Berlin Defense

3...Nf6 develops the king's knight with an immediate attack on e4 and bypasses the customary 3...a6. It is the principal alternative to that move and has a long-standing reputation for solidity, serving at elite level as a dependable way to neutralize 1. e4. Black does not need to fear 4. O-O, because taking on e4 is not a genuine material gain: after 4...Nxe4 White uses the e-file, for instance 5. Re1 hitting the knight with e5 standing behind it, to recover the pawn.

The critical test is 4. O-O Nxe4 5. d4 Nd6 6. Bxc6 dxc6 7. dxe5 Nf5 8. Qxd8+ Kxd8, the Berlin endgame; with 9. Nc3 Bd7 it is catalogued as the Berlin Wall. Queens disappear early, Black can no longer castle and carries doubled c-pawns, but the bishop pair offsets White's sound kingside majority, and opening lines to mobilize that majority tends to liberate Black's bishops. The endgame is exhaustively analyzed, and current opinion holds that careful defense secures the draw. Instead 6. dxe5 frequently ends in a well-known repetition, while 5. Re1 sidesteps the queen trade and is judged marginally more promising for White, though still drawish. Black's own deviations include 5...Be7, the Rio de Janeiro Variation favored by Emanuel Lasker, and on move four 4...d6, the Improved Steinitz Defense that can merge with 3...d6 lines, or 4...Bc5, the Beverwijk Variation, also reachable via 3...Bc5.

White often avoids all of this with 4. d3, the Anti-Berlin. The pawn protects e4 and also revives the threat to e5: with e4 covered, Bxc6 followed by Nxe5 can no longer be answered by a ...Qd4 fork. The cost is that d2-d4 later takes two tempi. Black most often replies 4...Bc5, restoring the tactic since ...Qd4 would then also eye mate on f2, or simply supports the pawn with 4...d6. Wilhelm Steinitz used 4. d3 with great success. Other tries are 4. Bxc6 and 4. Nc3, which transposes to the Spanish Four Knights; 4. Nxe5, the Halloween Attack, is unsound and Black scores heavily against it.

Baron von der Lasa published early analysis of 3...Nf6 in 1843, and the name was in German use by 1848, credited to Carl Jaenisch. Lasker and his contemporaries played it before it faded, and Vladimir Kramnik revived it in the 2000 World Championship, drawing four of Kasparov's white games with the endgame. It has since become a fixture of top-level play, adopted by Magnus Carlsen and Viswanathan Anand in their 2013 and 2014 title matches. The engine rates the position as roughly balanced, and in Lichess games White scores somewhat better.