concept of psychological effect or surprise.It seems to me Nh1 impressed someone like Keene for a variety of reasons: it is an original, and to a human eye (especially in its chess-cultural historical context) an "unnatural" move -- especially played against a classical player like Rubinstein. It may have had a strong psychological effect in the context of the game. It involves a long-term plan -- and it worked in the game. It befuddled a very strong opponent. I think RV is probably right that the move Nh1 is not a winning move, objectively. But what isn't clear to me is that the alternatives are really so much better than Nh1 that it could be described as squandering away a small advantage in order to appear clever. In fact I think it is still open that Nh1 leads to White advantage. But even if that turns out not to be true, Nh1 could still be in many ways described as a brilliant stroke in the human-human context. |
|
| Jan-08-10 | | zanshin: <psmith> In case you haven't seen it, here's an often-cited analysis of this game to add to this dicussion, Dvoretsky's 'Worst piece principle' http://www.chesscafe.com/text/dvore...
You have raised some valid points, but also a few more questions. I have come to the conclusion that Nh1 is not a move any engine will find because engines often include a penalty for Knights on the edges. I can't imagine the penalty for a Knight on a corner! If Nh1 was played to confound a classical player, this suggests Nimzo was playing the opponent and not the board. Dvoretsky argues that it's the start of a long term sequence to improve a position, which implies that the move is sound, regardless of opponent. I have to say that I could not find any measurable advantage arising from this move. So if Nh1 is psychological, then there is no way we can measure the advantage. Analysis by engines and/or humans can not predict whether an inferior or unsound move may lead to a win because it unnerved a specific opponent. Finally, I also agree with you that there are no clear alternatives - at least I could not find any. |
|
| Jan-08-10 | | zanshin: <So the real question is: which square is it more important to defend, d4 (where a pawn can be captured with check) or e7 (where a check with the Rook can win the Queen)? The answer: e7.
And this is real insight, gained from computer analysis, not just a questionable line and and an eval. Or so I claim.> <psmith> I've gone through this position and studied your comments. I agree with you. It's a moot point now, but Rybka was not the only engine that suggested <26...Be7> [+0.42] d=23 26...Be7 27.Qxd4 Bf6 28.Qc5 b6 29.Qf2 h6 30.Qe1 b5 31.Bb3 Bb7 32.Kh1 Qc6 33.Ng1 Re8 34.Qf2 Rxe2 35.Qxe2 Qd6 36.Nf3 Bd5 37.Bxd5 (1:34.28) 753923kN (Rybka 3) [+0.34] d=21 26...Be7 27.Qxd4 Bf6 28.Qf2 b5 29.Bb3 Re8 30.Rxe8 Bxe8 31.Ng5 Qd6 32.Qa7 Bd7 33.Be3 Qxd3 34.Kf2 Nc6 35.Ne6 Kh6 36.Nc5 Nxa7 37.Nxd3 Nc6 38.Bd5 Kg7 39.g3 h6 40.Bg2 Be8 41.a3 g5 42.Nc5 (1:47.03) 79787kN (New free engine) [+0.64] d=14 26...Be7 27.Qxd4 Bf6 28.Qf2 b5 29.Bb3 Qd7 30.Ng5 Qxd3 31.Re3 Qb1 32.Re1 Qd3 33.Bc1 Re8 34.Rxe8 Bxe8 35.Nf3 Qc3 36.Bd2 Qc7 (1:34.12) 3582kN (Stockfish 1.6) After <26...Bb6 27.Qe1> Rybka suggests saccing the Bishop because of the threats of Re7+ and Qe5+. 
click for larger view |
|
| Jan-08-10 | | psmith: <zanshin> I hadn't seen the Dvoretsky article. He says "27. Qe1 was the more energetic and stronger (observe the typical placement of heavy pieces on the open file, with the Queen behind the Rook) Be4 28. Nf2  (Nimzowitsch)"! So the improvement was known to Nimzo himself, though I don't know where that was derived from. |
|
| Jan-09-10 | | zanshin: <psmith> I think we agree on the key moves of the game so far. There are a few more moves where we can quibble as to whether Nimzo was playing the man or the board, but I'd just like to wrap this up for now with the other !! move <34.b5> White to play move 35:
click for larger viewHere are a few engine evaluations:
[+7.11] d=23 34.Ne6 Bd8 35.Nxd8 h5 36.Ne6 Bd5 37.Qf6 Kh7 38.Nf8 Kh6 39.Qg5 Kg7 40.Nxg6 Qe6 41.Nh4 Kf7 42.Qxh5 Kg8 43.Nxf5 Bxa2 44.Nxd4 Qf6 45.Qe8 Kh7 46.Qe4 Kg8 47.b5 axb5 48.Nxb5 Qa1 49.Kf2 Qb2 50.Qe8 Kh7 51.Qd7 Kh6 (0:09.05) 398577kN (Fruit 2.3.1) [+5.63] d=17 34.Nf7 Kg7 35.Nd8 Kh8 36.Nxc6 Qg7 37.Qe8 Qg8 38.Qe5 Qg7 39.Qb8 Qg8 40.Qxb7 Qxa2 41.Qe7 Kg8 42.Qf6 Qf7 43.Ne7 Kf8 44.Nxg6 hxg6 45.Qxb6 Qb3 46.Qf6 Kg8 47.Qxg6 Kh8 48.Qf6 Kg8 49.Qg5 Kh8 50.Qd8 Kh7 51.Qe7 Kg8 52.Qe2 Kf8 53.Be1 Qb1 54.h4 Qc1 55.Qe5 Qe3 56.Bf2 Qc1 57.Kh2 (0:57.35) 8305kN (New free engine) [+6.39] d=15 34.Ne6 (0:03.24) 16842kN (Rybka 3)
(9.57) Depth: 20/41 00:04:44 437mN
34.Ne6 Bd8 35.Nxd8 Qg7 36.Nxc6 Qxe7 37.Nxe7 Kg7 38.Nc8 Kf6 39.Be1 g5 40.fxg5+ Kxg5 41.Nd6 Kf6 (Fritz 10) The engines favor either <34.Nf7+> or <34.Ne6> as objectively strongest in this position. What's your opinion of <34.b5> then? Was it more of a psychological ploy because it again looks surprising and unnatural? Or is it possible Nimzo knew he had a win and chose a more unorthodox move to make a statement? |
|
| Jun-29-10 | | vintage geisha: I tried reading through all the analysis, but it just seemed to be more a more chess engines without anyone hitting this point dead on. With h6, I believe, black has created a light square complex on his kingside. Moves such as Qh5 are now far more devastating, keeping in mind that his bishop on c4 covers the g8 square, leaving the black king with little to no movement (h7), and a white knight which is fast coming up to the g5 square. Clearly, if black did play 1...kh6 at some point in hopes of supporting a g6 advance, white could follow with 2.Ng5 when the black king would have to return to his lowly position on h8 since 2...hxg5 would be followed by 3.Qh5# I'm sorry I couldn't present it in a more easy to read format, but it beats endless lists of computer eval's, don't you think? To conclude, h6 really doesn't prevent the knight from falling on g5; in order to prevent a further invasion of blacks camp, with moves like Qh5, Ng5 and Qg6 threatening mate, black would have to make serious concessions anyway. Hope this was insightful, because I, too, was overcome at first by the idea that if black had just played h6, this whole long knight maneuver would have be in vain. |
|
| Nov-28-11 | | albertfrank: Anyway, this game is not fantastic as it looks: 27.Qe1! (instead of 27.Qh4) was wining immediately. |
|
| Jan-14-14 | | nhnsn: Hello, I was analizing(Not whit a computer) the coments to Keene about 34.b5. At the end of his variant, what if Black plays 38...h4 ? I don't see any way to force the win, unless not quickly. A better variant, I think, is 35.♗b4!, and there is no way to prevent the bishop arrives to d6 and then e5, delivering the coup de grâce. |
|
| Jan-14-14 | | nhnsn: I rectify myself, after 35.♗b4 follows the simple 35...♕g7 36.♘f7+ ♔g8 and the Bishop doesn`t arrive on time. Anyway,Kenne's variant does not win either. |
|
Mar-01-14
 | | LIFE Master AJ: The great Nimzowitsch (himself) considered this his most elegant attacking game ... and over no less than a Rubinstein! |
|
| Sep-12-15 | | peirce: Hello everybody.
I would like to know how much time used Nimzowitsch to conceive
and move the 18 th move .
I have an hypothesis about the
problems which the need of the
clock is giving to players.
But first , I would like to
build my idea on a datum .
Is there a way to find out how much time Nimzo used , just as today for some tournaments the time used is indicated.Thank your for your answers. |
|
Feb-05-17
 | | plang: A terrific tournament who scored 8.5-0.5 with the draw against Alekhine. I think this was the first game where 5 e4 was played which has now become a main line. Nowadays 6..Nd3+ is the most popular move. 11..Bd6 was an inaccuracy; 6..Be7 was more solid. Nimzovich acknowledged that 12 Qg4 would have been powerful for White and if 12..0-0? then 13 Bg5..Be7 14 Bh6..Bf6 15 e5!..Bxe5 16 Bxg7..Bxg7 17 Nh5 winning. 26..Be7 giving up the d-pawn but strengthening the kingside defense may have been a tougher defense. 34 Ne6 would have been even stronger; ie. 34..h5 35 Qf6+..Kh7 36 g4!. A really nice game; the way Nimzovich alternated between positional and attacking play was very instructive. |
|
| Jul-05-21 | | George Wallace: < psmith: <RandomVisitor>: Hi. I am not looking for a fight, but we can disagree. I sometimes do not find I get any insight from your posting of Rybka lines and evals. (not always, but sometimes!)
On the basis of Rybka's evals you wrote: "White dissipates whatever advantage he has with the clever looking 18.Nh1 - 19.Nf2 positioning, but this just allows black good equalization chances with 19...h6 or 19...g6." and also "After 21...h6 black might even have a tiny advantage..." This on the basis of a Rybka eval of +.31 in two lines and of -.1 in another line. With this you reject years of GM opinion of the game. My point is first of all that in such positions it is not clear that Rybka evals are to be trusted over GM evals. Not yet.> I read this comment and decided to enter the position into Stockfish 14, which must have just been released, because I uploaded Stockfish 13 not long ago at all. I have the engine running on 4 cores, 16 gig of ram. The first thing Stockfish analyzes is 18. a3 and 19. h3. Then, after a few minutes of crunching on this position, it changed. Depth=37
=(0.26)
18. Nh1 h6
19. Nf2 Bd7
20. Nh3 Rae8
21. Rxe8 Bxe8
22. Ng5
I can't believe it found this idea. |
|
| Sep-17-21 | | DouglasGomes: 21.. h6 proves to be a controversial "computerish" move, but it seems to appears in all the variations. I then had this idea: Black can try to play Na5 for Bc6 to get rid of light squared bishop. But in order to do we have to contest the dark squared bishop. Black queen xray on c1 guarantees we can trade them. Hence I propose 21.. Bb4: 21... Bb4 22. Nh3 [illustrative] Bxd2 23. Rxd2 Na5 24. Bd5 Bc6 25. Re2 Qd7 26. Bxc6 Nxc6 27. Ng5 h6 28. Ne6 Rf6 [controls e6 square] 29. Nc5 Qd6 [black had to sort out back rank issues] However:
21... Bb4 22. Bxb4 Nxb4 23. Nh3 h6 24. Ng5 b5 25. Nf7 Kh7 (only move to hold the balance) 26. Ng5+ [perpetual] Stockfish however AVOIDS the draw and suggests a continuation that allows Black to play on. 21.. h6 22. Nh3 b5 23. Bb3 Bb4 24. Be1 Re8 25. Rxe8 Bxe8 26. Qe2 Bf7 27. Bxf7 [the exchange is forced] Qxf7 28. Bxb4 Nxb4 26... Qe7 is a alternative that would force the trade of queens. Advantage for Black is nonexistent. There are some ways for White go wrong, one pretty sloppy way for White to play that allows counterplay:
24. Bxb4? Nxb4 25. Ng5 Qc1+ 26. Kf2 Bc6 27. Qg3 [not allowing f4 to fall with check] Qb1 28. Ne6 [only move for White to not be complete losing and now Black is better] Stockfish likes 18. a3 unsurprisingly. |
|
| Sep-18-21 | | SChesshevsky: <...Stockfish likes 18. a3 unsurprisingly.> Since whites pieces seem so well placed compared to blacks at move 18, I'm guessing today's top GM's would also be suspicious of 18. Nh1. Willing to give black time to improve in a dynamic position in exchange for inconclusive threats on the king? Think the key here is for black to use that time to sort out position, exchange pieces to reduce threat, and best case make the N look bad for trying the tour. Maybe something like 20...b5 21. Bb3 Rxe2 22. Rxe2 Bb4 23. Bxb4 Nxb4 24. Re1 Bc6 does the trick and meaningfully reduces whites opportunities. |
|
| Jan-09-22 | | Mathematicar: Wonderful game. That last tactics, with Nh1 move, is also very instructive. |
|
| Dec-15-22 | | Seyealebiosu1985: I am not a fan of the nimzo opening https://cpnseomarketing.com.ng/best... |
|
| Dec-16-22 | | sfm: <George Wallace: ... psmith: ...RandomVisitor...
"My point is first of all that in such positions it is not clear that Rybka evals are to be trusted over GM evals". Not yet." ... 18. Nh1...
I can't believe it found this idea.
>
Wow! Amazing.
In any case, we are humans, and with human limitations.
This is an important reality, which computers do not as of now consider on our behalf. "The strongest move" is the move most likely to work for the player, against the opponent, playing that game. |
|
Dec-17-22
 | | kingscrusher: I absolutely love Nimzovich and the principle of rerouting a knight - and taking sometimes backward steps to go forward later - like some highly celebrated Karpov moves like Nb1. So Nimzovich has a profound influence on modern positional players and the principle of "Improve your worst piece" as a default positional plan. This is all brilliant and useful to any modern positional players. But - it seems as though 13 Qf3 was a bad inaccuracy. The Knight on g3 could have been beautiful and functional on e4. E.g. 13 a3 (kick b4 knight whilst Queen still covers c2) Nc6
e5 (vacate e4 for the g3 knight)
Bc7
Qh5 Ne7
16.Ne4
and it reminds me of my Knights tango with White (I like playing 1.nc3 with a view of encouraging d5-d4 from black and having a knight often at g3 later). It seems very pleasant for White here. Aron Nimzowitsch - Akiba Rubinstein 1-0 5.0, Dresden Dresden GER 1926
click for larger viewAnalysis by Stockfish 15 - 3 threads max:
1. +- (2.19): 16...Nf5 17.Bd2 Qh4 18.Qxh4 Nxh4 19.Rfc1 Bb8 20.a4 Bd7 21.b4 Bc6 22.Kf2 Ng6 23.g3 Rd8 24.b5 axb5 25.Bxb5 Bd5 26.Bb4 Bxe4 27.dxe4
White is clearly better
(Gavriel, 17.12.2022)
Okay but as Raymond Keene mentions - it is so easy to be critical of classic games with modern engines. The principle of "improving the worst piece" remains though and is kind of "immortal". And here it is emphasized more as it is not just a backward knight move, but a knight move to a corner square. So kind of like Karpov knight retreats to the maximum. The wider principle is that sometimes one has to take some steps back for a net forward in the end. So Nh1 is a great example of this too as well as "improving one's worst piece". |
|
| Dec-20-22 | | sfm: BTW, one of the best puns ever. |
|
 |
|
|
< Earlier Kibitzing · PAGE 4 OF 4 ·
Later Kibitzing> |
|
|
|
Kibitzing
Annotations by 