For many years 3.Bc4, a move which Keres burned a lot of midnight oil over, was considered White's best, but today “it seems” don’t offers White no advantage. Its very messy (and unnecessary !) and honestly I think Keres busted with this.
The move develops a piece, prepares to castle, and threatens an immediate assault on f7. It is even listed as the Main Line against the Latvian Gambit in Nunn's Chess Openings. It leads to some of the most complex situations to be found in the Latvian.
Yes indeed, there are many reasonable replies (f.i. 3..Nc6 or even transposing to Philidor counter gambit with 3..d6 - remember 1.e4 e5 2.Nf3 d6 3.Bc4 f5 -), but the more usual answers are 3..fxe4 4.Nxe5 Qg5 and 4...d5 (Svedenborg variation):
4...Qg5 ("Poisoned pawn" line) was the first idea against 3.Bc4; it produces some of the most hair-raising variations known in chess. After Main Line 5.d4 Qxg2 6.Qh5+ g6 7.Bf7+ Kd8 8.Bxg6! Qxh1+ (not taking the Rook has its drawbacks) 9.Ke2 and then it has been demonstrated only few time ago UNIQUE correct move is
9... Qxc1 10.Nf7+ Ke8 11.Nxh8+ hxg6 12.Qxg6+ Kd8
13.Nf7+! (the point of the interpolation of this move before 14.Nc3 is to deny Black the opportunity to rapidly relocate his Queen by 13..Qf4; thus, if 13.Qxg8 Qf4 14.Nd2 d6 15.Rg1 - or 15.Ng6 Qg4+ 16.Kf1 Nd7 17.Qf7 Qh3+ 18.Kg1 Qh6 19.Nxe4 c6 unclear - 15...Qf5 - also 15..Nc6!? - 16.Rg5 Qf6 17.Nxe4 Qf4 very unclear in F.Tejero (2165)-A.Melchor (2133), Spain Catalonia team ch. (9), 2006 but Black won in a few moves by a great mistake of first player ) 13..Ke7 14.Nc3! Qxc2+ 15.Ke1 d6 ((15..c6? 16.Nd6 etc. threatening mate in two is loser) 16.Nd5+ Kd7 17.Qxg8 ( slight White advantage, NCO )
17.. e3! (striving to get the Queen back into play. As well as this 17...Qxb2? 18.Rd1 e3?! was known, but after f.i. 19.fxe3 Qa3 Kosten recommendation follow 20.Qg4+ Ke8 21.Qh5!) 18.fxe3 ( If 18.Nxe3 Qxb2 19.Rd1 Nc6! or 18.Ne5+!? dxe5 19.Qf7+ Kd6 20.Nxe3 Qe4 21.Qxf8+ Kd7 as Elburg-Voliani, corr. e-mail friendship game, 2001 still looks unclear to Kosten, although White can force a draw if he desires) 18..Be7 19.Ng5 (19.Qg4+ Ke8 20.Qxc8+ Kxf7 is more confuse O'Connor-Domingo, cr. e-mail LADAC thema prel., 2006) 19...Na6 20.Qxe6+ Kc6 and:
- A) 21.Qg6 Qc4 22.Nxe7+ Kb6 23.Nf3 Bh3 24.Nd2 Qe6 25.Qxe6 Bxe6 26.Ng6 Bf7 (26...Nb4!?) 27.Nf4 Rh8 (27...Nb4) 28.h3 Nb4 draw in 53, Rouzaud-Rosenstielke, cr. e-mail 5th. LG World Ch. sf. B, 2004/05
- B) 21.Nxe7+ Kb6 22.Qb3+ Qxb3 23.axb3 Nb4 24.Kd2 Bd7 25.Rf1 Rh8 26.Rf2 c5 27.dxc5+ dxc5 28.e4 Rh5 and draw again in 36 moves, Koudelka-Rosenstielke, cr. e-mail 5th. LG World Ch. sf. B, 2004/05, but we think White can improve a bit the game if he tries 22.Nxc8+ Rxc8 23.Qxc8 Qxh2 - necessary if Black Queen can threaten annoying lateral checks - 24.Qe6 Qg1+ 25.Ke2 Qxa1 26.Qb3+ or 24.Rd1 Qg1+ 25.Kd2 Qf2+ 26.Kc3 Qxe3+ 27.Rd3 Qxg5 28.a3 as Morcillo-Melchor, cr. e-mail Spain Cup prev., 2012/13 although anyway game ended draw three moves later.
For decades the capture 9..Qxc1 was considered dubious, in these games, extensively analyzed, White didn't demonstrate an absolute forced win, and in fact, both games was draw, BUT White can improve definitively upper hand: HOW ? ….