Lloyd Retallick v Chris Purry
Cornwall v Somerset 2013/14 (Bd.6)
Lloyd Retallick has supplied detailed notes from his deep analysis of this complex game from the 2013 Somerset match
The Tarrasch Defence
A Chess opening principle is to favour developing short range knights before long range bishops – other things being equal – to preserve flexibility. Here the pawn on d5 doesn't need defending and it's too early to commit the light-squared bishop to e6.
9.Bg5 would likely transpose into main line theory. On e3 the bishop creates tension before black has castled.
Since the pawn is poisoned black goes for complications.
12.b4 Nxe3 13.Qd3 Qd8 14.fxe3 gives white an extra pawn and a better position – the doubled pawn can be readily traded at an opportune moment and white's pieces are well placed. Any attempt to open the Queenside by either a5 or b6 can be repelled by b5.
14.e3 would consolidate a small but thematic advantage, aiming for long term pressure against the isolated QP. Here the retreat Nf6 plays into that plan, whereas Ne5 attempts to complicate.
Whoops – I didn't miscalculate the consequences of this move, rather completely overlooked it. That's careless, given the red flags inherent in the position: 1) the proximity of the Ng4, targetting the f2 and e3 squares & 2) my evalated evaluation of the position – never look a gift horse in the mouth.
Taking back the initiative and leaving residual pressure against the black queenside at the cost of the exchange.
What's wrong with 15.Rc2, keeping the rook on the c-file? I remember rejecting it during the game, but have forgotten my reasoning. Initially during my post-game analysis I concluded it failed to a knight fork from e3, e.g. 15...Be3 16.fxe3 Nxe3 17.Nxd5 (17.Qd3 Nxc2) 17...Nxc2 winning for black. But white doesn't have to take the bishop – 16.Nxd5 Qxd4 17.Nxe3! and white emerges a pawn up. 15...Nxf2 is an alternative way for black to try and exploit the dark square weaknesses, but after 16.Rxf2 Be3 17.Nxd5 Bxf2+ 18.Kxf2 white is fine. Thus, subject to computer analysis showing better lines, 15. Rc2 was the best alternative. What this emphasises though is that in (human) practice we are continually relying on imperfect analysis and, under time pressure, our games are typically littered with inaccuracies.
After this sequence we are left with an extraordinary position, where I count no fewer than 10 candidate moves for black: 18...a6, 18...a5, 18...b5, 18...f5, 18...h5, 18...Kh8, 18...Nf6, 18...Nxf2 and 18...Qe5). To understand these moves, it helps to work through the various latent motifs, e.g. Ne7+ forking c8 and thus preventing an immediate Rc8 to attack Nc1 which is itself defending Pe2; a fork is also possible on c7 hitting the Rook and Bishop, the latter defending the Ng4; and more. Another pertinent consideration is what endgame to target – for example is R vs B with 3 kingside pawns each winning or drawn? (answer below)
Given that both players have used considerable time getting to this point, simplification is a practical choice for black. An alternative approach, playing for a middlegame showdown, is exemplified by the sample variation 18...b5 19.Qa6 Rad8 20.Nf4 Bc4 when white is invited to nab the Pa7 but will struggle to survive in the centre.
Since the Queen is about to be evicted from b2 anyway, bringing forward Qb6 is sensible here.
White can claim a small victory here, maintaining pressure on the queenside pawns and thereby making black's task a little harder.]
A little 'shadow boxing' of no great consequence.
Black initiates a series of Knight moves here, in line with his quest for a simplified endgame. The question mark is not so much for this move, rather the whole sequence, omitting a prophylactic kingside pawn move to rid black of any back rank mate threats.
29...Nb5 was the move I feared during the game.
The a6 pawn can't be defended because of the back rank mate threat.
That's the point – the bishop retreat simulanesouly blocks the check and frees the a6 square as a safe haven for the pawn.
Although black could carry on risk free for a while if he chose to, the position with the extra pawn is an easy draw. Without that pawn (to address the question posed above) it should be winning. In his book 'Endgame Preparation' John Speelman shows that with pawns on f2-g3-h4 against pawns on f7-g6-h5 black, with the Rook, has an easy win if white's bishop is of the same colour as his pawns. With the opposite coloured bishop, as here, he reckons after several pages of analysis that black is probably winning with best play, but not conclusively so!
Notes by Lloyd Retallick