View Full Version : FM guidelines

11-10-2007, 10:39 PM
Some of the usual book guildelines can be misleading, and some useful rules of thumb don't appear in most books. So here is some correction and addition:

With R + minor v R + pawns, do not follow the rule "exchange pieces when material up". Rooks are much better at attacking than defending, more than other pieces. The minor piece can defend a pawn or entry square quite cheaply, allowing the R to go on the attack. This will tie the enemy R to passivity, so the R, minor and K can coordinate an attack. A good example is Tarrasch vs Chigorin Ostende 1907 (http://www.chessgames.com/perl/chessgame?gid=1036654).

Bad bishops are sometimes better than good bishops, for a related reason: they can defend pawns cheaply, while the opponent will have to tie down a R. I won't repeat what I've already said in When bad bishops are good (http://www.chesschat.org/showthread.php?t=6403).

Sometimes the bad B is better than a N. E.g. in a KID formation with Bd3 Ps c4,d5,e4 v Nc5 Ps c7,d6,e5, the B should often retreat to c2 rather than allow an exchange. From there, it hinders ... f5, while if it is exchanged, Black will benefit even more from ...f5 and opening lines.

The rule to place one's pawns on the opposite colour of one's B with same-coloured Bs should not be carried too far for another reason: they can ruin winning pawn breaks. See Dvoretsky's comments on Sveshnikov vs Kasparov, Minsk 1979 (http://www.chessgames.com/perl/chessgame?gid=1069793); also Bronstein criticized 37. f4 in Taimanov vs Kotov, Zurich 1953 (http://www.chessgames.com/perl/chessgame?gid=1084367) for taking away the options of e4 and g4. Conversely, note 47... f6 instead of ... f5 in Thomas vs Alekhine (http://www.chessgames.com/perl/chessgame?gid=1012322), preparing the decisive break with ... e5.

In a typical K-side v Q-side game, the exchange of Qs favours Q-side attacker because there is less chance of the K-side attack ending in mate with Qs off. So White should like a Q exchange in many KIDs and the QGD Exchange variation (http://www.chesschat.org/showpost.php?p=153028&postcount=2), while Black should like it in many Sicilians. (NB: here I am talking about the normal situation with both Ks castled short).

With B v N, don't follow too blindly the rule that Bs like open games and Ns like closed ones. But in the opening, the side with a Ns often has a development advantage, so the best strategy is for that side to open the game, make use of the tactical abilities of the N, and force pawn moves that create permanent outposts. So the side with the Bs should seek to stabilize the position, catch up in development, then open up the game when ready, so the bishops can display their strength. And if the Ns avoid opening with pawn exchanges, the side with the Bs just gains more and more space with pawn advances. Conversely, the Ns often can't just sit back and wait behind a closed position, because this is just waiting to be crushed. Watson's books have good examples.

An outpost on an open file is best supported by two pawns. Then an exchange of a piece will result in a protected passed pawn. Also, if only one pawn supports a piece, then there is a danger of destroying the guard tactics unless this pawn is defended by another pawn.

Don't underestimate the B-pair. There are some teachers who rail against the "dogmatism" of the B-pair, but replace it with their own dogma of the precise equality of B and N. As Watson pointed out, this is a remarkably categorical claim for two pieces with completely different moves and history.

11-10-2007, 11:31 PM
That is a great posting Jono. The two books by Watson on the midle game are very very good.

12-10-2007, 12:00 AM
That is a great posting Jono. The two books by Watson on the midle game are very very good.
Thanx David. Yes, Watson's books were a significant addition to chess literature, the best in a few decades most likely, and fully deserve all their awards. My only quibble is his disrespect for the older masters, but he certainly brings out the way the great players are not rule-bound.