< Earlier Kibitzing · PAGE 3 OF 3 ·
|Aug-20-11|| ||Creg: I'm not going to pretend to see all the lines, but I think it best for white to ensure an open diaganol on the a1-h8 line for his bishop.|
22.Nd5 exd5 23.Bd4 Ng7 (else 24.Nh6+ is mate) Now things get messy. Is it 24.Nh6+ Kh8 then 25.Qh5. The idea is to play 26.Nf7+ followed by 28.Qh6, then there's all types of mate threats, including one line that brings a rook lift via f3 and h3.
This is as far as I'm going to give this one a try.
Why am I not surprised I got this wrong... :-)
|Aug-20-11|| ||agb2002: White has a bishop and a knight for the bishop pair.|
White would be insterested in controlling the dark squares around the black king (if the bishop were on d4 then 22.Nh6 would deliver mate at once). Therefore, 22.Nd5 exd5 (the queen can't defend the DSB) 23.Nh6+:
A) 23... Kg7 24.Qf7+
A.1) 24... Rxf7 25.Rxf7+ Kh8 26.Bd4+
A.1.a) 26... Bf6 27.Rxf6
A.1.a.i) 27... Nxf6 28.Bxf6#.
A.1.a.ii) 27... Ng7 28.Rf7 Qc5 (28... Rg8 29.Bxg7+ Rxg7 30.Rf8+ Rg8 31.Rxg8#) 29.Bxc5 dxc5 (29... Be8 30.Rf8#) 30.Rxd7 + - [R].
A.1.a.iii) 27... Qc5 28.Rf8#.
A.1.b) 26... Nf6 27.Rxf6 looks similar to A.1.a.
A.1.c) 26... Ng7 27.Bxg7#.
A.2) 24... Kh8 25.Qxe7 with multiple threats: 26.Bd4+, 26.Rxf8+, 26.Rf7, etc.
B) 23... Kh8 24.Bd4+
B.1) 24... Rf6 25.Qxf6+
B.1.a) 25... Bxf6 26.Rxf6 transposes to A.1.a.
B.1.b) 25... Nxf6 26.Rxf6 transposes to A.1.b.
B.2) 24... Bf6 25.Qxf6+
B.2.a) 25... Rxf6 26.Rxf6 transposes to A.1.a.
B.2.b) 25... Nxf6 26.Rxf6 Rg8 27.Rf8#.
B.3) 24... Nf6 25.Qxf6+ transposes or looks similar to previous lines.
B.4) 24... Ng7 25.Qf7
B.4.a) 25... Rxf7 26.Rxf7 transposes to A.1.a.ii.
B.4.b) 25... Bf6 26.Qg8+ Rxg8 27.Nf7#.
|Aug-20-11|| ||kevin86: The intermezzo of the knight sacrifice divert the black pawn from interposing and blocking the attack on the long diagonal.|
|Aug-20-11|| ||realbrob: <sevenseaman> I played the position against DroidFish, which is damn good to be a free chess program for Android, and it also answered 22..Qb8. There is no big tactical shot after that, but I concluded that 22..Qb8, 23.Nxf7+, I'm up a piece, that's enough to win.|
|Aug-20-11|| ||sevenseaman: <realbrob> <22..Qb8. There is no big tactical shot after that, but I concluded that 22..Qb8, 23.Nxf7+, I'm up a piece, that's enough to win.>|
Thanks, the mathematical consolation will have to do.
|Aug-20-11|| ||Marmot PFL: <Sorry - I remebered this one, I had seen it before. (Its one of Geller's more famous combinations, I think its in his book.)>|
You have a good chess memory. I have the same book but didn't recall the game.
All the same, I thought this looks like Sicilian, so Nd5 comes to mind. (This from another book, by Levy, Sacrifices in the Sicilian.) But what purpose could that serve? Then I saw Nh6+ and Bd4+, and after ed5 this pawn can't block the bishop. But even here it isn't easy. 24 Bd4+ Nf6 25 gf6+ Bxf6 26 Bxf6+ Rxf6 27 Qxf6+ Kxh6 28 Rf4 wins easily, but what about 24...Bf6 25 gf6+ Kxh6+ 26 Be3+ g5? Geller solves this with the brilliant 24 Qf7+, a move I missed.
|Aug-20-11|| ||jheiner: 22.White to play. Geller vs Anikaev, 1979|
Material. Even. Nf7 is loose, otherwise everything is solid.
Position. White has a space advantage. White has infiltrated the squares around the Black K. White has control of some dark squares around the K. The White K is secure from immediate checks. The long dark diagonal has been opened up, but no one has claimed it yet.
Candidates: Nh6+, e5, Bd4, Nd5
22.Nh6+ is for later. Once the dark squares are secure, this move becomes mate.
22.e5 closes the dark diagonal, which White does not want. But how to keep Black from e5?
If 22...e5 23.Nd5 Qxc2 (moves?) 24.Nxe7+ Kg7 wins a B with a strong attack
If 22...Ng7 23.Nh6+ Kh8 24.Bxg7+ Kxg7 25.Qf7+ Rxf7 26.Rxf7+ Kh8 and I am not sure how to proceed.
If 22...Ng7 23.Nh6+ Kh8 24.Qf7 Rxf7 25.Rxf7 threatns Bxg7#
Black has multiple spoilers.
25...e5 26.Nd5 but now Qd8 is possible.
This looked like the win, but let us look at the last candidate and alternative move orders.
If 22...exd5 (else Q moves and 23.Nxe7+ wins) 23.Bd4 threatens Ng7#
If 23...Ng7 24.Nh6+ Kh8 24.Bxg7+ Kxg7 25.Qf7+ Rxf7 26.Rxf7+ Kh8 again doesn't work.
I've got my eye on Qh4 at the right time. After Ng7 when the h-file is open.
If 23...Ng7 24.Qh4 threatens Qh6 and capture on g7.
24...Rxf7 25.Rxf7 Kxf7 26.Rf1+ Ke1 27.Qxh7
What is the count? White is 2N for R and P with control over the long dark diagonal, menacing the Ng7 which has no escape. The White K is still secure. White also threatens checks on the back rank winning yet another R.
This is not decisive, but the extent of my board vision. Time to check.
Hmmm. going to have to check the kibitzing. I think I had the main idea, but 23...Kg7 allows a series of checks leading to mate. I wonder if my Kh8 and Ng7 was stronger defense by Black.
|Aug-20-11|| ||Once: <erniecohen> For starters, there's today's puzzle. I found 22. Nd5 much faster than Fritz 11, and I suspect many of us here were the same. Of course, Fritzie analysed the variations far better than I could, but that's not really the point.|
In general, many of the Sunday and Saturday POTDs are ones that computers take a little while to find. This is particularly noticeable where there is a long series of moves before the combination restores material parity.
|Aug-20-11|| ||OhioChessFan: Wonderful deflection of the e Pawn. I might have found it if I hadn't given up so quickly. I kept wanting to start with Nh6+|
|Aug-20-11|| ||erniecohen: <Once> Houdini 1.5 zeroes in on 22. ♘d5 as the best move in about 1/20 of a second. It reaches an evaluation of +7.7, which is pretty close to its final value, in under half a second. I think that't probably at least 2 orders of magnitude better than any human. |
I just ran a dozen random Sunday puzzles, and the worst time to solve one was about 2 seconds, and most of them were well under a second. You have some hard ones in mind?
|Aug-20-11|| ||TheRavenPK: <Once> I am 19 years old, but I am an old school not-computer user. It can help, but mostly I think it kills the chess. Young people of my age are totally obsessed with finding tactics and all that sht and absolutely unable to find anything on their own. I am a -great- fan of Mikhail Tal's style "I make this move because I like it, and you prove me it is wrong". Not only in chess, but also in life I really hate "waiting for luck to turn up, because the life becomes very boring"..|
All I wanted to say - I support your theory. :)
|Aug-20-11|| ||Once: <erniecohen> 1/20th of a second? That's fast! Fritz 11 on my PC takes 50 seconds to find the same move.|
But if Houdini is that fast then 2 seconds to solve a Sunday POTD is an eternity, relatively speaking.
There used to be a number of set puzzles that computers couldn't solve - eg where the winning move needed more than a certain number of plies to be evident. I wonder if these still work for monsters like houdini?
|Aug-20-11|| ||David2009: Geller vs Anikaev, 1979 White 22?|
I am very late to this puzzle. 22.Nd5!? forks Q and B so forces 22...exd5 23.Nh6+ K moves Bd4+ and Black cannot block the check with the e-pawn.
Main variation: 22.Nd5 exd5 23.Nh6+ Kh8 24.Bd4+ Ng7 when White has at least a draw by perpetual check (in the line 25.Qxf8+ Rxf8 26.Rxf8+ Bxf8 27.Nf7+ etc) but can try for more OTB.
Well I got the first move but doubt if I would have won it OTB. 24 Qf7+! was a very nice move.
click for larger view
Crafty End Game Trainer check of the puzzle position:
The EGT declines the N with 22...Qg8 which should lose quickly (but White has to be careful - I failed to win first time round). Here's a supplementary Crafty EGT link to the game position after move 22:
Enjoy exploring the variations!
|Aug-20-11|| ||David2009: Geller vs Anikaev, 1979 postscript:
<sevenseasman>: After 22 Nd5 Qb8 try 23.Nxe7+ Kg7 24.Nd5 Qe8 25.Nc7! Qxf7 26.Qxf7+ Rxf7 27.Rxf7+ Kxf7 28.Nxa8 etc.|
A line to avoid in the above:
25.Nxd6 (instead of 25.Nc7) Rxf2 26.Nxe8+ Rxe8 27.Rxf2 exd5 28.Bd4+ Kg8 29.exd5 Re4 30.c3 Nf4 and we reach
click for larger view
and Crafty EGT has enough counter-play to draw.
|Aug-20-11|| ||WhiteRook48: ugh... if I'd seen this before, of course I would've gotten it.|
|Aug-20-11|| ||sevenseaman: <David2009> Going to bed last night, a thought occurred to me. The N, after it has captured e7B (in response to 22...Qb8), must retrace its step back to d5 pronto.|
You have done that; experience counts. Your advice about avoiding the 2nd temptation is also very valuable. Thanks.
Crafty is a resourceful defense tool but like any automated programme, at times it manages to look an idiot that can be exploited.
|Aug-20-11|| ||drnooo: Alekhine won a nice game with a similar theme that came out of a Philidor where he castles queenside and uses the sac forcing the pawns to line up helplessly the same way in fact the whole thing right down to the infamous sting at the end of the combo which is a good five, six moves deep looks like he hard at work, except it ain't or even, yikes, Geller it's some bloody unknown taking his fifteen minutes of fame straight to the Warhol factory with a smile|
|Aug-21-11|| ||CHESSTTCAMPS: White has a knight for a bishop and a strong king-side attack based on control of the f-file. White has a tempting-looking continuation in 22.Bd4 which threaten mate on the move (Nh6) and wins in the case of 22... e5?? 23.Nd5 or 22... Ng7?? 23.Bxg7 Kxg7 24.Nd5! exd5 26.Qd4+. However, the defense 22... Bxg5! breaks white's attack and I couldn't find a way to make 22.Bd4 work.|
I took a fresh look this morning and reexamined an alternative candidate that I'd considered previously. This time I found threats I'd been missing.
22.Nd5!! immediately is the key. The point is to give white uncontested ownership of d4 and a winning attack on the long diagonal
22... exd5 23.Nh6+! and now
A) 23... Kg7 24.Qf7+!! (an idea I'd examined without 22.Nd5 first) Rxf7 25.Rxf7+ Kh1 26.Bd4+ N(B)f6 27.Rxf6! B(N)xf6 28.Bxf6#!
A.1) 27.... Kg7 28.Rf7#
A.2) 27... other 28.Rf8#
B) 23... Kh8 24.Bd4+ Ng7 25.Bxg7+ Kxg7 26.Qd4+ B(R)f6 27.Rxf6 R(B)xf6 28.Qxf6#
B.1) 24... Nf6 25.Qxf6+!! Bxf6 (Rxf6 26.Rxf6 and mate next) 26.Rxf6! Rg8 27.Rf7+ Rg7 28.Bxg7#.
B.2) 24... Bf6 25.gxf6! (Qxf6+?? Rxf6 26.Rxf6 Ng7! 27.Rf7 Rg8 defends!) Qd8 (to prevent 28.f7+ Ng7 29.Qf6) 28.Qf3! de 29.Qxh5! gxh5 30.f7+ Qf6 31.Bxf6#
B.3) 24... Rf6! 25.gxf6 Bf8 26.f7+ Ng7 (Bg7? 27.Bxg7+ Kxg7 28.Qd4+ Kxh6 29.f8=Q) 27.ed and white is an exchange up with a protected pawn on f7 and black very passive.
Time for review. This one was tough - I may skip Sunday.
|Aug-21-11|| ||CHESSTTCAMPS: Nailed the game continuation, but stumbled into a draw against Crafty EGT on first attempt.|
|Aug-23-11|| ||erniecohen: <<Once> There used to be a number of set puzzles that computers couldn't solve - eg where the winning move needed more than a certain number of plies to be evident. I wonder if these still work for monsters like houdini?>|
This happens mostly in cases where the computer just doesn't understand that a position is won, typically because it is an endgame position that's to big to fit into the tablebases. But for fun, I thought I'd look for a game where the combination depends on a long tail, and indeed, it takes Houdini about 11 seconds (from cold start) to find 30. ♗a3 from the famous Botvinnik-Capablanca game. Nevertheless, I'd guess that it took Botvinnik a lot more time than that.
|Aug-23-11|| ||nummerzwei: <erniecohen: it takes Houdini about 11 seconds (from cold start) to find 30. a3 from the famous Botvinnik-Capablanca game.>|
Strange. I've always thought this one was relatively easy for computer programs. The main difficulty is seeing there won't be a perpetual. I find it difficult to believe that a computer could have the idea there might be a perpetual when there isn't one.
Regarding Botvinnik's speed, it is well known that he did not calculate the double sacrifice to a win at all when he made it. He only convinced himself that he had at least a draw in hand.
On the other hand, I'm sure he saw the <idea> of the sacrifice even before the position arose.
|Aug-24-11|| ||erniecohen: We should perhaps be continuing this kibbitz on the page of the game itself, but just to be clear, Houdini also hits upon 30. ♗a3 almost immediately, and thinks that it's the best move for about .1 sec, but decides that it just leads to a draw because of the exposed white king. It then sees that it can get force a trade down to the following position:|
click for larger view
which it evaluates as +.74, and so spends almost all of its time wondering whether this is a winning endgame (again, its lack of endgame knowledge is the Achilles heel). But eventually, it spends a little more time thinking about 3. ♗a3 and realizes white can escape the perpetual, and that this is a leads to a winning advantage.
|Aug-24-11|| ||nummerzwei: <erniecohen>: Actually, I don't see any points that need further clarification. I'm grateful for your explanations though.|
|Jan-10-12|| ||chegado: The f-file is not the way to gradma's home, Anikaev!|
|Mar-13-12|| ||wordfunph: "Who knows how my game with Yuri Anikayev would have gone, had I not been in an especially aggressive mood that evening. I did not just want to win - this is almost always the case. I wanted to sacrifice, I wanted to attack. And the attack succeeded! I will not hide the fact that the finish gave me considerable pleasure and...a special prize."|
- GM Efim Geller
Source: The Application of Chess Theory by Efim Geller
< Earlier Kibitzing · PAGE 3 OF 3 ·