The London System: Choose a Plan, Not Just a Setup

The London is often taught as a fixed sequence: Bf4, e3, c3, Nd2, Nf3, Bd3, and castle. The sequence works until Black plays ...c5 and ...Qb6, when the queen attacks b2 and the position asks which pieces are actually defending it. In this game the queen's knight has gone to d2 before f3, and that detail decides the answer. With the knight on d2 the pawn on e3 is ready to become e4 later, and the queen can go to b3 without leaving anything loose.

Look at the position after 5...Qb6. A good defence of b2 has to pass three tests: it must not cost a tempo, it must not loosen the c3 pawn or the long dark diagonal, and it must not take the b3 square away from the queen. Most natural-looking defences fail at least one of them. 6.Qb3 passes all three and adds something of its own: it offers a trade that helps White. After a capture on b3 the a-pawn recaptures, the a-file becomes half-open for the rook with a7 as its target, and the doubled b-pawns are ready to advance. If Black keeps the queens with 6...c4, the retreat 7.Qc2 keeps a small edge, so the offer costs nothing.

Black chose 6...Bf5, a natural developing move, and White took the chance to fix the queenside first: 7.dxc5 and only then the recapture 8.axb3. Black must spend a move on ...e5 and another on ...Bxc5 to win the pawn back, and each b-pawn advance then comes with a threat: 10.b4 hits the bishop, 11.b5 hits the knight. By move 12 Black's minor pieces have moved several times and White has gained a queenside space advantage from a roughly balanced opening position.

When the e-pawn attacks the bishop on move 8, notice where it goes. 9.Bg3 stays on the h2–b8 diagonal and presses e5. If Black spends a move on 9...Nh5 to attack it, 10.b4 simply keeps the extra c5 pawn, and ...Nxg3 hxg3 hands White a half-open h-file on top; the engine rates that about two pawns better for White. A retreat to g5 would leave the diagonal and invite ...h6 with another tempo. In the London the bishop's diagonal is worth more than any single pawn, and the retreat square is a decision, not a reflex.

The final position after 20...Bf7 is level in material, with White's pieces a little more active and a knight on a5 on the edge of Black's queenside. The engine rates it as level. The lesson is not that 6.Qb3 wins; it is that a queen trade offered from a good structure keeps the initiative, and that the move order Nd2 before Nf3 is what made the offer safe.

When Black plays ...e6 and ...Bd6, the offer is to trade White's best piece for a bishop Black is happy to exchange. The engine rates three answers about equal: keep the bishop with 5.Bg3, take on d6 yourself, or develop with 5.Nbd2 and let Black take on f4, when the e-pawn recaptures and White gets a firm grip on e5 at the price of doubled f-pawns. They are three different games, and the choice is a matter of plan. This chapter follows the first one, because it produces the clearest lesson about what a lost bishop can buy.

Keeping the bishop invites 5...Ne4, a knight that attacks g3 and offers the trade a second time. Retreating again would lose time, so White develops with 6.Nbd2 and lets Black decide. When Black takes, 7.hxg3 is the recapture to remember: the h-pawn, not the f-pawn, because the e3 pawn keeps its neighbour on f2 and the h1 rook now looks down a half-open file. In these structures the file is worth more than Black's bishop pair, and the rook is already standing on it.

The knight on d2 has a second job: it supports e4. After 7...Nd7, a solid but unambitious square, there is no black knight on f6 to hit e4, so the break comes at once, and 8...dxe4 9.Nxe4 attacks the bishop on d6, which retreats to e7 having moved twice. 8.Bd3 and 8.c4 were rated slightly higher; 8.e4 is the most direct way to open the centre while Black is still unpacking the queenside.

Black's 10...f5 gains a tempo on the knight but weakens e5 and e6 and opens the king's own diagonal. Where the knight goes matters: it should stay in the game and keep an eye on d5, and it must not step where a bishop and queen can pick it off. By move 12 White is clearly better, and the rest is a demonstration of levers: the king goes to the safer wing, the knight lands on e5, and f4, Bb5+ and g4 arrive one after another against a king that never found shelter.

The final position after 22...Be6 is level in material. Black's king stands on f7 in the middle of the board, both white rooks are ready, and the bishop on b3 and the queen on e5 both bear down on d5. The engine rates White decisively better. The lesson is that in the London the bishop is a means, not an end: if Black spends three knight moves to remove it, take the file and the tempi and open the centre.

The first five moves of this game reach the same kind of position as chapter one: a London bishop on f4, pawns on c3, d4 and e3, and a black queen arriving on b6. The difference is where the knights stand. This game used the 2.Nf3 order, so White has developed the king's knight to f3 instead of the queen's knight to d2: b1 is still occupied and g1 is empty. Two knights have swapped roles, and that changes the correct answer to the sortie.

Go back to the three tests from chapter one: a defence of b2 must not lose time, must not weaken the queenside, and must not take the b3 square away from the queen. 6.Qb3 still passes them. The difference lies in what happens after 6...c4, a reply Black has in both move orders. In chapter one, 6...c4 would have been met by 7.Qc2 with a small edge, because b1 was empty: after 7...Bf5 White could take on f5, since ...Qxb2 is answered by Qb1. Here 7...Bf5 hits the queen and 8.Qxf5 loses to 8...Qxb2, which forks the rook on a1 and the knight on b1; the queen ends up on c1 after three moves, and Black has gained space for free. The engine prefers the modest 6.Qc2 from the start, by about a third of a pawn. That number is small, and this chapter is not about a blunder. It is about a third of a pawn that grows.

Look at the position after 12...Rfc8. Black is fully developed, the c4 pawn cramps White's queenside, and the plan ...b5–b4 is two moves away. White's only way to change the character of the position is e3–e4, and it needs preparation: the bishop off the e-file, via d1 to c2, where it also supports the push, and a rook on e1 or a knight on e5 to contest the centre first. The game shows the arithmetic. The manoeuvre takes three moves, the b-pawn takes two, and when 16.e4 finally comes the engine already prefers a different move. Every white move from here is reasonable; each one is also a little late.

Notice how the drift accumulates: 15.Re1 instead of 15.Ne5, 16.e4 instead of 16.Bxf5, 18.Bd2 instead of 18.Bg3. None of these costs more than a fraction of a pawn, but they push Black's edge from about a third of a pawn towards a full pawn by move 20, before it settles at about half a pawn at 21...Qa5, the end of the shown line. At that point the b-file is open for Black's rooks, the c4 pawn is still there, and the queen hits c3 and a2. Material is level, and the slide continued afterwards.

The lesson is the title of this guide. The setup was the same in chapters one and three; the plan could not be. With the knight on d2, 6.Qb3 is an offer that works. With the knight on b1, the same move hands Black the initiative for nothing, and the correct London answer is the quieter queen move and a faster e4.