< Earlier Kibitzing · PAGE 3 OF 3 ·
|Oct-12-08|| ||JohnBoy: While I solved this, I am not entirely satisfied. It seems that 23...Qxe5 holds longer than the path chosen. Any easy demolitions of the black position after this try?|
|Oct-12-08|| ||Jimfromprovidence: To me, the position is extremely challenging and noteworthy after 23…Bd7 24 Bg8 Qxe5.
click for larger view
25 Qxh6+ is dead wrong because after 25...Kxg8, all white is left with is a draw by repetition.
The winning move is the gutsy 25 Rxf6!, (because it sacrifices the a rook next move and puts the king in check.).
Now, the combination continues with 25…Qxa1+ 26 Kg2 Qb2+ 27 Kh3 e5+.
click for larger view
But, after 28 Ne6 Bxe6+ 29 Rxe6 black is finally out of magic.
click for larger view
29…Qb1 is unavoidable to stave off mate. 30 Qxb1 Rxg8 31 Rxe7 Rg7 32 Rxg7 Kxg7 33 Qxb7+ finishes beautifully for white.
|Oct-12-08|| ||kkshethin: 19.Bxh6 gxh6 20.Bh7+ Kh8 21.Qg6 Qxb2+ 22.Kh1 Nf6 23.Ng5
(23...Qxe5 24.Qxh6 Nxh7 25.Rxf8+ Bxf8 26.Nf7+ Kg8 27.Qg6+ Bg7 28.Nxe5 Nf6 29.Qf7+ Kh8 30.Ng6+ Kh7 31.Nf8+ Kh6 32.Qg6#)
(24...Qxf6 25.Rxf6 Bxf6 26.Bg8 Rxg8 27.Qh6#)
25.Bg8 Qc2 26.Qh6+ Qh7 27.Qxh7#
|Oct-12-08|| ||VooDooMoves: With black queen, queen rook, bishop and knight all pretty much stranded on the Q-side the position is ripe for a king hunt. Here's what I found (<TheAn> I set up the board correct this time but what I found is so weird you may wish to LOL again today.If I can't solve these puzzles at least I have the power to bring smiles to people's faces :)|
19. Bh7+ Kh8 (Kf7?? 20. Qg6#) 20.Bxh6! and the threat is (21. Bxg7+ Kxg7 22.Qg6+ Kh8 23. Bg8! threatening mate on h7 and the two defenses, 23...Nf6 and 23...Rxg8, lose to 24. exf6 and 24. Qh6 # respectively) 20...gxf6 21. Qg6, threatening the same as before except with black to move and he has the strange defense 21...Qxb2+ (although it still fails). 22. Kg1 Nf6!? and now 23. exf6 Qxf6 allows black to trade and thereby relieve the pressure but...23. Rab1! Qc3 24. Rec1! This seems to win.
|Oct-12-08|| ||johnlspouge: So as to avoid filling the kibitzing with errors, I analyzed with Toga and replaced faulty analysis <<>as indicated>.|
Serper vs Barsov, 1988 (19.?)
White to play and win.
Material: Even. The Black Kg8 has 2 legal moves (including f7, on the same file as the White Rf1). White has a battery Qc2 and Bd3 controlling the weak light squares around Kg8 and able to check at h7. The activation of Qc2 is desirable, because the attack would be much more powerful if Qc2 and Bd3 were interchanged. The White Bc8 controls some complementary dark squares around Kg8 and in particular, attacks Ph6. The White Nf3 has little scope at present, but protects Pe5, which controls f6 and prevents the Black Nd7 from contributing to K-side defense. The White Ra1 requires activation, either by clearing Bc1 and reloading Rf1 or by a R-lift Ra3. The primary action will be on the K-side, where the White Q and Bs provide some local superiority, but any sacrificial combination should hinder the development of the Black Q-side and in particular, the bottleneck Nd7.
Candidates (19.): Bh7+, Bxh6
Candidate order is not very important, but 19.Bh7+ 20.Bxh6 is more forcing than 19.Bxh6 20.Bh7+.
19.Bh7+ Kh8 [Kf7 Qg6#]
20.Bxh6 gxh6 [Rxf3 21.Bxg7+ Kxg7 22.Qg6+ Kf8 23.Qg8#]
[else, drop a critical P without compensation]
21.Qg6 (threatening 22.Qxh6 23.Bg6+ 24.Qh7#)
As usual, Black has 3 types of response: counterattack, reinforcement, or retrenchment.
Counterattack is infeasible:
<Here I went for the flashy 22.Rf2 but missed
22…Qxf2+ 23.Kxf2 Nxe5
My second choice was a lemon also:
22…Qxf2+ 23.Kh3 Nf3 24.Ng5 hxg5 25.exf6 <e5+>
Note: The game reply 23.Kh1 does not permit 25…e5+.>
<<>Here I went for 22.Qxh6 (threatening 23.Bc2+ Kg8 24.Qh7#) but missed 22…Ng4, removing the discovered check.>
Retrenchment is impossible because Kg8 cannot run, so reinforcement of Ph6 is the only remaining option:
(3) 21…Bg5 22.Nxg5 (threatening 23.Nf7+ 24.Qg8# and 23.Rxf8+ 24.Qg8#)
Black is faced with mate. <Toga gives a mate-in-4.>
|Oct-12-08|| ||johnlspouge: I would like to know the candidate order engines or humans prefer, and (in the case of computers, speculation on) how the preference of the order is determined. To forestall nugatory responses, I do understand that in the present case, move order does not matter much, but that just makes it an interesting example to me.|
Toga preferred 19.Bxh6, where I preferred 19.Bh7+ as more forcing.
|Oct-12-08|| ||5hrsolver: 23...Qxe5 24.Qxh6 threatening 25.Rxf6 Qxa1 26Bb1 check and mates |
if 24..Qxg5 25QxQ Kxh7 26Rf4 threatens mate on the h file
It took me some time to figure (10 hrs) that the key to the solution are all the weak white squares around the black king and the potential Bg8 sacrifice.
|Oct-12-08|| ||ku0826: The final move,Bg8 looks like the Columbus'Egg.|
|Oct-12-08|| ||5hrsolver: In considering move order (19.Bxh6 or 19.Bh7) it may be useful to consider that black can refuse the Bxh6 sacrifice hence it may be better to sacrifice first before committing your other forces.|
|Oct-12-08|| ||JohnBoy: <Jimfp ... after 23…Bd7 24 Bg8 Qxe5, 25 Qxh6+ is dead wrong because after 25...Kxg8, all white is left with is a draw by repetition.> Not quite. 26.Rxf6 and black can't easily defend h7. Death is imminent.|
|Oct-12-08|| ||johnlspouge: < <JohnBoy> wrote: <Jimfp ... after 23…Bd7 24 Bg8 Qxe5, 25 Qxh6+ is dead wrong because after 25...Kxg8, all white is left with is a draw by repetition.> Not quite. 26.Rxf6 and black can't easily defend h7. Death is imminent.>|
Toga II 1.3.1 evaluates 26.<Rxf6> as follows. (Humans can improve near the end of the full computer variation.)
[ply 15/58 time 00:07 value (to White) -9.13]
23…Bd7 24.Bg8 Qxe5 25.Qxh6+ Kxg8 26.<Rxf6> Qxa1+ 27.Kg2 Qb2+ 28.Kh3 e5+ 29.g4 Qc3+ 30.Kh4 Qe1+ 31.Kh5 Bxg4+ 32.Kxg4 Qe2+ 33.Kh4 Qxh2+ 34.Kg4 Qg1+ 35.Kh5 Qh1+ 36.Kg4 Qxh6 37.Rxh6 Rf4+ 38.Kh5 Rxa4 39.Re6 Bxg5 40.Kxg5 e4 41.Re7 b5 42.Rc7
My chessforum gives instructions for downloading Toga, which is freeware.
|Oct-12-08|| ||MiCrooks: I found this quickly by feel as well, though I need to check. I saw both that I could transpose the first two moves and chose to play the less commital move first (Bh7+). I did miss the impact of the Queen check. It was incidental but it did force White to be more precise (as the rooks couldn't really come into play).|
So didn't fully realize the solution but I would have played this over the board. I have a real weakness for sacking bishops into the opponents kingside!
|Oct-12-08|| ||MiCrooks: Oh and as to all of the computer talk, that Queen move actually hurts their ability to find the right line here. Most PC's do some selective trimming of their search trees to try to speed things up. Hiarcs I think is supposed to be very good at this.|
Here the in between move with Qxb2+ adds to the tree as well as the negative evaluation. Both hurt any computers ability to find the right response. This is 20+ ply deep to find the right move, and very few computer setups can get that deep in any reasonable time frame.
If you go 7-ply into the variation (after Kh1) the computers start to see that things aren't so rosy. Go another two ply along the game line, even though it is the same line the computer thought was best, and it finally gets to where it sees Black is losing. But that is because the searches are going into what is effectively the low 20's at that point.
|Oct-12-08|| ||zooter: Why 23.Ng5 at all? Why not directly exf6 followed by Bg8?|
|Oct-12-08|| ||MiCrooks: On the Bd7 line White does walk a very fine line, but it is one that is pretty easy to find. I say easy because it is easy to see that the alternatives lose quickly! It is in positions where lots of things look possible that it is easier to go astray.|
In the end Black has the same choice as in the game...drop his Queen on c2 and trade it for the Bishop or get immediately mated.
That final position is quite nice though, as illustrated above in Jim's third and final diagram.
|Oct-13-08|| ||MostlyAverageJoe: <MiCrooks: Oh and as to all of the computer talk...>|
I agree with your comments. The engines have no problems finding the continuations after the line is pushed some 6-7 plies forward, but somehow they must be pruning the leading moves and concentrate on what seems to be more promising side branches.
A pleasant surprise from Toga II 1.3.1, which I just added to my set of engines: in inifinity analysis, it did find 19. Bh7 and 19.Bxh6. After 19 plies (1h40m of runtime), its diagnosis of the two best lines is:
1. (+2.71) 19. Bh7 Kh8 20. Bxh6 Rf5 21. Bxf5 exf5 <the rest is less reliable: 22. Bg5 Bxg5 23. Nxg5 Nf8 24. Rad1 Qg6 25. h4 Bd7 26. Qc7 Bxa4 27. Qxb7 Bc6 28. Qf7 Rb8 29. Rd2 d4 30. Kg1 Qxf7 31. Nxf7 Kg8 32. Rxf5>
2. ( : +2.48) 19. Bxh6 Rf5 20. Bxf5 exf5 21. Bg5 Bb4 <the rest is less reliable: 22. Rac1 Nf8 23. Qd3 Ng6 24. Qxd5 Be6 25. Qd4 Qa6 26. h4 Ne7 27. Bxe7 Bxe7 28. Rc7 Bd8 29. Rc3 Be7 30. Qd3 Qb6>
|Oct-13-08|| ||znprdx: Well I'm glad I came back to this a day later - the immediate 19.Bh6 seemed needlessly dramatic - and I doubt Serper ‘saw’ 25.Bg8 since technically Bxh7+ first is more forcing as <5hrsolver:> pointed out and <McCrooks> elaborated: "It is in positions where lots of things look possible that it is easier to go astray."
I think the key factor here would be time. After the provocative 19.Bh6 both...Bf5 and Rx[N]f3 are plausible spoilers. There must surely be something better than ...21Qxb2+ I’ll be back....but I notice that others have tried using computers: tsk-tsk Chess is played between people. Would a computer have ended up with a dead Queen Bishop?|
|Oct-13-08|| ||agb2002: Black has problems to finish his development and the white squares around the black king are weak.|
A) 19.Bh7+ Kh8 20.Qg6
A.1) 20... Rd8 21.Bxh6 Qxb2+ 22.Kh3 Bf8 22.Bg8 Kxg8 23.Ng5 followed by Qh7 or Qxg7 mate.
A.2) 20... Nc5 21.Bxh6 Qxb2+ 22.Kh3 gxh6 23.Bg8 Rxg8 24.Qxh6 mate.
B.1) 19... gxh6 20.Bh7+
B.1.a) 20... Kg7 21.Qg6+ Kh8 22.Qxh6 Nf6 23.exf6 Qxb2 24.Kh3 Qxf6 25.Bg6+ Kg8 26.Qh7#
B.1.b) 20... Kh8 21.Qg6
B.1.b.1) 21... Qxb2+ 22.Rf2
B.1.b.1.i) 22... Qxa1 23.Bg8 Rf7 24.Bxf7 threatening Qg8 and Qh6 mate.
B.1.b.1.ii) 22... Qxf2+ 23.Kxf2 Rxf3+ 24.Kg2 Rf8 25.Bg8 as in the previous line.
B.1.b.2) 21... Nxe5 22.Nxe5 Qxb2+ 23.Kh3 Qxe5 24.Rxf8+ and 25.Qg8 mate.
B.2) 19... Rxf3 20.Bh7+
B.2.a) 20... Kh8 21.Bxg7+ Kxg7 22.Qg6+ and 23.Qg8 mate.
B.2.b) 20... Kf8 21.Bxg7+ Ke8 22.Rxf3 and black looks very bad.
B.3) 19... Rf7 20.Bh7+
B.3.a) 20... Kh8 21.Qg6 Qxb2+ 22.Kh3 and the rook seems to be lost.
B.3.b) 20... Kf8 21.Ng5 Bxg5 22.Rxf7+ Kxf7 23.Qg6+ Ke7 24.Bxg5+ winning.
I think I’d try line A). Time to post and check.
|Oct-13-08|| ||agb2002: Today's mistakes:
1) I didn't see that 22... Nf6 was not met with 23.exf6 but 23.Ng5 attacking through the f-column and indirectly black' back rank, although I saw this possibility in my line B.1.b.2).
2) I preferred 22.Kh3 instead of 22.Kh1 because the queen could take the rook on a1 with check in some lines, as pointed out by <Jimfromprovidence>
|Oct-13-08|| ||kevin86: Black is doomed-an effort to dodge the mate on h7 by 25...♖xg8 loses to the epaulette mate by 25 ♕h6#|
|Oct-13-08|| ||johnlspouge: < <MostlyAverageJoe> wrote: [snip] A pleasant surprise from Toga II 1.3.1, which I just added to my set of engines: in inifinity analysis, it did find 19. Bh7 and 19.Bxh6.>|
I paid for Rybka, but I use Toga. 'Nuff said.
|Oct-24-08|| ||patzer2: For the difficult Sunday Oct 12, 2008 puzzle solution, White wins with the deep demolition of pawn structure combination 19. Bxh6! |
However, it would appear from <MostlyAverageJoe>'s Toga II 1.3.1 analysis above that 19. Bh7+! Kh8 20. Bxh6! works as well or better.
An interesting aspect of the combination is an examination of the not-played side-line 23...Bd7, as first examined by <jimfromprovidence>. Jim indicates <after 23…Bd7 24 Bg8 Qxe5 25 Qxh6+ is dead wrong because after 25...Kxg8, all white is left with is a draw by repetition.> He goes on to demonstrate a fantastic White win after 23…Bd7 24 Bg8 Qxe5 <25. Rxf6!>. However, <johnboy> and <johnlspouge> indicate that after 23…Bd7 24 Bg8 Qxe5 25. Qxh6+ Kxg8 <26. Rxf6!> White is still winning.
In any case, this combination makes for a good study of a deep demolition of pawn structure mating combination.
|Oct-24-08|| ||Jimfromprovidence: <patzer2> <However, <johnboy> and <johnlspouge> indicate that after 23…Bd7 24 Bg8 Qxe5 25. Qxh6+ Kxg8 <26. Rxf6!> White is still winning.>|
Thanks for the comments, but for the record, <jls>'s analysis supported my line, not <johnboy>'s.
|Dec-29-08|| ||Domdaniel: Nobody plays ...h6 in the French without considering the possibility of Bxh6 by White - especially when white, like Serper, was rated 2450 at the time of the game. This is conscious provocation by Barsov: actually encouraging Bxh6 because he believed his defences could hold, particularly with the neat move 22...Nf6. But white had an even neater one in 25.Bg8!!|
Lasker and Steinitz have nothing to do with it. Forget aphorisms - this is raw calculation. In fact, if white had passed on the sac with, say, 19.Bb5, then Black has a promising exchange sac with 19...Rxf3 20.Rxf3 Nxe5, which is roughly equal.
Essentially, both players had to see the whole combination - with all its sidelines - at move 18/19. Black must have felt that the extra piece, the counterplay with ...Qxb2+, and the clever ...Nf6 was sufficient to take the risk. It's called heroic defence, and every French player knows it.
And no, they probably didn't spend much time on Lasker or Steinitz - who wouldn't have understood moves like 7...a5 or 8.a4 very well. But I'm sure they studied Botvinnik, Korchnoi and Vaganian.
|Dec-29-08|| ||Domdaniel: Interesting comments all round today, btw. Fritz 'finds' 19.Bxh6 at 18-ply, but rates it as equal, 0.00 -- it has to go another couple of plies before finding the win. Interestingly, it also rates the 19.Bb5 Rxf3 line as dead equal.|
In cases like 19.Bh7+ vs 19.Bxh6, it's true that the sac might be refused, thus the check is more forcing. But, from an engine POV, the optimal lines transpose - hence there is absolutely no difference between them, and the selection order is irrelevant.
If analysing with an engine I always display 2 or 3 lines at least, so the problem of finding one key move is avoided. It's also useful in cases where one line is clearly better than the rest, as they can be eliminated.
Finally, I accept that *some* players would have played 19.Bxh6 'on instinct' and quite possibly found the rest of it, move by move. I just don't think that happened here.
< Earlier Kibitzing · PAGE 3 OF 3 ·