< Earlier Kibitzing · PAGE 2 OF 2 ·
Later Kibitzing> 
Mar0420   Walter Glattke: Not so easy after 31.Kd2 Qxb2+ 32.Ke1 Re3+ (Tartakower: No game was won by
resign) 33.Qb5+? 34.Bd3 Rxd3? perpetual by Qh5Qd8 or B) above with 33.Qb3, possibly 34.Rxd5!? Qxb1+ 35.Kg2 cxd5 36.Qxd5+ Kh8 37.Qh5+ Qh7! wins for black. 

Mar0420   stacase: 32...Qxa3+ and White throws in the towel. I am reminded of one of those little sayings when you post that says, "No one ever won a game by resigning. Over the board, I'd be very happy to accept White's resignation after 32...Qxa3+ 

Mar0420   stacase: And guess what little saying popped up when I posted that? Ha ha ha ha ha ha! 

Mar0420   saturn2: No perpetual after <Walter Glattke's 30...Rxc3+ 31. Kd2 Qxb2+ 32. Ke1 Re3+ 33. Kf1 Qb
5+ 34. Bd3> Now black still has 34... Qb3 35. Be2 Nc3 36. Re1 Qe6
and wins. 

Mar0420   saturn2: or instead
35. Qh7+ Kf8 36. Qh8+ Kf7 37. Qh5+ Ke6 38. Bf5+ Kd6 and no more checks. 

Mar0420
  thegoodanarchist: Tough one today. 

Mar0420   BwanaVa: Stacase...after 32...Qa3+ its mate in one, so why not resign? 

Mar0420
  chrisowen: Mitigate. 

Mar0420
  chrisowen: Just know that... 

Mar0420
  drollere: i also solved the puzzle as 30. .. Rxc3+, 31. Kd2 because pxc3 Qxc3+ is merely suicidal. then 31. .. Qxb2+, 32. Ke1 Re3+, 33. Kf1 are forced. now the black R prevents the perpetual check on e8h5, and it seems the white K can escape checks starting at Qh7+. i would probably just play 34. Qxa3 

Mar0420   stacase: <BwanaVa: Stacase...after 32...Qa3+ its mate in one, so why not resign?> Because I'm a wood pushing patzer and didn't see it. But now that you mention it yes, White has two two squares available to get out of check, and so 33...Qed3# or 33...Nc3# would do the job. Had Black continued I would have seen it. These "puzzles" and over the board when the other side resigns, I sometimes don't see why. I don't usually do a postmortem when I win, and I do win sometimes (: But thanks for the comment, I never know if anyone reads these comments. 

Mar0420   Damenlaeuferbauer: After long pondering, the great Silesian chess player Theodor von Scheve, whose most famous game is unfortunately his loss against the "Praeceptor Germaniae" Dr Siegbert Tarrasch, Leipzig 1894, where he made the decisive error already on the 7th (!) move, finally found the win with 30.,Rxc3+! 31.Kd2 (31.bxc3,Qxc3+ 32.Bc2,Qxa3+ 33.Kd2 [33.Kb1,Nc3#] 33.,Qe3#) 31.,Qxb2+ 32.Ke1 (32.Bc2,Qxc2+ 33.Ke1,Re3+ 34.Kf1 [34.Qe2,Qxe2#] 34.,Qh2 35.Qg4 [to prevent 35.,Qh3+] 35.,Rxg3 36.Qc8+,Kh7 37.Qf5+,g6 +) 32.,Re3+ 33.Kf1 (33.Qe2,Qxe2#) 33.,Qh2 34.Qh7+ (34.Bh7+,Kf8 35.Qg4 [to prevent 35.,Qh3+] 35.,Rxg3 36.Qc8+,Kf7 37.Qg8+ [37.Bg8+,Kg6] 37.,Ke7 38.Re1+,Kd7 39.Qf7+,Kd6) 34.,Kf8 35.Qh8+,Kf7 36.Qf7+,Kd6 and the black king escapes) 36.Qh5+,Ke7 37.Qg4 (to prevent 37.,Qh3+) 37.,Rxg3 +. 

Mar0420   Walter Glattke: Saturn 2, you are right, I set a question mark für Rxd3, otherwise of course no perpetual. 

Mar0420
  AylerKupp: <<al wazir> The win is less clear after 31. Kd2. In fact, I'm not sure black does win.> (part 1 of 2) There seemed to be some uncertainly as to whether Black wins after 30...Rxc3 31.Kd2 although several players posted analyses that indicated that Black still won. So I had some engines analyze the position after 30...Rxc3 31.Kd2:
click for larger viewTheir conclusion: <White> has either a forced mate or a clear win after 30...Rxc3 31.Kd2. The following are the engines Principal Variations: <Stockfish 11>: [+M25], d=46, < 2 mins of calculation: 31...Qxb2+ 32.Ke1 Re3+ 33.Kf1 Qb5+ (I can only assume that Black realized that it is lost so the horizon effect kicks in and Black tries to delay the inevitable as long as possible. But this is a move that likely no human would consider since it stops Black's attack and any winning chances) 34.Qxb5 cxb5 35.Bg6 Kf8 36.Rd8+ Ke7 37.Re8+ Kd6 38.Rxe3 (Black could, of course, safely resign at this point if not earlier but this is an engine, and engines don't resign) 38... a5 39.Re8 b4 40.axb4 a4 41.Ra8 b5 42.Be4 f5 43.Ra6+ (43.Bf3 would have been rubbing it in since Black's king can then not penetrate into either the qside or the kside and after White's king penetrates in the center will eventually have to play ...a3 and lose all its pawns) 43...Kd7 44.Bxf5+ Kc7 45.Be4 (and now White accomplishes the same thing I suggested except that its pieces are even better placed than in my line) 45... Kb8 46.h5 Kc7 47.Rg6 a3 48.Rxg7+ Kb6 49.Rg8 a2 50.Ra8 Kc7 51.Rxa2 Kb8 52.h6 Kc8 53.h7 Kc7 54.h8Q Kd6 55.Ra7 Ke6 56.Qe5#
click for larger viewStockfish actually found a matein30 at d=41 and then found successively shorter mates until it found the line listed above. And alternatives to 31...Qxb2+ fare much worse, matein9 after 31...Qxb1 and matein6 after 31...Rd3+. <Houdini 6>: Houdini found the mates in 6 and 9 moves easily enough at d=15 after 4 secs of calculation after 31...Rd3+ and 31...Qb1 respectively, but it did not quite find a mate after 31...Qxb2+ after almost 2 hours of calculation when I lost patience. However, its evaluation of [+96.12], d=38 is pretty much a win after 31...Qxb2+ 32.Ke1 Re3+ 33.Kf1 Qb5+ (same approach as Stockfish even though it effectively means abandoning all winning chances) 34.Qxb5 cxb5 35.Bg6 (it appears that this is the winning move and Black can probably reasonably resign here since the loss of the rook can't be prevented) 35...Kf8 36.Rd8+ Ke7 37.Re8+ Kd6 38.Rxe3, and it found this at d=9 after less than 1 sec of calculation.
click for larger viewAnd even I would resign this position as Black. 

Mar0420
  AylerKupp: <<al wazir> The win is less clear after 31. Kd2. In fact, I'm not sure black does win.> (part 2 of 2) Houdini's analysis continued: 38...a5 39.Re8 b4 40.axb4 a4 41.Rg8 (here Houdini deviates from Stockfish, tempting Black to advance its apawn but it's just a tease (hmmm, can engines tease?), White can get its rook to the afile in time) 41...a3 42.Ra8 (White could prolong its teasing by 42.Ke2 and if 42...a2, 43.Ra8 but I guess Houdini is not programmed to carry the tease to its ultimate conclusion) 42...Kc6 43.Rxa3 Kb5 44.Ra7 Kb6 45.Ra8 (45.Ra5 would have been teasingly more consistent) 45...Kb5 46.Rg8 Ka4 47.Rxg7 b6 (inexplicable, why not 47...b5 and 48...Bxb4? Not that it matters) 48.Rd7 Kxb4 49.h5 Kb5 50.Ra7 Kc5 51.h6 Kd4 52.h7 f5 53.h8Q+ Ke4 54.Qe5+ Kd3 55.Qxf5+ Kc3 56.Qd3+ Kb4
click for larger viewI think that we will all agree that this is a winning position for White. <Komodo 12.3> Komodo also didn't have any problem finding the mates in 6 and 9 moves easily enough after 4 secs of calculation after 31...Rd3+ and 31...Qb1 respectively. And after 31...Qxb2+ it had found 32.Ke1 Re3+ 33.Kf1 Qb5+ 34.Qxb5 cxb5 35.Bg6 Kf8 36.Rd8+ Ke7 37.Re8+ Kd6 38.Rxe3 at d=9 and less than a second of calculation. So I didn't bother letting it calculate much longer and I won't post its line since I feel that it would probably take it along while to reach as high an evaluation as Houdini reached in about 2 hours of calculation (Komodo reached an evaluation of [+17.30] at d=26 after about 2½ mins of calculation. The search for alternatives for Black will continue. 

Mar0420   RandomVisitor: After 20...Rad8 white has a winning move:
click for larger viewStockfish_20021515_x64_modern:
<50/89 38:04 +4.32 20.Bh3 Qb6 21.f5> Qxd4 22.Rxd4 Bh7 23.Re7 Nc5 24.b4 Na4 25.Rxb7 Nc3 26.Rd3 Ne2+ 27.Kb2 Nxg3 28.Re7 g6 29.f6 g5 30.Rxg3 Rxd6 31.hxg5 Rb8 32.Kc3 Ra8 33.Rge3 Bg6 34.Bg2 a6 35.Kb3 Kh7 36.c4 Rc8 37.Ka4 Rd1 38.c5 Rg1 39.Re2 Rf8 40.Bxc6 Rxg5 41.Re8 Rxe8 42.Bxe8 Kh6 43.Ka5 h4 44.a4 h3 45.Rh2 Bf5 46.Kxa6 Kg6 47.b5 Kxf6 48.b6 Bc8+ 49.b7 Bxb7+ 50.Kxb7 Rxc5 51.Rxh3 Ke5 52.Kb6 Rc1 53.Bxf7 

Mar0420   RandomVisitor: <AylerKupp>31.Kd2 Qxb2+ 32.Ke1 Re3+ 33.Kf1 <Qh2> is likely good for black click for larger viewStockfish_20021515_x64_modern: TB6
67/48 06:39 6,495,725k 16,277k M24 33. ... Qh2 34.Qh7+ Kf8 35.Qh8+ Kf7 36.Qh5+ Ke7 37.Qg4 Rxg3 38.Qxg3 Qxg3 39.Rd3 Qxf4+ 40.Ke2 Qc1 41.Rb3 Qc4+ 42.Rd3 Nc3+ 43.Kf2 Nxb1 44.Rg3 Qxh4 45.Kg2 Kf7 46.Rb3 Qe4+ 47.Kg3 Qe1+ 48.Kh2 Qe2+ 49.Kg1 Qd1+ 50.Kf2 Qxb3 51.Ke2 Qc2+ 52.Ke3 Nc3 53.Kf3 Qd3+ 54.Kg4 Ne4 55.a4 Qg3+ 56.Kf5 Nd6+ 

Mar0420   cormier: 30...Rxc3 31.Kd2 Qxb2+ 32.Ke1 Re3+ 33.Kf1 Kf8 maybe ..... 

Mar0420   5hrsolver: I too did not find a good continuation after 30...Rxc3+ 31. Kd2 Qxb2+ 32. Ke1 Re3+ 33. Kf1 The chessgames engine here gives 33... Qh2 as the right continuation. It makes sense because it takes advantage of whites very weak kingside. If black can. capture the g3 pawn with check then checkmate follows quickly. For example. 34. Rd3 Qh3+ 35. Kf2 Qxg3+ 36. Kf1 Re1# or 34. Bd3 Qh3+ 35. Kg1 Qxg3+ 36. Kh1 Qh3+ 37. Kg1 Rg3+ 38. Kf2 Qh2+ 39. Ke1 Rg1+ 40. Bf1 Qg3+ 41. Kd2 Qc3+ 42. Ke2 Qe3# Of course the other question is if Blacks king can avoid checkmate or perpetual. I think the black king can hide behind the rook on e3. Here is the actual engine line that shows black can survive the checks 34. Qh7+ Kf8 35. Qh8+ Kf7 36. Qh5+ Ke7 37. Qg4 Rxg3 and white has to give up the queen to avoid checkmate on g1. if 38. Re1+ then Kd8 avoids the checks. 

Mar0420   TheaN: Rather straightforward idea, where after <30....Rxc3+ 31.Kd2 (bxc3 Qxc3+ 32.Bc2 Qxa3+ 33.Kd2 Qe3#/Kb1 Nc3#) Qxb2+ +> Black has broken the king's shield. It's not entirely trivial but Black can escape perpetual and is winning. <AylerKupp> second time now (Monday too) you posted tenths of lines of analysis from a wrong position, you omitted the knight on d5. Of course Black's lost in that case. 

Mar0520   cormier: Mar0420 RandomVisitor: <AylerKupp>31.Kd2 Qxb2+ 32.Ke1 Re3+ 33.Kf1 <Qh2> is likely good for black Stockfish_20021515_x64_modern: TB6
67/48 06:39 6,495,725k 16,277k M24 33. ... Qh2 34.Qh7+ Kf8 35.Qh8+ Kf7 36.Qh5+ Ke7 <37.Qg4> Rxg3 38.Qxg3 Qxg3 39.Rd3 Qxf4+ 40.Ke2 Qc1 41.Rb3 Qc4+ 42.Rd3 Nc3+ 43.Kf2 Nxb1 44.Rg3 Qxh4 45.Kg2 Kf7 46.Rb3 Qe4+ 47.Kg3 Qe1+ 48.Kh2 Qe2+ 49.Kg1 Qd1+ 50.Kf2 Qxb3 51.Ke2 Qc2+ 52.Ke3 Nc3 53.Kf3 Qd3+ 54.Kg4 Ne4 55.a4 Qg3+ 56.Kf5 Nd6+ 

Mar0520
  AylerKupp: <<TheaN> <AylerKupp> second time now (Monday too) you posted tenths of lines of analysis from a wrong position, you omitted the knight on d5.> Well, unfortunately that just validates my post of Z Sturua vs V N Kozlov, 1975 (kibitz #35) that, because I probably post more than you do, that means that my posts are sillier than yours might be. 

Mar0520
  AylerKupp: <<cormier> <AylerKupp>31.Kd2 Qxb2+ 32.Ke1 Re3+ 33.Kf1 <Qh2> is likely good for black> Yes, but as <TheaN> pointed out, not if Mieses gets knight odds by removing von Scheve's knight from d5! I don't know if I should flag my careless error with a :) or a :( 

Mar0520
  AylerKupp: Sigh. I had just finished a 3engine analysis using Stockfish 11, Houdini 6, and Komodo 12.3 which indicated that White has a forced mate after Black's "best" move 33...Qb5+ but, as <TheaN> pointed out, all the analyses were based without Black's Nd5. I was then going to see if there were improvements in Black's play prior to 32...Re3+ since (without Black's knight on d5) after 33.Kf1 Qb5+ 34.Qxb5 cxb5 35.Bg6 that allows the forced win of Black's rook after 35...Kf8 36.Rd8+ Ke7 37.Re8+ Kd6 38.Rxe3. But, as often happens <RandomVisitor> beat me to it and found that White has a winning move after 20...Rad8 so all those analyses are moot in addition to being just wrong. Curse you, Red Baron! But thankfully you spared me having to do additional analyses with the wrong initial position. I think that after this goof I'm going to take a few days off from posting further analyses. However, <RandomVisitor>, your analysis threw me off since the position you showed was after 19...Rad8 and not after 20...Rad8. So after "20...Rad8" I couldn't see how White could play 20.Bh3. A small nit, certainly when compared to leaving out a piece in the initial analysis position.. As you know, long engine analyses should be taken with a grain of salt. Forward sliding is needed to validate it. And this is a complex position so an engine's best line can change significantly from search ply to search ply. For example, I let Stockfish 11 analyze the position after 19...Rad8 and at d=36 it considered that the best line started with 20.Bh3 c5 21.Qe5 .... At earlier plies it considered that the best line started with 20...Qa5 (d=34), 20...Rd7 (d=31), 20...Qb6 (d=30), 20...b6 (29), and so on. And, yes, this analysis was also done without the knight on d5 but that was OK since the knight didn't move to d5 until 27...Nd5. How consistent was your Stockfish_20021515 in considering 20...Qb6 as Black's best reply to 20...Bh3? And as you also know, multicore engines' nondeterminism is such that one person's analysis may not be duplicatable by others. So attempting to find Black's actual losing move would not be easy. Shades of trying to find a forced win for White after 29...Bxh2 in Spassky vs Fischer, 1972. Which, BTW, has a nifty title that I don't remember noticing before, "Bish, You Were Here". :) 

Mar0720   morfishine: After <29...Qa1+> I just knew Black had something up his Scheve 


< Earlier Kibitzing · PAGE 2 OF 2 ·
Later Kibitzing> 


