Paying Homage to the Master
Nick Larter, Oct. 2009
Ennis Chess Club and IM Gerard Welling have ‘Previous.’ It’s 2005 in the sleepy Val’d’Aostan backwater of Saint Vincent, our first venture into the dizzy heights of the European Club Cup and our captain John Cassidy and this creative and hugely respected IM serve up a classic c3 Sicilian battle that Welling, playing Black, finally shades when his king marches fearlessly up the board to support his remaining Queen and Knight [1]. His team HMC Calder whitewashed us 6-0 that day and so, four years on in the delightful southern Macedonian city of Ohrid there’s a bit of pride at stake when we draw the palpably misfiring 21st seeds in the last round –our reward for our best ever ECC return of four points – and it’s my turn to test my mettle against the man.
From my preparation it seemed that my 1.e4 would be most likely met with one of 1…d6, 1…c5, 1…g6 or 1…d5. I could aim to transpose any of the first three into my usual Botvinnik English type set-up, confident in the expectation that a few well timed off-thewall moves from my opponent would soon get us into unknown territory. For the last, I was faced with daring to play my normal transposition into the Blackmar-Diemer Gambit against a player who, certainly in his younger days if maybe somewhat less so today, was an acknowledged BDG aficionado and expert.
So this is the story that unfolds below and whilst I would generally be reticent in assuming that anyone would be much interested in one of my chess games, I’m persuaded to make an exception here, firstly because of the amusing back-story that emerges and secondly because of the brain-twisting and highly theoretical denouément that may be new and of interest to some, as it certainly was to me. Scene setting done, let’s see what happens…
N. Larter 1799 – G. Welling IM 2372
European Club Cup, Ohrid, 2009
Round 7: Ennis Chess Club v HMC Calder
1.e4 d5 2.d4 So BDG it is dxe4 3.Nc3 Nf6 4.f3 exf3 5.Nxf3 c6 and naturally enough we enter the Ziegler defence, a bastard cousin of the Caro-Kann, which usually forms the basis of the purported refutations of the BDG that appear in print from time to time, of which IM Andrew Martin’s ‘Shopping for a Tombstone’ [2,3] is perhaps the best known example. 6.Bc4 Bf5 7.Ne5 e6 8.0–0 Bxc2 I was playing for this line and was a little surprised that Welling went into it. 9.Nxf7!?
And for those of you still in the dark, let’s hand over to IM Gary Lane to reveal the novelty of what is transpiring, for of this move he writes [4] This tremendous move by Welling has revived the entire line – yes I’m playing the very variation that my opponent himself had invented almost 30 years ago!
By now our wry grins across the board had attracted the attention of the neighbouring games and one or two of the HMC Calder players, including GM Giri, came over for a look and immediately grasped the joke. 9…Bxd1 (clearly not 9…Kxf7?, which gives White a raging attack, eg.; 10.Qg4 Qd7 11.Bg5 Na6 12.Bxf6 gxf6 13.d5! Nc7 14.dxe6+ Nxe6 15.Rae1 Re8 16.Ne4 Bxe4 17.Rxe4 Bc5+ 18.Kh1 Re7 19.Qh5+ 1-0, Welling – Marzik, Biel 1981) 10.Nxd8 Kxd8 11.Rxd1 Nd5 12.Re1 Kd7 Alternatively, in Welling – Wrobel, Luxembourg 1982, Black tried to free his position through some exchanges with 12…Bb4 13.Rxe6 Nxc3 14.bxc3 Bxc3 but failed and lost after 15.Bg5+ Kd7 16.Re7+ Kd6 17. Rd1 h6 18.Re6+ Kd7 19.Bf4 Re8 20.Rd6+ Ke7 21.Rd3 Bb4 22.Re3+ Kf8 23.Rf6+ 1-0
But back to the game. 13.Bg5 h6 14.Bd2 I think this is best – the immediate 13. Bd2 doesn’t create the potentially useful hole for me on g6, whilst the retreat 14. Bh5, although weakening the important f6 post for Black’s Knight after 14…g5 15.Bg3, gives him the a1-h8 diagonal and leaves my piece offside – it’s more flexibly placed on d2. Be7 15.Ne4 b5 16.Be2 Again more flexible than
16.Bb3 and the pressure on e7 can potentially be renewed later with Bg4. Nf6 17.Nc5+ Bxc5 18.dxc5 a5 19.Bf4 Na6 20.Rad1+ Nd5 21.Bd6
So Black has engineered a passed pawn but White has a lot of positional compensation – the threat 22.Rf1 is difficult for Black to meet (21…Ke8 fails to 22.Bg4; 21…Nf6 fails to 22.Bf8+) and I think Welling, in opting to sacrifice the exchange to thwart this move whilst removing the annoying piece on d6 in compensation, finds the best plan. 21…Rhe8 (Not 21…Rhf8? 22.Bxf8 Rxf8 23.a4! 22.Rf1 Re7 23.Bxe7 Kxe7 24.Rfe1 Nxc5?! (I was initially shocked when he played this, since he must have seen that White will immediately win the pawn back with a tactic. I felt at the time that the c5 pawn was so weak that Black had time to prepare to capture it more safely and gain a two pawn advantage for the exchange. However on reflection maybe it’s not so clear cut, since the natural 24…Ndc7 neutralising 25.a4 and with the idea 25…Nxc5 does not work after 25.Rd6 Nxc5 26.Rxc6 and anything else beginning with 24…Nac7 is probably too slow, eg; 25.a4 bxa4 26.Ra1) 25.Bxb5 Kd6 (If 25…cxb5 26.Rxd5) 26.Bc4 Rb8 27.b3 Nd7
Here I start to drift and it’s where I lose any realistic winning chances. Having overlooked the elementary 29…Rb7, it’s clear I’m not going to improve my position any further by one-move threats and I would have been better employed organising a blockade of black’s trump card – the passed e pawn. One line could go 28.Re2 N7d6 29.Rde1 Nf4 (or else 30.g3) 30.Rxd6+ Nxd6 31.Rxd6+ Kc5 with chances for an edge for White. 28.Re3 N7f6 29.Rg3? Rb7 30.Rf3 e5! Naturally enough. 31.h3 e4 32.Rf5 Re7 33.g4 Kc5 34.g5 I could try to restrain the King with 34.a3 but then 34…g6 35. Rff1 e3 36. Be2 Ne4 is unpleasant for White. Alternatively just a simple and natural 34…a4 would threaten to open up my Queen’s side. hxg5 35.Rxg5 Kb4 36.Be2
sa
Pretty much forced. Simplifying into a materially equal endgame with 36.Bxd5 cxd5 37.Rgxd5 Nxd5 38.Rxd5 is superficially attractive, but after 38…e3 39.Kf1 Rf7+ 40.Ke1 Rf3 it’s Black who has all the active play. 36…Ka3 The Black King invades, but maybe I can turn it to my advantage by sacrificing a pawn to open some lines. Alternatively, 36…Nf4 37.Rd4+ Kc3 38.Rc4+ Kd2 would have become very messy for both sides. 37.Rd4 Kxa2 38.Ra4+ Kxb3 39.Rxa5 Nc3 40.Kf1 Nfd5 41.Rc5 Ne3+ 42.Kf2! Anything else loses eg; 42.Ke1? Nc2+ 43.Kd2 (43.Kf1 Rf7+ 44.Rgf5 Ne3+ 0-1)
Rd7+ 0-1 Nc2 43.Bc4+! Covering the vital f7 square.43…Kb4 44.Rxc6 And after some aggressive defence White has reduced the pawn deficit again. 44…e3+ 45.Kg2 45.Kf1 allows 45…Rf7+! and moving anywhere else loses to a Knight fork. 45…Nd4 46.Rc8 e2 47.Bxe2 Ncxe2
Now my plan becomes very simple – to sacrifice all my remaining material for Black’s Rook and pawn, leaving him with a King and two Knights versus my King. There now follows a long passage of play with White alternately threatening the g pawn and if a Knight moves to cover, switching to cutting off the Black King. 48.Rg4 Rf7 49.Rg8 Nf4+ 50.Kh2 Nfe6 51.Rc8 Kb5 52.Rg2 Nc7 53.Rg8 Nde6 54.Rc2 Nc5 55.Rg2 N7e6 56.Rc8 Kc4 57.Rd2 Rf3 58.Rh8 Kc3 59.Ra2 Kd4 60.Rh4+ Ke3 61.Rg4 Nd3? 62.Rg3 Rxg3 63.Kxg3 Part 1 accomplished. Ke4 64.Ra8 Ne5 65.Rg8 Kf5 66.h4 Nf7 67.h5 Nh6 68.Rxg7?! Nxg7
69.Kf3 After 68…Nxg7 I had kind of assumed that I had reached the finishing line with a draw, but with IM Welling playing on,it slowly dawned on me that this was not so. In fact, as he confirmed to me after the game, the position after Black’s 68th move is a theoretical win for the second player. The pawn needs to be on the 6th rank for White to be sure to draw on best play. What’s going on here of course is that with the pawn on, White has to lose a tempo with a pawn move while the Black Knights and King
are stalemating their opponent, enabling checkmate to be delivered the move following. In my play I had only considered the case of mate being delivered where the White King is hemmed in, the vital escape square being blocked by his own pawn: I thought by running across to the other side of the board I would be safe. Not so. However, theory is one thing, practice is another. The game concluded thus:- 69…Ne6 70.Ke3 Ke5 71.Kd3 Kf4 72.Kc3 Ke3 73.Kb2 Nd4 74.Ka2 Kd3 75.Kb2 Kc4 76.Ka3 Kb5 77.Kb2 Kb4 78.Ka2 Kc4 79.Kb1 Kd3 80.Ka2 Kc3 81.Kb1 Ne2 82.Ka2 Kb4 83.Kb2 Nd4 84.Kc1 Kc3 85.Kb1 Ne2 86.Ka2 Nc1+ 87.Kb1 Nb3 88.Ka2 Nd2 89.Ka3 Nb3 90.Ka2 Nd4 91.Kb1 Ne2 This in fact is a threefold repetition but I did not notice it at the time – it is immaterial since by this point IM Welling had decided to throw in the towel and this move was preparatory to an elegant draw offer made via the capture on h5. 92.Ka2 Nf4 93.Kb1 Nxh5 draw agreed. After the game Welling told me that in his original analysis of 9.Nxf7 he had eventually concluded that Black stood better after 12…Kd7, which would explain why he was happy to go into the line, ‘but clearly that assessment’s a load of rubbish!’ he added with a smile. And from my perspective, thanks to IM Lane for taking a more optimistic view of Welling’s invention than the man did himself – as he puts it [4] The ending after 9…Bxd1 underlines the significance of White’s superior development.
For the record, Ennis lost the match 4.5 to 1.5 – a fighting performance, with IM Petr Neuman and Rory Quinn also drawing on boards 1 and 4 respectively. But let’s return to the position after 68…Nxg7. This brings us into the arcane realm of Troitzky Endgames, named for Russian analyst A.A. Troitzky – a world I never knew existed until penning this write-up. Before looking at Troitzky’s ideas in more detail, let’s return to the problem facing IM Welling after move 68 – and lest we were to conclude that Welling failed just because he is ‘only’ an IM, my Ennis colleague IM Petr Neuman told me that strong GM V. Babula had had a similar position against him not long ago and couldn’t do it either – we can quickly discern some general principles that illustrate why it is so difficult to bring home the win:-
1. Black has to be confident he can calculate the mating solution before moving the blockading Knight in for the kill since as soon as he does so he is committed.
2. Black has to take into account what will happen after the pawn queens. If this move delivers check or attacks the mating piece, then the win has gone.
3. It is hard to steer the enemy King just using King and Knight. From point 2 above, we can deduce that depending on which file the remaining pawn is on, the solution differs. This leads us to the concept of the Troitzky Line
If the pawn is blockaded on or before this line then the attacker will win on best play. If the pawn is blockaded after this line then the board can be divided into two zones – a winning zone and a drawing zone –and the defending King, if not confined by the opponent, should aim to get to the drawing zone to secure the half point.
This is not the complete story however. Some of the Troitzky wins take more than 50 moves on best play – the longest is 115 moves! If we were to create a 50 move Troitzky line it would look like this
Apparently a win can also be achieved in 50 moves or under with a pawn blockaded on b2/b3 or g2/g3 in about 99% of instances. For the present game I’ll begin by exploring a few ideas worked out with my colleague Rory Quinn on the way back from Ohrid during a lengthy flight delay at Budapest airport, that give a flavour of what is involved.
We set up a checkmate with the defending king on a1, then working backwards we found that one solution for the game begins with the following starting position
The Knight could equally well be on d3 as e2. Checkmate can then be delivered thus 1.Ka1 Nc1 2.Kb1 Na2 3.Ka1 Nf5 4.h6 Nd4 5.h7 Nc3 6.h8=Q Nc2#
Well this is just dandy if the White King cooperates by sitting on a1, but what if instead of a pawn move White plays at some point Kb1? Can Black apply corrective action to his plan in mid flow? In a word yes, eg; 5.Kb1 Ne2 6.h7 Nac3+ 7.Ka1 Nd4 8.h8=Q Nc2#. The King move at other points in the sequence is solved the same way. Given this is so, the crux must surely be how the attacker forces the defender into one of the starting positions to begin with, which may be easier or harder to do depending on whether or not the defender realises the danger in the first place. In my defence in the game I opted to run near to a corner without voluntarily entering the corner itself – looking back through the final 20-odd moves of the game, Welling never achieved either of the starting positions against me, though that was more by luck than judgement on my part. Rory reckoned that it would be best for the defending King to stay in the middle of the board for as long as possible until forced towards the corner by the attacker. The only caveat here I can see is to avoid straying too close to the blockading Knight from where it could passively influence the position.
We didn’t draw any conclusions on what other viable starting positions existed in addition to the two we found. Nor did we look at which is the best location for the defending King to run to – in the game I opted for the vicinity of a1, but it would be interesting to see how running to the vicinity of either a8 or h1 would change the solution. To try and shed some light on this, the final part of this discussion describes what I found out when I ran the position through the Nalimov Tablebases. Those whose interest has been piqued by the above can also check out two articles by GM Karsten Mueller [5,6] that give more theory and a lot more examples of solutions to Troitzky Endgames from different positions.
The Nalimov Tablebases give the solutions for all chess endings with three to five men and for some with six men. The one drawback with them is that they do not currently take account of the 50 move rule. There are several implementations available on line, the nicest of which is probably at the K4IT website [7]. The outcome of the exercise was illuminating to say the least. Some of the general points which emerged can be summarised as follows:-
1. From the starting position after 68…Nxg7 the win, on best play takes 90 moves;
2. If it were Black’s move in the same position, the win would only take 31 moves on best play; 3. On worst play by the defender, best by the attacker, it’s over very quickly – 69.Kh4 Kf4 70.Kh3 Kf3 71.Kh2 Kf2 72.Kh1 Ng4 73.h6 Nf5 74.h7 Ng3#.
3. On worst play by the defender, best by the attacker, it’s over very quickly – 69.Kh4 Kf4 70.Kh3 Kf3 71.Kh2 Kf2 72.Kh1 Ng4 73.h6 Nf5 74.h7 Ng3#.
4. The win on best play does not respect the 50 move rule, so the position for all theoretical, if not practical purposes, is a draw. Initially I thought otherwise, since mid-way through the solution, the Knight blockading on h6 moves away allowing h5-h6, before the other Knight moves in to renew the blockade on h7. However if the purpose of this finesse is to zero the 50 move clock then it ends up being futile as the pawn advance h5-h6 occurs on move 101 and it is 55 moves later before it advances any further. Probably the purpose is to set up the possibility of Samma’s mate (see 7 below);
5. The forcing line drives the defending King around the whole board clockwise, before finally delivering mate in the vicinity of h1;
6. There are at least six instances where the attacker has to find the only move, the point being that otherwise the defending King will be able to temporarily escape, entailing threefold repetition as it is being rounded up again, before mate can be delivered;
7. The crux of the solution is where the defending King is driven close to the blockading Knight, so that it has fewer flight squares. This Knight then has to be left en prise, the point being that if the defending King captures, Samma’s mate follows. Since the Knight is poisoned, the King has to move away and gets trapped on the h1-h5 file, which directly leads to the final mating sequence. The forced win is as follows – it’s not a unique line – there are some transpositions possible along the way:- 69.Kf3 Ke5 70.Ke3 Kd5 71.Kd3 Ne8 72.Kc3 Ke4 73.Kc2 Kd4 74.Kd2 Nd6 75.Ke2 Nc4 76.Kf2 Kd3 77.Kf3 Ne5+ 78.Kf4 Kd4 79.Kg5 Neg4 80.Kf4 Kd5 81.Kf3 Ke5 82.Ke2 Ke4 83.Kd2 Kd4 84.Kc2 Ne5 85.Kb3 Nd3 86.Ka4 Kc4 87.Ka5 Ne5 88.Kb6 Kb4 89.Kb7 Kb5 90.Kc7 Kc5 91.Kb7 Nd7 92.Ka7 Kb5 93.Kb7 Nf6 94.Kc7 Kc5 95.Kb7 Nf7 96.Ka6 Kb4 97.Kb6 Ne5 98.Ka6 Nc4 99.Kb7 Kc5 100.Kc7 Kd5 101.h6 the pawn moves 101…Nh7 102.Kd7 Nb6+ 103.Kc7 Kc5 104.Kb7 Nd5 105.Ka7 Kc6 106.Ka6 Nb6 107.Ka5 Kc5 108.Ka6 Nc4 109.Ka7 Kd6 110.Kb7 Kd7 111.Ka7 Kc7 112.Ka6 Kc6 113.Ka7 Nd6 114.Kb8 Kb6 115.Ka8 Kc7 116.Ka7 Nb7 117.Ka6 Kc6 118.Ka7 Nc5 119.Kb8 Kd7 120.Ka8 Kc8 121.Ka7 Kc7 122.Ka8 Kb6 123.Kb8 Nd7+ 124.Kc8 Kc6 125.Kd8 Nb6 126.Ke8 Kd5 127.Ke7 Ke5 128.Ke8 Ke6 129.Kd8 Kd6 130.Ke8 Nc8 131.Kd8 Na7 132.Ke8 Nc6 133.Kf7 Kd7 134.Kg7 Ke7!
This is the crux, with Samma’s mate if White takes the bait: 135.Kxh7?? Kf7 136.Kh8 Ne5 137.Kh7 Nd7 138.Kh8 Nf8 139.h7 Ng6# 135.Kg6 135…Ke6 136.Kh5 Kf5 137.Kh4 Kf4 138.Kh3 Kf3 139.Kh2 Kf2 140.Kh3 Ne5 141.Kh4 Kg2 142.Kh5 Kf3 143.Kh4 Nf7 144.Kh3 Nfg5+ 145.Kh2 Kf2 146.Kh1 Ne6 147.Kh2 Nf4 148.Kh1 Kg3 149.Kg1 Ng2 150.Kf1 Kf3 151.Kg1 Ne3 152.Kh2 Kg4 153.Kg1 Kg3 154.Kh1 Kf2 155.Kh2 Ng5 156.h7 the pawn moves again (156.Kh1 Ng4 157.h7 Ne4 158.h8Q Ng3# 156…Ng4+ 157.Kh1 Ne4 158.h8Q Ng3#
What else remains? Well the one last exerecise I tried was to put the actual moves from the game through the Tablebases, to see how well we were both doing. The results are tabulated below:
Move No | W move | Moves to Mate | B move | Moves to Mate | Notes |
69 | Kf3 | 90 | Ne6 | 90 | |
70 | Ke3 | 89 | Ke5 | 89 | |
71 | Kd3 | 88 | Kf4 | 90 | |
72 | Kc3 | 89 | Ke3 | 89 | |
73 | Kb2 | 84 | Nd4 | 88 | |
74 | Ka2?? | 44 | Kd3 | 44 | No. moves to mate halved |
75 | Kb2 | 43 | Kc4 | 43 | |
76 | Ka3 | 42 | Kb5 | 42 | |
77 | Kb2 | 40 | Kb4 | 40 | |
78 | Ka2?? | 29 | Kc4?? | 43 | |
79 | Kb1?? | 32 | Kd3?? | 44 | |
80 | Ka2 | 40 | Kc3 | 40 | |
81 | Kb1?? | 25 | Ne2?? | 41 | |
82 | Ka2 | 40 | Kb4 | 41 | |
83 | Kb2 | 40 | Nd4 | 40 | 83 Kb1?? Or 83 Ka1?? are both 8 moves to mate. |
84 | Kc1 | 39 | Kc3 | 39 | |
85 | Kb1?? | 25 | Ne2?? | 41 | |
86 | Ka2 | 40 | Nc1+ | 40 | |
87 | Kb1?? | 11 | Nb3?? | 39 | 87 …Nd3 is correct, see below |
88 | Ka2 | 38 | Nd2 | 40 | |
89 | Ka3 | 39 | Nb3 | 39 | |
90 | Ka2 | 38 | Nd4 | 40 | |
91 | Kb1 | 25 | Ne2 | 41 |
The table shows that neither of us had much idea of what we were supposed to be doing. Every time I blundered into a bad position Welling immediately gave me all the moves back again. The crux is that at one point (move 87) I blundered into a mate in 11. Now I’m sure that this would be calculable to a finish, but only if the attacker knows that the position has suddenly shifted some thirty moves in his favour – without that knowledge the mistake is surely academic.
For the record, it would go thus:- 87…Nd3 88.Ka1 Kc4 89.Ka2 Kb4 90.Ka1 Ka3 91.Kb1 Kb3 and we’ve reached one of the starting positions analysed by Rory and I above.
Notes
[1] Cassidy – Welling, Saint Vincent 2005
http://www.icu.ie/games/display.php?id=11841
Welling told me that it has since been published as an illustrative game in a German book on the c3 Sicilian – I’m guessing that it’s Zeller’s book on Murey’s Antidote (1.e4 c5 2.c3 b6) but I so far haven’t been able to confirm this.
[2] Martin A. (2004) Shopping for a Tombstone Part 1
http://www.jeremysilman.com/chess_bits_pieces/110103_blackmar
_dmr_gmbt.html
[3] Martin A. (2004) Shopping for a Tombstone Part 2
http://www.jeremysilman.com/chess_bits_pieces/110103_blackmar _dmr_gmbt_2.html
[4] Lane G. (1995) The Blackmar-Diemer Gambit. Batsford. p.73
[5] Mueller K. (2003). The Damned Pawn.
[6] Mueller K. (2003). The Damned Pawn 2.