|
< Earlier Kibitzing · PAGE 5 OF 5 ·
Later Kibitzing> |
Jan-30-11
 | | Once: Ah, you want me to put the bunny back in the box, eh? Apparently I've got an Athlon II X4 630 2.80 ghz with 4GB of RAM. Whatever that means... |
|
| Jan-30-11 | | RandomVisitor: Does white have an earlier way to win? After 23.Bxh7! Rxh7 24.dxc6 black has poor survival chances:
click for larger view Rybka 4:
<[+1.66] d=23 24...bxc6> 25.Na2 Qb5 26.Qf3 g6 27.Rb3 Qa4 28.Nb4 Rb8 29.Qh3 Rhh8 30.Qd7 Rb7 31.Qxc6 Qxc6 32.Nxc6 Rxb3 33.cxb3 Bd8 34.Rd1 Bc7 35.Ne7 Kb7 36.b4 Rf8 37.Nd5 Bd8 38.Rd3 Kc6 39.Kc2 Bb6 |
|
| Jan-30-11 | | johnlspouge: < <Once> wrote: [snip] Apparently I've got an Athlon II X4 630 2.80 ghz with 4GB of RAM. Whatever that means... >
Yours is a little faster than mine (2.80 GHz vs. 2.4 GHz, a "Hertz" being a cycle per second), so Fritz and Toga performed comparably. Yours is also about as big as mine, a statement <JG27Pyth> might call a "BPB". I would explain that acronym, too, but the explanation would probably be censored (again)... Hasta la vista, baby :) |
|
Jan-30-11
 | | scormus: I was OK until 27... Bd8, for some reason I imagined ... Qd7 would be played instead. Then 28 Rc3 and I suppose the only way to parry the threat of Rc7 is 28 ... Bd8 and the BQ is in the wrong place. Yes, a wonderful combination, brilliantly topped off with 31 Rdc3! I admit I'd never have forseen that. Sigh. Seems not only can I not solve these puzzles, I'm not getting the puns either, I thought the Wizard of Oz was Magnus. |
|
| Jan-30-11 | | johnlspouge: < <scormus> wrote: [snip] I'm not getting the puns either, I thought the Wizard of Oz was Magnus. > The pun puzzled me, too: "Oz" = "Auz" = "Australia", where Ian Rogers hails from. |
|
Jan-30-11
 | | patzer2: For today's insanely difficult Sunday puzzle, Ian Rogers 26. Bxc6+!! sacrifices a piece to destroy pawn cover and expose the weakly protected King and the unprotected Bishop and Queen to a decivise attack. As <Once> observes it is extremely difficult to calculate all the variations, but the positional advantages scream for the sacrifice: 1. Weakly protected King position
2. Unprotected Queen and Bishop
3. Bad White Squared Bishop completely out of play.
4. White's pieces are all in play and potentially active. 5. 26. Bxc6!! Qxc6 27. Nd5! makes a threat and clears the way for all White's pieces to quickly mobilize and join in the attack. Even so, preceisely calculating all the possibilities is extremely difficult. I wonder if Rogers calculated all of this out in advance or simply played the sacrifice on general principles and figured it out as he went. P.S.: White's winning position came about because of the miscue 25...Qc7?! Instead, 25...Rb8 stops the threat of 26. Bxc6 and fully equalizes. |
|
| Jan-30-11 | | Patriot: <David2009> <There is a major hole in my analysis: after 27...Bd8 28 Nb4, 28...Qb5! pins the N.> I didn't consider 28...Qb5 either. In your variation A, white may have the better ending but I would rather try to retain my active pieces with 30.Qa6 instead of 30.Qd3. For instance, 30...Rxd6 31.Rxd6 Rd8 32.Qc6+ Kb8 33.Qb5+ and go from there. 33...Ka8? 34.Qd5+ wins. 33...Kc8 34.Qc5+ Kb7/Kb8 (34...Qc7 35.Rc6  ) 35.Rb6+ wins the queen. I didn't see all of that at first but was thinking more as a general principle, when you have an attack, try not to exchange all of the attackers (especially the queen). |
|
| Jan-30-11 | | Patriot: <patzer2> <I wonder if Rogers calculated all of this out in advance or simply played the sacrifice on general principles and figured it out as he went.> Good question and we may never know the answer. I see now that I did not assess the position correctly--the bishop on d7 is NOT trapped! This takes the problem in a different direction. If it were trapped then Bxc6+ must be played. Then I would wonder how the bishop became trapped and if white calculated everything from previous moves, which would be even more impressive! I think he may have calculated it out because he still had Bh3 as an option. But I don't think it's completely necessary because the h7-bishop may as well not be on the board. |
|
| Jan-30-11 | | CHESSTTCAMPS: This middlegame position looks somewhat more comical than insane. White is effectively a minor piece ahead, given the ugly position of the black LSB on h7, which has apparently buried itself with the maneuver Bf5-Bh7-g6. With both rooks and both bishops languishiing on the kingside with low mobility, black is ill-prepared to cope with a queenside invasion, especially considering the difficulty of extricating the buried bishop. The following sacrifice makes it easier to invade on the light squares: 26.Bxc6+ Qxc6 (Kb8? 27.Nd5 wins quickly) 27.Nd5
The knight on d5 is already stronger than the black bishops combined. Furthermore, white gains tempi for mobilizing the majors by attacking the Q and DSB. A) 27... Bd8 28.Rc3 Qb7 29.Rb3 Bb6 30.Nxb6 axb6 31.Rxd6 Ka7 32.Qd3 and the threat of Rd7 wins. A.1) 28... Qd7 29.Qa6 Be7 (otherwise 30.Rc8+) 30.Rc7 wins A.2) 28... Qa4 30.Rc4 Qa5 31.Rb4 f5 32.Qc4 Qc5 (Bb6 33.Qc6+ Kb8 34.Rxb6 axb6 35.Qc7+ Ka8 36.Nxb6+ wins) Qa6 and black can resign. I present the preceding lines as strong evidence that black can't defend the a8 diagonal if white continues correctly. Because I have already exceeded my time limit, I will not analyze the defenses B) 27... Re8 C) 27... Qb7 and D) 27...Qd7. Time to test against Crafty EGT and view the game. |
|
| Jan-30-11 | | CHESSTTCAMPS: My analysis was OK as far as it went, but I lost to Crafty on the first try. Later I will review to determine where I missed my chance. |
|
| Jan-30-11 | | CHESSTTCAMPS: <Phony Benoni> <First of all, resist the temptation to say that Black's a piece down. Once he plays ...f6, that LSB will have something to say about the rest of the game....> This is a fair point and makes the case for careful analysis. I would add ...f5 to the comment. |
|
Jan-30-11
 | | agb2002: White has a bishop and a knight for the bishop pair. Black threatens 26... Qxd7 and is probably ready for ... f6 followed by doubling rook on the f-file and ... Bg8, recovering the lsb. White would like to pass the rooks to the queen side to attack the black king. However, the knight blocks the way. Therefore, 26.Bxf6+, freeing d5 for the knight and further weakening Black's castle: A) 26... Qxc6 27.Nd5
A.1) 27... Bd8 28.Rc3
A.1.a) 28... Qb7 29.Rb3
A.1.a.i) 29... Bb6 30.Nxb6 axb6 31.Rxd6 and the b-pawn is lost with a winning attack. A.1.a.ii) 29... Qc6 30.Rdd3 Ba5 (30... Qa6 31.Rdc3 Qxe2 32.Rc8#) 31.Ne7 Qd7 32.Rd5 followed by Qa6 and White's attack seems to make progress. A.1.b) 28... Qd7 29.Qa6 Kb8 (29... Be7 30.Nb6+ Kb8 31.Nxd7 + -) 30.Rb3+ Ka8 (30... Bb6 31.Nxb6 axb6 32.Rxb6+ + -) 31.Rdd3 and the threat Rdc3-Rc8+ looks difficult to stop. A.1.c) 28... Qa4 29.Rc8+ Kb7 30.Rc4 followed by Rd3 and Rb3 and White seems to have compensation for the material. A.2) 27... Re8 28.Rc3 Qa4 29.Rc4 Qa5 30.Nc7+ Kb7(8) 31.Nxe8 Rxe8 32.Rb4+ Kc7(8) (32... Ka8 33.Qc4 with the threats Qc6+ and Qxf7) 33.Qc4+ Kd8 34.Rb8+ Kd7 35.Rb7+ Kd8 36.Qc6 + -. A.3) 27... Qb7 28.Rb3 Qd7 29.Qa6 is similar to A.1.b. A.4) 27... Qd7 28.Nxe7 (28.Rb3 Rb8 29.Qa6 Rxb3 30.cxb3 Rb8 31.Nxe7 Qxe7 32.Rxd6 and perhaps White doesn't have enough compensation) Qxe7 29.Rxd6 Rd8 30.Qa6 is unclear. B) 26... Kg8 27.Nd5
B.1) 27... Qxc6 28.Nxe7 Qd7 29.Rxd6 Qxe7 30.Rd7 Qe6 31.R1d6 followed by Qd5+ winning. B.2) 27... Qd8 28.Rb3+ Kc8 29.Qb5 + -.
I don't know, but I probably would try 26.Bxf6+. |
|
| Jan-30-11 | | RandomVisitor: <patzer2><25...Rb8 stops the threat of 26. Bxc6 and fully equalizes.>Possibly, but deep computer runs suggest a rough ride ahead:
click for larger view
Rybka 4
<[+0.87] d=24 26.b3> Qc7 27.Bh3 Rhd8 28.f4 exf4 29.e5 d5 30.Bg2 (1:52:43) 1685908kN |
|
| Jan-30-11 | | WhiteRook48: I suck at sunday puzzles |
|
| Jan-30-11 | | CHESSTTCAMPS: As <David2009> mentioned earlier, Crafty EGT deviates from the game with 30... Bb6, giving the following position:
click for larger viewI lost twice from this position before I took the time necessary to find a problem-like combination. I will follow David's lead and not give the solution away until folks have had a chance to take a crack at it. |
|
| Jan-30-11 | | M.Hassan: "Insane" White to play 26.?
Materials even.
A sane move would be 26.Bh3. an insane move is:
26.Bxc6+ Qxc6
27.Nd5 Bd8
28.Rc3 Qd7
29.Qa6 Be7
30.Rc7 Rxc7(or else 31.Qxa7#)
31.Nxc7+ Kb8
32.Nb5
It may be the time for Black to resign.Time to check to learn more. |
|
| Jan-30-11 | | CHESSTTCAMPS: Regarding the problem diagammed in my previous post, I posted a solution to my forum. |
|
| Jan-31-11 | | TheBish: I Rogers vs G Milos, 1992 White to play (26.?) "Insane"
Black's Bh7 looks to be buried for God knows how long, so I think White can sac his bishop to allow a prime attacking square for the knight. 26. Bxc6+!? (Maybe !!, don't know yet.)
26...Qxc6 27. Nd5 Bd8 28. Rc3 Qb7
This stops 29. Qa6. Looks like we need to get the other rook into play. 29. Rb3 Qc6 30. Rdd3 Ba5
Now we need to somehow remove the bishop's protection of c7, as well the queen's. 31. Rdc3!! Bxc3 32. Qa6!
Maybe two exclams for the queen sac, but you had to see the rook sac first as well as the follow-up, and the queen sac is rather easy after the first. 32...Rc8
Of course 32...Qxa6 33. Nc7# is easy, as is 32...Qd7 33. Nb6+. 33. Nb6+ Qxb6
Or 33...Kb8 34. Nxc8+ Kc7 35. Rxc3 is hopeless.
34. Rxb6 Rb8 35. Rxb8+ Rxb8 36. Qc6+ Rb7 37. Qxc3 and White has an easy win in the endgame. |
|
Jan-31-11
| | scormus: <johnlspouge: "Oz" = "Auz" = "Australia", where Ian Rogers hails from.> Thanks! Of course, they play a form of chess there, under Aussie Rules ;) |
|
| Jan-31-11 | | gofer: I missed this weekend's puzzles as I was on duty at my folks. Late last night I saw this and got as far as
29 Rb3 ... but thought black would reply 29 ... Bb6 accepting the long term loss of the bishop as the a6 pawn
can creep up on the bishop and the a7 pawn cannot move. <26 Bxc6+ Qxc6 27 Nd5 Bd8 28 Rc3 Qb7 29 Rb3 Bb6 > But what I was really interested in was how on earth Bh7 had managed to get to h7 in the first place. Now this would
have been an interesting puzzle in its own right - it stumped me! So I looked at the game and it was obvious!
Black used 7 moves out of 23 (nearly a third) to create a inpenetrable fortress for its LSB that would take
black another 4 moves to get it out of!!! Surely this can't be deemed as "practical" - to "wall-in" one of your minor
pieces? |
|
Jan-31-11
 | | AylerKupp: <<johnlspouge:> This problem provides an opportunity to test the puzzle-solving ability of different chess engines. My timing tests used an Intel 4GB RAM 2.40 GHz dual processor.> I thought that you might be interested on my results (time, depth, PV, etc.) for finding 26.Bxc6+ on the chessgames.com puzzle, I Rogers vs G Milos, 1992, Jan-29-11:using an Intel Q9400 4-core, 2.66 GHz processor system with 4GB RAM and a 1024 KB hash table. All the engines initially considered 26.Bh3 to be the best move.
click for larger viewGame continuation: 26.Bxc6+ Qxc6 27.Nd5 Bd8 28.Rc3 Qb7 29.Rb3 Qc6 30.Rdd3 Ba5 31.Rdc3 Bxc3 32.Qa6 1-0. Houdini 1.5: 1.564 mins, d=20, eval=[+4.62], 478,066,432 total nodes evaluated at an average rate of 4,900 Knodes/sec. PV=26.Bxc6+ Qxc6 27.Nd5 Qc4 28.Nxe7 Rb8 29.b3 Qa6 30.Qd2 Qb7 31.Nd5 f5 32.gxf6 Bg8 33.f3 Rf8 34.Nb4 Rxf6 35.Rxd6 Rxd6 36.Qxd6 Bf7 37.Qxe5 Rb8 38.Rd6 Rc8 39.c4 Re8 40.Qf6. Rybka 4: 0.454 mins, d=15, eval=[+2.77], 4,053,506 total nodes evaluated at an average rate of 152 Knodes/sec. PV=26.Bxc6+ Qxc6 27.Nd5 Qc4 28.Nxe7 Rb8 29.Qf3 Rb7 30.Rc3 Qe6 31.Nd5 Rc8 32.Rxc8+ Qxc8 33.Qc3 Qxc3 34.Nxc3 Rd7 35.Nb5 Kb8 36.Rxd6 Kc8 37.Rxd7 Kxd7 38.Nxa7 Ke6 39.a4 f6 40.gxf6 Kxf6 41.a5 Stockfish 2.0.1: 0.784 mins, d=25, eval=[+3.87], 131,001,437 total nodes evaluated at an average rate of 2,799 Knodes/sec. PV=26.Bxc6+ Qxc6 27.Nd5 Bd8 28.Rc3 Qb7 29.Rb3 Qc6 30.Rdd3 Bb6 31.Nxb6+ axb6 32.Rxd6 Qxd6 33.Qa6+ Kb8 34.Rxb6+ Qxb6 35.Qxb6+ Kc8 36.Qa6+ Kc7 37.Qa7+ Kc8 38.b4 Rd8 39.b5 Rhe8 40.b6 Rd1+ 41.Kb2 Toga II 1.3.4 (DM3 Compile): 10.832 mins, d=19, eval=[+2.53], 695,426,113 total nodes evaluated at an average rate of 1,071 Knodes/sec. PV=26.Bxc6+ Qxc6 27.Nd5 Qc4 28.Nxe7 Rb8 29.Qd2 Rb7 30.Rc3 Qd4 31.Nc8 Rd7 32.Qxd4 exd4 33.Rxd4 Rhd8 34.Rd5 Kb7 35.Rb5+ Ka8 36.f3 d5 37.e5 Toga II 1.4.5c (beta): 11.420 mins, d=22, eval=[+2.82], 678,880,142 total nodes evaluated at an average rate of 3,884 Knodes/sec, PV=26.Bxc6+ Qxc6 27.Nd5 Qc4 28.Nxe7 Rb8 29.Qd2 Rb7 30.Rc3 Qd4 31.Nc8 Rd7 32.Qxd4 exd4 33.Rxd4 Rhd8 34.Rd5 Kb7 35.Rb5+ Ka8 36.f3 d5 37.e5 This exercise made me curious, so I’ve downloaded and will be running other free engines that have been highly ranked (Critter, Komodo, Gull). I have given up on Crafty 23.4 since it only found 26.Bxc6+ after more than 7 hours (!) during its preliminary evaluation at d=32 (and it’s still not done with it’s full evaluation at d=32!). No wonder it’s not highly ranked but, of course, this is only one instance. |
|
Jan-31-11
| | AylerKupp: <<johnlspouge:> This problem provides an opportunity to test the puzzle-solving ability of different chess engines. My timing tests used an Intel 4GB RAM 2.40 GHz dual processor.> Here are additional results from other engines highly ranked by IPON (http://www.inwoba.de/ ) as of Jan-31-11. Some of the engines gave initially impressive results, particularly Critter and Protector, in how quickly they discovered 26.Bxc6. But then they want back to 26.Bh3 and only after some time returned to 26.Bxc6. So I listed the performance statistics not following the initial discovery of 26.Bxc6 (Protector 1.4 selected 26.Bxc6 as early as the first move at d=1!) but the first suggestion of 26.Bxc6 after the engine had settled on it as the best next move by white along with a positive evaluation for white (Protector 1.4 didn’t give 26.Bxc6 a positive eval until d=23). I was leery of Gull’s low eval=[+0.33] for white compared to all the other engines’ evals for white (all above [+2.00])and thought about waiting for the eval to rise but it didn’t do that in spite of waiting for several plies. Maybe Gull is just pessimistic. Critter 0.90: 0.556 mins, d=19, eval=[+2.59], 101,901,503 total nodes evaluated at an average rate of 3,143 Knodes/sec. PV=26.Bxc6+ Qxc6 27.Nd5 Qc4 28.Nxe7 Rb8 29.Nd5 Rhc8 30.Re3 Qxe2 31.Rxe2 Rc6 32.Nf6 Bg8 33.Red2 Rcb6 34.Nxg8 Rxg8 35.Rxd6 Rxd6 36.Rxd6 Re8 37.Rd7 Rf8 38.Kc1 f6 Gull 1.1: 6.304 mins, d=20, eval=[+0.33], 699,896,267 total nodes evaluated at an average rate of 1,851 Knodes/sec. PV=26.Bxc6+ Qxc6 27.Nd5 Bd8 28.Rc3 Qb7 29.Rb3 Qc6 30.Rdd3 Ba5 31.Rdc3 Bxc3 32.Qa6 Rc8 33.Nb6+ Qxb6 34.Rxb6 Hannibal 1.0a: 2.015 mins, d=21, eval=[+2.33], 97,293,129 total nodes evaluated at an average rate of 8,0481 Knodes/sec. PV=26.Bxc6+ Qxc6 27.Nd5 Qc4 28.Nxe7 Rb8 29.Nd5 Rhc8 30.Nf6 Bg8 31.b3 Rb6 32.Qd2 Kb7 33.Rxd6 Rxd6 34.bxc4 Rxd2 35.Rxd2 Kc6 36.Kb2 Rb8+ 37.Kc3 a6 Komodo 1.3: 49.264 mins, d=27, eval=[+3.88], 1,982,964,128 total nodes evaluated at an average rate of 6,706 Knodes/sec. PV=26.Bxc6+ Qxc6 27.Nd5 Qc4 28.Nxe7 Rb8 29.Qf3 Rhe8 30.Nd5 Rec8 31.Rc3 Qb5 32.b3 Rxc3 33.Qxc3 Qc5 34.Qxc5 dxc5 35.Nf6 Bg8 36.Nd7 Rc8 37.Nxe5 Kb7 38.Kb2 Rc7 39.Nc4 Kb8 40.Kc3 Rc8 41.Ne5 a6 42.Nc4 Kc7 43.Rd5 Kc6 44.Rd6+ Kb7 45.Rd7+ Kc6 Protector 1.4: 34.510 mins, d=23, eval=[+3.84], 1,533,879,139 total nodes evaluated at an average rate of 651 Knodes/sec. PV=26.Bxc6+ Qxc6 27.Nd5 Qc4 28.Nxe7 Rb8 29.Qf3 Rhe8 I’ll summarize the results of all engines in my next post. |
|
Jan-31-11
| | AylerKupp: Best Chess Engine. A question that frequently comes up is “Which chess engine is best ?”. The first problem is defining what “best” is. The engine that is “best” for blitz may not be “best” for OTB play and may not be “best” for correspondence play. The most common method of trying to determine which engine is “best” is to conduct engine-engine tournaments at various time limits, and rank the engines according to their standing at the end of the tournament. My interest is to determine which engine(s) are best for analysis and correspondence chess at long time controls, and I don’t know that the engines that rank the highest in engine-engine tournaments are necessarily the best engines to use for analysis or CC. Inspired by <johnlspouge:>, I ran some tests on chessgames.com’s puzzle on Sunday, Jan-30-11, the game I Rogers vs G Milos, 1992. I think (but that’s to be proven) that ranking engine performance in solving puzzles using Infinite Analysis mode is a more accurate method of determining the “best” engine for analysis and CC than using the results of engine-engine tournaments at much faster time controls. What are the minimum requirements and considerations for ranking engine’s strength in analysis/CC? Here’s my criteria and I would love to get some feedback: Best Engine: The engine that provides the correct move in the shortest possible time with a reasonable evaluation of the position. Of secondary interest is which engine most closely followed the actual game continuation, but this is more an evaluation of the player than the engine since the player may not have found the best continuation! Of course, without providing the correct move to solve a puzzle nothing else is important. But since we typically want to run multiple analyses speed is also important, all other things being equal. And consistent reasonable evaluations of positions inspires confidence that other analyses will be correct more often than not. I analyzed the I Rogers vs G Milos, 1992 position following black’s 25th move to see how well several engines (all but Rybka free, I’m not made of money) would do in coming up with white’s 26.Bxc6. The analyses were done using an Intel Q9400 4-core, 2.66 GHz 32-bit processor system with 4GB RAM and a 1024 KB hash table under Windows XP. All the engines but one (Protector 1.4) initially considered 26.Bh3 to be the best move. Below are the results, with the engine’s ranking according to the latest IPON ((http://www.inwoba.de/ ) tournaments in parenthesis. I listed the performance rank not following the initial discovery of 26.Bxc6 (Protector 1.4 selected 26.Bxc6 as early as the first move at d=1!) but the first suggestion of 26.Bxc6 after the engine had settled on it as the best next move by white along with a positive evaluation for white (Protector 1.4 didn’t give 26.Bxc6 a positive eval until d=23) One note regarding the IPON rankings: I used the highest ranking for each engine, and this would usually correspond to the 64-bit version of that engine, not the 32-bit version. Sorry that the “table” looks so ugly, but there aren’t many formatting options. Rank Engine Eval Time (mins)
1 (2) Rybka_4 [+2.77] 0.454
2 (7) Critter_0.90 [+2.59] 0.556
3 (4) Stockfish_2.0.1 [+3.87] 0.784
4 (1) Houdini_1.5a [+4.62] 1.564
5 (45) Hannibal_1.0a [+2.33] 2.015
6 (17) Gull_1.1 [+0.33] 6.304
7 (62) Umko_1.1 [+2.62] 10.158
8 (N/A) Toga II_1.34 [+2.53] 10.832
9 (53) Toga II_1.4.5c (beta) [+2.82] 11.420
10 (28) Protector_1.4 [+3.84] 34.510
11 (11) Komodo_1.3 [+3.88] 49.264
12 (74) Crafty_23.4 [+3.90] 421.918
Summary and Observations:
a. Rybka 4 was the fastest in finding 26.Bxc6.
b. Gull 1.1 had a suspiciously low evaluation of white’s position ([+0.33]) compared to the other engines, but this evaluation didn’t rise substantially at greater depths. Gull 1.1 was also the engine that most closely followed the actual game, up until black’s move 32. c. Crafty was barely able to find the proper move and I gave up before it output a PV. |
|
Jul-12-16
 | | perfidious: <Jonathan Sarfati: Rogers should have won the prize. It is totally corrupt to have a contestant for the prize be one of the judges, as Kasparov was. Any fair judge would have recused himself.> Until reading a post elsewhere by <saffuna> today, I had forgotten there was any sort of controversy over this Olympiad--not that there should have been in a just world, as <Jonathan>'s remark above says it all. |
|
| May-06-17 | | Saniyat24: absolutely mind-blowing sequence of moves....just wow..!! |
|
 |
|
|
< Earlier Kibitzing · PAGE 5 OF 5 ·
Later Kibitzing> |
|
|
|
Kibitzing
sac: 26.Bxc6+