Tuesday, 31 October 2017

028. Two Lines of a Rook Endgame


White: J. De Waard - North Sea Team Championship, 1998

One of my mates once told me: “if you want to play the Four Knights as Black, play the Petroff”. That's because, as I mentioned here, 3 Nc3 is a common reply to 2...Nf6. Whereas it isn't so much against 2...Nc6. Checking my own OTB games, going back to 1989, I see I've only faced 3 Nc3 six times. In correspondence chess, it's even more of a rarity: just three games in all competitions. Those nine in total continued with 4 d4 (four times), 4 g3 (thrice), and 4 Bb5 (twice). I won just once.

Hmmm. It hardly seems worthwhile studying an opening that you only face once every five years. But still... only one win as Black. How'd you like them apples?! Not very much, in truth. Maybe I should do something about that.

On the other hand, my one win was quite nice. Glek's variation (4 g3) led eventually to a double rook endgame where I held the initiative, and thence to a single rook endgame. The final result turned, after 45...Rb1, on two move 49 positions, both of which I assessed as won for Black:



a) 46 Ra2 Rxb3 47 Ra6+ Ke7 48 Rxh6 Rxd3 49 Kxf5 Rd4;
b) 46 Re2 Rxb3 47 Re5 Rxd3 48 Rxf5+ Ke6 49 Rxc5 Rxh3.

Black ends up with rook and two vs. rook and one in each case. And the second line (with split passed pawns) is indeed won, as the Lomonosov databases now confirm. But the first one (with connected passed pawns) isn't: 50 Ke5 Rxc4 51 Rb6 or 50...Kd7 51 h4 Rxc4 52 Kd5 and draws (Lomonosov again).

Guess it was lucky my opponent went for line ‘b’.


Sunday, 22 October 2017

027. Emotions in Correspondence Chess


White: A. Richards - BCCA DJKO 33/1, 1998

The main point of Georgi Popov's 3...a5!? defence to the Ruy Lopez is to follow up with 4...Na7 5 Bc4 b5, gaining the usual queenside space for Black while driving the Lopez bishop away – and trapping it should it drop back to b3. White's 4 d3 in the game below is therefore quite obliging as it blocks a retreat to the centre. Thus 4...Na7 5 Bc4 b5 (etc) and the bishop is lost.

My opponent in the game below wasn't bothered by that, responding 6 Bxf7+ Kxf7 7 Nxe5+ Ke7 and then 8 Nf7!?.



I don't recall ever seeing a similar idea before: White gives up two pieces on f7 in order to win an unprotected rook on a8 with queen checks. Later on he surprised me again by putting his other bishop en prise to a pawn (21 Bd6).

Going over the game with Houdini now, it seems neither of these surprises is particularly good for White. The engine prefers the bishop and knight to White's rook and two pawns in the first instance. In the second, it suggests that Black simply takes the piece: 21...cxd6 22 Nxd6 Nf4 and then answer 23 Nxc8 with 23...Nxg2!, which looks very strong. Whether I actually considered that, I'm unable to tell, since the game is somehow missing from my cardboard Postal Games 1996-98 folder.

What I do remember is feeling unsettled, unable wholly to trust my assessments throughout the game. In the end, I went completely wrong (31...Ng7?) and lost.

It might be assumed that emotions don't much matter in correspondence chess. On receiving a shock, for instance, you don't have to find a reply right there with the clock ticking. You can let your feelings subside before studying a position... calmly, rationally... two or three days later maybe. But it can be difficult all the same. An underlying sense of unease, that you're not in control, that things are going wrong, can pervade a whole game, and is no less disturbing for being long drawn out.

Or so I have found anyway. Your mileage may vary.


Friday, 29 September 2017

026. A Spectacular Queen Hunt


White: MarcShaw - Thematic tournament, ChessWorld.net, 2014

Hunts in chess are always “spectacular”, aren't they. In this one the white queen rushes out early to h5, snaffles the rook on h8, is driven back by busy minor pieces and seeks shelter on the queenside, only to suffer the indignity of perpetual attack from a lowly bishop, forcing a draw.

Okay, this hunt is not really so spectacular. The initial queen sally is a known theoretical line, and her subsequent harassment forms the basis of Black's compensation for the sacrificed exchange. All the same, it's quite a nice sequence of events: Qd1-h5-h8-h7-h4-a4-a3-a6-a4-a6 and draws.

It also shows that, in the 3 Bc4 variation at least, The Latvian Gambit Lives!, even if Black's path to apparent safety remains quite narrow: 3...fxe4 4 Nxe5 d5! 5 Qh5+ g6 6 Nxg6 hxg6! and so on. Everything else (e.g. 3...d6?!, 4...Qg5?!, 6...Nf6?!) is just good for White.

Since the 3 Bc4 Latvian can – and in fact did – arise via the Calabrese Counter-Gambit, it's a good thing for the latter that this line of the former is acceptable for Black. Regarding which, the game below is theoretically interesting.



From the diagram, the game continued 11...Ne5 12 Nd2 exd3 13 0-0 Nf6 14 Qh4 Qd6, which I think represents best play for both sides.

Instead, GM Kosten – working in, if not exactly pre-computer days, then certainly weaker computer days – gave a few erroneous assessments which are perhaps worth noting:
a) “11...exd3! and Black is at least equal”. Here 12 0-0 is good for White; e.g. 12...dxc2 13 Bxc2 Nf8 14 Qh4 Qxh4 15 gxh4, MarcShaw-Reprimand, ChessWorld.net 2013.
b) “13 f4! looks right, followed by Nf3-g5”. This lands White in serious trouble after 13...Nf6 14 Qh4 Neg4 15 Nf3 Qd6 16 Ng5+ Ke7.
c) 13...Qf6!? 14 f4 Nc4 15 f5 gxf5 “unclear”, H.Hage-F.Van der Muysenberg, correspondence 1995. This should have been met by 16 Nf3! with advantage.

At the time I was a bit worried about 15 Qf4, but now Houdini shows that 15...g5! 16 Qxg5 Rh8 17 Re1 Rh5 is fine for Black. I love computers :)


Wednesday, 6 September 2017

025. Two Terrible Blunders


Black: juliangon - thematic tournament, ChessWorld.net, 2017

I'm currently competing in a Latvian Gambit tournament, all the games to start with 1 e4 e5 2 Nf3 f5. I've played the Latvian on and off over the board with reasonable results, and still bring it out in blitz chess from time to time. However, apart from 3 Bc4 which can arise via the Calabrese Counter-Gambit (1 e4 e5 2 Bc4 f5 3 Nf3), I've not looked at the theory in years, especially the state of the 3 Nxe5 Qf6 4 d4 d6 5 Nc4 fxe4 main line.

Well, I have to say that this seems now to be virtually unplayable for Black, so that anyone who manages to scramble a draw from this position has, I think, done very well. So far I've not managed to do that and have been crushed twice. From the other side, I am (or was) looking at two solid wins as White. My third opponent might well draw as Black (well done!), while I might draw my third game too (hurrah!).

All the same, just ½/3 and having to work extremely hard for it... No, I don't think I'll bother anymore, thanks. If it's really true that The Latvian Gambit Lives, as GM Tony Kosten entitled his second book on the opening (Batsford 2001), it's only with a finger or two keeping the coffin lid from being nailed shut.

My two losses have come against 6 Nc3 Qg6 7 Ne3 and 7 f3 Be7 8 fxe4!. In the latter line, I look like drawing with each colour after 8 Nxe4. I'm a pawn up in the endgame as White, but it will probably resolve itself to three vs. two on the kingside, which won't be an easy win. Whereas two pawns up and three vs. one should be simple enough (as has occurred in another game after 3...Nf6 4 Bc4 Qe7 5 d4 Nc6 6 Nc3!).

In my final game as White, I went for Bronstein's 6 Be2, both for a change and because I've never been particularly happy facing this myself. Indeed, it all turned out very well for me, until my final move...



Here I intended 30 gxf6 Rxe5 31 fxg7 Kxg7 32 Ne4 and wins, while the tricky 31...Re1!? is refuted by 32 Bxh7+! Kxh7 33 g8Q+ Kxg8 34 Rg3+ (saving the rook) and 35 Qxe1. Instead, I apparently entered 30 Bxf6??, which is a terrible blunder since 30...Re1! now works and even wins for Black; i.e. 31 Qxe1 Qxf3+ 32 Kh2 Qxh3+ 33 Kg1 c4+ etc. As I wrote in an earlier post: “human stupidity can never be prevented entirely”.

Fortunately, I only discovered my mistake when glancing over an email notifying me that my opponent had resigned, which of course was a terrible blunder too.

Naturally, I was both surprised (at my mistake) and pleased (that it hadn't immediately cost me the game), if more surprised than pleased. In fact “shocked” and “relieved” would be more to the point there. But at least it was first thing in the morning rather than last thing at night. Otherwise I might have been tossing and turning for hours.


Thursday, 31 August 2017

024. Rook vs. Bishop and Pawn


Black: D. Nightingale - John Elburg anniversary tournament, 2001

Petroff's Defence, or the Russian Game, is an extremely peace-loving opening. By choosing a symmetrical move, Black is as much as saying: whatever you do, I'll follow you; I'm a meek and mild-tempered fellow.

“If White responds to Black's peace initiative, then the moves flow like water off a duck's back – a dozen exchanges, and the clocks are stopped. Draw.

“But White has only to show a touch of arrogance and obstinacy, and go in for an open fight, when Black immediately replies blow for blow, and then for a long time it is difficult to say who is attacking and who is defending.

“But that rarely happens. More often White complies; one exchange, then another – and then the judge rushes to the table....


Such was Bronstein's witty description of the Petroff back in 1970. Nowadays, even if the draw percentage remains high, White is generally unwilling to concede one quite so easily. Thus Black is obliged to demonstrate the opening's fighting qualities, as in the game below.

This followed a then topical variation with 3 d4 and 7...Qh4, which seems now to be a theoretical backwater, almost a historical footnote. White prefers 3 Nxe5 five or sixfold over 3 d4 – and 3 Nc3 too is seen nearly three times as often. Even when 3 d4 is brought out, subsequent play hardly ever gets as far as 7...Qh4, which features in just 0.002% of recent Petroff games.

Nevertheless, an “open fight” duly ensued, with both sides pushing their pawns forward vigorously, and a wholly correct exchange sacrifice led to a material imbalance of rook vs. bishop and pawn.



In this position Black's bishop (hiding behind the white d-pawn) safeguards his king, and the advancing kingside pawns should give him sufficient compensation, whether or not queens come off. In the end, all the major pieces left the board, and the g- and h-pawns were traded for White's c- and d-pawns, to leave a pure rook vs. bishop endgame with three vs. two on one flank.

As Averbakh showed many years ago, Pa5, Pb6, Bc6, Kb7 is a fortress even without the black c-pawn. So all that remained was for the metaphorical judge to rush to the table.


Saturday, 19 August 2017

023. Winning By "House"


White: J.J. Wagenbach - friendly thematic, 1998

Twenty years ago (really?!) we used to play friendly thematic games to test out variations of the Wagenbach. One line which saw quite a bit of action was 4 Bc4 h4 5 Ne5!?, whereby White tries to improve on 5 d4 g5 6 Ne5 (as in Game 1) by attacking f7 before Black has played ...g7-g5, so that a later ...Qf6 does not also defend the rook in the corner. After a while, 5...Nh6 6 Qh5 Qe7! was worked out to be fine for Black and the line was more or less abandoned, but not before we'd had several interesting games. The one below is one such.

However, I'm not showing this game for any reason of Wagenbach "theory", nor subjecting it to modern computer analysis, but rather because of a possible position at move 32. Here White might have played 32 Rh1!? and claimed a win by "house"; i.e. by filling in a complete line on the board – and, very oddly, on the h-file.



Obviously, there's no such rule in chess, so János opted for something more sensible in 32 Bg3 (and only subsequently went wrong). All the same, I might as White have played 32 Rh1!? anyway, just because.


Thursday, 27 July 2017

022. Queen vs. Rook, Bishop and Knight


White: N. Christophe - ICCF thematic tournament, 2000

Back in my very first post, I mentioned that my pet lines include the Calabrese Counter-Gambit (2 Bc4 f5!?), Wagenbach King's Gambit (2 f4 exf4 3 Nf3 h5!?), Traxler Two Knights (2 Nf3 Nc6 3 Bc4 Nf6 4 Ng5 Bc5!?), and Schliemann Ruy Lopez (2 Nf3 Nc6 3 Bb5 f5). Subsequent posts have featured two or three of each of those so far – with the exception of the Traxler (aka Wilkes-Barre Variation), so here's one of those now, which also fits into the ongoing Material Imbalances theme.



In the diagram position, the titled piece configuration is only temporary. Black can quickly regain the e4-pawn for a start, while another game (Holzner-Tait) soon reached Queen and Two Connected Passed Pawns vs. Two Rooks and Bishop – which I lost. As it happens I seem to be the only person to lose (and lose twice) as Black from this position, though the other games in the database hardly represent best play either.

Black has tried a variety of moves here: 20...a6, 20...Qh4+ (going for the pawn straight away), and the anodyne 20...Kb7 (my choice). That move looks bizarre to me now. White is getting close to consolidating the kingside and I'm bothered about losing the a-pawn?! Obviously Black should be trying to generate counterplay and, to that end, might consider either 20...h5 (intending ...h5-h4, when 21 Rh1 Qe5 has at least disrupted White's artificial castling) or 20...h6 (probably stronger), followed by ...g7-g5. Houdini marks the latter down as only slightly better for White.

As for 20...Kb7?!, someone else did win with this move (with a bit of help), and in Holzner-Tait 28...h5 (rather than 28...g5?! etc) 29 a4 Qg2 might have offered Black more chances. But in the game below I lost without any fight whatsoever.

In his recent book, Black Gambits 2 (Quality Chess 2012), GM Boris Alterman stops at 18 Nf3 and writes: “Black has no more attack, and White's rook and two pieces will eventually overpower the black queen.” Certainly that proved to be the case in both my games, though whether it's entirely inevitable is not so clear. All the same, the standard 7...Qh4 is probably the right way to go.