|
< Earlier Kibitzing · PAGE 2 OF 2 ·
Later Kibitzing> |
| Mar-02-20 | | saturn2: 39...Qc1 was quiete a blunder.
Instead of 35...gxf5 I would prefer exf5 |
|
| Mar-02-20 | | goodevans: <saturn2: 39...Qc1 was quiete a blunder.> Black doesn't have anything better. If <39...Qb4> then <40.Qh6 Qf8 41.Rg7>. |
|
| Mar-02-20 | | lentil: I looked at Rg8+ first, but didn't notice that ...Rxg8 hung the Q/c1 (gaaak) |
|
| Mar-02-20 | | malt: 40.Rg8+ K:g8
(40...R:g8 41.Q:c1 )
41.Qg3+ Kf8 42.Qg7+ Ke8 43.Qg8#
<Walter Glattke:> Yesterday ChessGames were offline (for me). Myself as well. |
|
| Mar-02-20 | | zb2cr: I initially thought of 40. Qh6, but as the morning's coffee started to kick in, realized the Black Queen at c1 spoiled that. A second, more careful look, and I found 40. Rg8+. I have nothing to add to the variations worked out by <SamAtoms1980> and <agb2002>. |
|
| Mar-02-20 | | Damenlaeuferbauer: With the last move before time control, my old Georgian friend Zurab found the win of the black queen or the mate in 5 moves with 40.Rg8+,Rxg8 (40.-,Kxg8 41.Qg3+,Qg5 42.Qxg5+,Kf8 [42.-,Kh8 43.Qg7#] 43.Qg7+,Ke8 44.Qg8#) 41.Qxc1 +-. |
|
| Mar-02-20 | | TheaN: After <40.Rg8+> Black has the unpleasant choice of 40....Kg8 41.Qg3 Qg5 42.Qxg5+ Kf8 (Kh8 43.Qg7#) 43.Qg7+ Ke8 44.Qg8#, or <40....Rxg8 41.Qxc1 +->. The ending after Rxg8 is not entirely trivial, though White should be easily winning. The combination of controlling the c-file, the pressure on the g-file, the potential to open up the diagonal to h7, the outpost on b5 and the weak b6-pawn is enough to keep the rooks occupied one way or another; the bishop eventually wins White the game. |
|
Mar-02-20
 | | chrisowen: Lamb of god! |
|
Mar-02-20
 | | gawain: 40 Rg8+ deflects the Black rook and wins his queen. (Or else leads to checkmate.) I liked this puzzle. |
|
Mar-02-20
| | chrisowen: No idea! |
|
| Mar-02-20 | | landshark: Another W plan to win after winning the Q is to pile up on e6 which seems to be impossible to defend without losing the b pawn. Then sac the B for the e and f pawns and win with the connected passers on e5 and f6 |
|
| Mar-02-20 | | TheaN: The win after 41.Qxc1: Black cannot effectively stop 42.Qc6 so he has to be prepared to hold off the queen alone. 41....Rdd8 fails because of 43.Qc7 and a pawn's already gone. So <41....Rb7 42.Qc6 Rbb8>. I'd say Rgb8 is worse, as this confines the rooks completely: after 43.Qd6 White simply prepares 45.Ba6 and Black can no longer coordinate rooks. <43.Qc7 Rgf8> or already give up b6, but 43....Rbf8 45.Qxb6 is lost <44.h6> mind that Black hasn't had time for h6 himself or it would have been tricky to confine the king.
click for larger view
Now Black can no longer hold b6 as after 44....Kg8 45.Bh5 Black has no practical defense against Bxf7+: 45....b5 goes too slow and White wins after 46.Bxf7+ Kh8 (Rxf7 47.Qxb8+ +-) 47.Bxe6 and Qg7# is only preventable by two more moves. So, <44....Rbc8 45.Qxb6 Rc2 46.Bf3 Rfc8> Ra8 fails on 47.Qb7 Rf8 48.Qe7 Rcc8 (Kg8 is similar) 49.Bh5 Kg8 (Rce8 50.Bxf7! wins practically 50....Rxe7 51.fxe7 Rb8 52.Kg3) 50.Bxf7+! Rxf7 51.Qxe6 Rf8 52.Qxd5 wins all center pawns for the bishop. But after <47.Qxa5>:
click for larger view
Black has now accomplished to double rooks, but being the crucial a-pawn down is enough for the White win. |
|
| Mar-02-20 | | SugarDom: This one-star difficulty took me 4 minutes to solve. Darn. |
|
Mar-02-20
| | chrisowen: bubble eg! Hmmm daag? |
|
Mar-02-20
| | chrisowen: Gains hard? |
|
| Mar-02-20 | | TheaN: Perhaps I should not have omitted <47....R2c4 48.Bh5!> the bishop's no longer needed to defend g2; White prepares the fatal blow. 48....Kg8 49.Qd2 whoops. 48....Rf8 fails on 49.Bxf7 as f7 isn't actually defended. After 48....Rxd4 49.Qa7! proves White's won enough pawns after 49....Rh4+ 50.Kg1! (don't allow a check on the g-file) 50....Rxh5 51.Qxf7 Rg5 52.Qxe6 Rc1+ 53.Kf2 Rc2+ 54.Ke1! Rcxg2 55.Qe8+ Rg8 56.f7 and White simply walks out of checks. So <48....R4c7> but now <49.Bxf7! Rxf7 50.Qb6> and Black has no proper defenses against both the a-pawn and Qxe6. |
|
| Mar-02-20 | | Momentum Man: Inexplicable blunder by Black, who was an IM |
|
Mar-02-20
 | | AylerKupp: <Phony Benoni> I swear on a stack of Stockfishes, Houdinis, Komodos, etc. that I didn't even consider 40.Qh6 at any time. Given that it's a Monday puzzle I knew that it needed to be a forcing move. And the mate by Qg7 seemed like the most likely outcome. So how does White <forcibly> get its queen to g7? By giving check on g4, of course. But, given that Black's king is not on the g-file, how does White force Black's king to the g-file? By 40.Rg8+ of course. Too bad that loses the rook by 40...Rxg8. But, wait! That leaves Black's Qc1 undefended! So after 40...Rxg8, 41.Qxc1. And for once this thought process was almost instantaneous for me. Pattern recognition is a wonderful thing – when it works. But, after all, it's a Monday puzzle, it's supposed to be easy. And for me it was – this time. Tomorrow, unfortunately, is not a Monday. But so far I'm 1/1 on the way to 1/7! |
|
Mar-02-20
| | AylerKupp: <A stackfull of engines> (part 1 of 2) But <Phony Benoni>'s comment about a stack of Stockfishes, etc. made me think, always a bad thing. And it would be utterly unlike me to let it go after a reasonably short post. How would a stackfull of chess engines approach the situation after 39...Qc1?
click for larger view<Spoiler alert>: All these engines found, or thought they found, a forced mate-in-8. But there were several different approaches and most of them are interesting and possibly funny. And two of the engines didn't even calculate the number of moves needed for a mate in their Principal Variation properly. So here's what <some> of my stackfull of chess engines determined, all within 30 secs or less: <Stockfish 11>: 40.Rg8+ Rxg8 41.Qxc1 Rc7 (a rather drastic way to discourage 42.Qh6! But several engines used this approach so the horizon effect must be at work) 42.Qxc7 Rf8 43.Bd3 (but after this preparatory move Qh6 cannot be prevented) b5 44.Qc1 Rg8 45.Qh6 Rg6 46.hxg6 fxg6 47.Qg7#
click for larger view<Houdini 6>: 40.Rg8+ Rxg8 41.Qxc1 b5 (who's afraid of the big, bad Qh6?) 42.Bd3 (but White is in no hurry since mate can't be prevented) 42...Rdd8 43.Qh6 Rxg2+ (the horizon effect at work again) 44.Kxg2 Rg8+ 45.Kh2 Rg6 46.hxg6 fxg6 47.Qg7#
click for larger view<Komodo 12.3-ABP>: 40.Rg8+ Rxg8 41.Qxc1 Rc7 42.Qxc7 Rf8 43.Qe7 (Who needs Qh6? I can almost hear Komodo saying "Qh6? I don't have to play no stinking Qh6!) 43...Kg8 44.Bd3 h6 (after all, mate is unavoidable and allowing 45.Bxh7+ would reduce Black's evaluation of the position since it would have one less pawn) 45.Bh7+ Kxh7 46.Qxf8 b5 47.Qg7#.
click for larger view<ShashChess 8.0>: 40.Rg8+ Rxg8 41.Qxc1 Rc7 42.Qxc7 Rg7 (an equally drastic way to delay Qh6!) 43.fxg7+ Kxg7 44.h6+ Kg6 45.Qd8 Kf5 46.Qh4 b5 47.Qg4#
click for larger view.
<Rybka 4.1>: 40.Rg8+ Rxg8 41.Qxc1 b5 (Rybka also doesn't bother to try to delay Qh6 because if 42.Qh6 then 42...Rdd8 43.Bd3 (mate is now unavoidable but it can still be delayed by an orgy of sacrifices) 43...Rxg2+ (more horizon effect-driven moves) 44.Kxg2 Rg8+ 45.Kf2 Rg2+ 46.Kxg2 Kg8 47.Qg7# ) 42.Bd3 Rdd8 43.Qh6 (but this is just a transposition of moves) 43...Rxg2+ 44.Kxg2 Rg8+ 45.Kf2 Rg2+ 46.Kxg2 bxa4 47.Qxh7#
click for larger viewApparently Rybka found 47.Qxh7# earlier in its search tree than 47.Qg7#. Besides, it won a pawn so its position evaluation was probably higher. :-) <Bouquet 1.8>: 40.Rg8+ Rxg8 41.Qxc1 Rdd8 42.Qh6 Rxg2+ 43.Kxg2 Rg8+ 44.Kf2 b5 45.Bd3 Rg6 46.hxg6 fxg6 47.Qg7#
click for larger view |
|
Mar-02-20
| | AylerKupp: <A stackfull of engines> (part 2 of 2) <Critter_1.6a>: 40.Rg8+ Rxg8 41.Qxc1 Rc7 42.Qxc7 Rf8 43.Bd3 b5 44.Qe7 Kg8 (with this move order Critter can't afford to save a pawn by 44...h6) 45.Bxh7+ Kxh7 46.Qxf8 bxa4 47.Qxf7+
click for larger viewAnd here Critter stops displaying moves even though it announced +M8 after 39...Qc1 but from this position an extra move is needed, either 47...Kh6 48.Qg6# or 47...Kh8 48.Qg7#. Apparently Critter can't count, but maybe it saw 47.Qg7# and simply "touched" the Pf7 and was forced to capture it. :-) But who needs to count properly when you can force a mate? <Spike1.4>: 40.Rg8+ Rxg8 41.Qxc1 Rc7 42.Qxc7 Rf8 43.Qe7 (again, who needs qh6?) 43...Kg8 44.Bd3 Ra8 (yet another way to try to delay the unavoidable result) 45.Bxh7+ (anyway!) 45...Kxh7 (if 45...Kh8 then 46.Qxf7 Rg7 47.Qxg8#) 46.Qxf7+ Kh8 47.Qg7#. And, of course, if 46...Kh6 then 47.Qg6#.
click for larger view<Hannibal 1.4>: 40.Rg8+ Rxg8 41.Qxc1 Rc7 42.Qxc7 Rg7 (yet another way to delay the mate!) 43.fxg7+ Kxg7 44.h6+ Kg6 45.Qd8 Kf5 46.Qh4 Kg6 47.Qh5+.
click for larger viewHannibal, like Critter, announced +M8 after 39...Qc1 but in this position an extra move is also needed after 47...Ke4 48.Qf3# But wait! There's more! All the engines above are "classic" chess engines using a hand-crafted evaluation function and minimax search combined with alpha-beta pruning (ABP to find the best move. With the success of AlphaZero, LeelaC0, and other neural-network based engines that don't do a minimax search perhaps they can do better. I can't run LeelaC0 because I still only have a 32-bit computer (some day!), but Komodo offers a version that uses Monte Carlo Tree Search (MCTS) instead of minimax search with ABP. How did this version do in comparison with Komodo 12.3-ABP? Not too well. After 40.Rg8_ Rxg8 41.Qxc1 b5 it got fixated on capturing the Pb5 either by 42.axb5 as its Principal Variation until d=33 after almost 2 hours of calculation. <Komodo-12.3 MCTS>: 40.Rg8+ Rxg8 41.Qxc1 Rdd8 42.Qh6 Rd7 43.Bf1 (apparently Komodo MCTS is anticipating the rook sacrifice(s) on g2 and makes a move to neutralize them) 43...Rdd8 44.Kh1 Rd7 45.Bb5 Rgd8 46.Bxd7 (inexplicable, why not 46.Qg7#?) 46...Rxd7 47.Qf8#
click for larger viewSo Komodo MCTS does find a mate-in-8 like the classic engines did, but it took it a lot longer and not without much help from Black plus a moment of chess blindness. So perhaps MCTS is not a panacea over minimax + ABP pruning. |
|
Mar-02-20
 | | HeMateMe: It must hurt to be on the losing end of this. |
|
| Mar-02-20 | | TheaN: Good posts <AylerKupp> and definitely interesting for that position but perhaps don't omit Pf5. Otherwise my posts look so silly. I guess they do anyway :> |
|
| Mar-02-20 | | RandomVisitor: A very recent development build of Stockfish thinks that black can delay checkmate for a while by aiming for a fortress-type defense:
click for larger viewStockfish_20022713_x64_modern:
<54/78 53:47 +27.21 40.Rg8+ Rxg8 41.Qxc1 Rdd8> 42.Qc7 Rdf8 43.Qxb6 Rc8 44.Qa7 Rgf8 45.h6 f4 46.Qxa5 Rc2 47.Bd1 Rc1 48.Bh5 f3 49.Bxf3 Rfc8 50.Bh5 R1c7 51.Qb4 Rc2 52.a5 Ra2 53.Qb7 Rg8 54.Bf3 Rg6 55.Qc7 Ra1 56.Qa7 Rb1 57.a6 Ra1 58.Qe7 Ra2 59.Qd8+ Rg8 60.Qc7 Rg5 61.a7 Rg6 62.Qb7 Rg8 63.a8Q Rgxa8 64.Qxf7 Rg8 65.Qb7 Rf2 66.Qe7 Rc2 67.Kg1 Rb2 68.Qd7 Rc2 69.Kf1 Rc1+ 70.Kf2 Rh1 71.Qe7 Rb1 72.Ke3 Rb3+ 73.Kf4 Rd3 74.f7 Rxf3+ 75.gxf3 Rg4+ 76.Ke3 |
|
Mar-04-20
| | AylerKupp: <<TheaN>Good posts <AylerKupp> and definitely interesting for that position but perhaps don't omit Pf5.> I'm glad that you thought that they were good posts because my posts seldom are. But I'm not sure what you mean by not omitting Pf5. From my usual narrow minded perspective (focused sounds better but a rose by any other name ...) I was only looking at the position after 39...Qc1.
click for larger viewAll other moves may be interesting (or even better) but as the saying goes for those that don't want to do additional work, that's beyond the scope of my analysis. :-) And don't bother about your posts looking silly. Most of my post do and I think that I post more than you do, so I'm the sillier one. |
|
Mar-05-20
 | | moronovich: Hi <AK> !
In the position above white has a direct win with Rg8+.But you perhaps know already. |
|
 |
|
< Earlier Kibitzing · PAGE 2 OF 2 ·
Later Kibitzing> |
Kibitzing
sac: 26.f5