(1.e4 e5 2.Nf3 f5 3.Nxe5) 3...Nf6 was played 216 times in the Melchor database, score 60% for White. The overwhelming majority of these games, I'd estimate 75% or more, were corr. games.

Only in 49 games found White 4.Bc4 (57% score); 34 of those were corr. games (score 60%). Black replies 4...Qe7, the Melchor database contains 129 games, often in a different move-order (score 50%). The right response is 5.d4, in 118 games (53%).

Black answers 5...Nc6. The database has 78 games (51%). The critical move 6.Nc3! appears in 20 games (score 58%). So the probability that these White opponents (mainly corr. players) reached this critical position was 5.3% - and much less if you include cases with other third moves.

Black can try 6...fxe4 (14 games; 77%), hoping that White doesn't find 7.Ng4! (in those 14 games, no one did). Or the untested 6...d6? 7.Nf7 d5; or the best 6...Nxe5! 7.dxe5 Qxe5 8.0-0 fxe4 (22 games, 62%). In the latter case, 10 White players found 9.Nd5. As we have seen, even this position isn't hopeless, as in +-, but I admit that it is +/- (one pawn behind without compensation).

In today's corr. play, to end in this painful situation and have to defend it for months is no fun. So I'll avoid it in corr. chess. But a chance of 2% of pain vs maybe 30% of White failure [= slight plus or more for Black] and else "normal life", well, I have done stranger things with Black. 

Stefan Buecker wrote on 07/14/11 at 21:51:22:

This Latvian gambit is just like Portuguese variation in Scandinavian. Very hard to play in corr. chess, but yet very pleasant OTB. White can walk straight into the trap with every dubious move. 

MNb wrote on 07/13/11 at 23:15:28:


The Latvian has a reputation as being both tactically and positionally suspect. That reputation is fairly well deserved, but only if White knows his stuff and plays accurately. When it is sprung on an unsuspecting opponent, its aggressiveness can occasion a surprisingly muted response. Weak players with little or no book knowledge are often reduced to 3 d3, which simply hands the initiative to Black. Players who know little about the Latvian shy away from the most critical lines because they don’t want to be the victim in a game

I personally think Latvian is only += ... if Blak know he is doing .. and specially if you have -2200 FIDE. Of course, is also dangerous in CC, even today WITHOUT machines !? 

g2-g4 wrote on 07/13/11 at 18:03:14:

After 12.d5 Nb4 13.Rf4 Qd7 14.Bf5 Qd8 15.Bxc8 Qxc8 T.Kosten gave 16.Nxd6 cxd6 17.Rxb4 0-0 18.Bd2 Bd8! =
But there are two improvements.
18.Rf4 Bd8 19.Qd3 Ng4 20.Rxf8+ Kxf8 21.Ne4 Bb6+ 22.Kh1 Ne5 23.Qe2 Qg4 24.Qxg4 Nxg4 25.Bf4 Re8 26.Nxd6 (26.Bxd6?= 1/2 Viola,M (2393)-Elburg,J (2350)/SEMI email 2002) Nf2+ 27.Kg1 Nh3+ 28.Kf1 Nxf4 29.Nxe8 Kxe8 30.c4 +=
16.Qe2 1-0 Vegjeleki,A-Voracek,M/cr ICCF WSTT/1/07/2 e-mail 2007 (41) and 1-0 Zielinski,S-Acedo,H/cr LADAC thema prel 2006 (32)
Or was 15...Rxc8 found to be better?



Ops !, g2-g4 you are a tough interlocutor, but this is fine !? and maybe we could to write a book on Latvian gambit !!.

This line started with 14.Bf5 was known on a remarkable game Rublevsky-Maljutin, Jurmala, 1991 ( as you must to know ), and published in the "Informator". In the aboved game Viola-Elburg perhaps 18.Rf4 Bd8 19.Qd3 Bb6+ 20.Be3 Qc7 is better - with some White advantage ( more space ). Tatlow and Ruggeri later on developed some analysis.

16.Qe2!? 0-0 17.Ne3 is certainly more promising and I don't see Black's game improvement.

15..Rxc8 is discredited as you can see in Kosten book.

Nimzowitch suggested 14.Nb6!? axb6 15.Rxb4 and Karl Bething in 1940 told after 15..0-0 (=) it's unclear that White's spacial advantage amounts to anything. I think personally after 16.Bd2 first player chances are a little bit favourable; and so Kosten give 15..Ng4!?

Thus I recommend 7..Nc6!? and 7..Be7!? for instance to avoid 16.Qe2! etc. and even variations with 10.Qxg4 or 10.Qf2!?
Both Diagram moves was developed after Kosten's book was published:



Remember, the Latvian is extremely complicated and has lots of possible lines, many of these leading to promising situations for Black. Many players below master level and indeed some masters are not familiar with the maze of variations that arise from this gambit so it could be a useful weapon in your arsenal and you may be able to score some good victories with Black using it's variations.

g2-g4 wrote on 07/11/11 at 18:16:05:


So called "relatively" classical Main line 1.e4 e5 2.Nf3 f5 3.Nxe5 Qf6 4.d4 d6 5.Nc4 fxe4 6.Nc3 Qg6 sets up probably one of the most encountered position in the Latvian, but now some new ideas has been found as Black. It's important to note second player must be prepared to answer White's considerable array of 7th. moves, but at the moment come on with your suggestion: 7.f3 exf3 8.Qxf3 Nf6 9.Bd3 Qg4 10.Qe3+ Be7 11.0-0. Now 11..Qh5?! is very dubtious; I know Buecker put two exclamations marks once ( and Kosten in his own book an interesting one indeed, !? ), but both are wrong !!. Your analysis above are convincing and even it was played two times by the same Spanish player Jordi Domingo in 2007 and 2010 respectively ... losing both games ... ( f.i. 12.Nb5 Na6 13.h3! Kd8 14.a4! Re8 15.Bd2 etc. Mueller-Domingo, FICGS class M015, e-mail, 2007 1-0, 32 ).

The key question is Black must force a trasposition with 11..Nc6 to the Main line; thus 8..Nf6 9.Bd3 Qg4 10.Qe3+ Be7 11.0-0 Nc6 ( Kosten, "The Latvian Gambit Lives !" p.27 ). However, Black can, with accurate play, a playable position, being perhaps White "best" reply positional 12.d5


     
The true problem with 8..Nf6 is 9.Ne3! ( followed by 10.Bd3 ). I can not write here vaste theory about, but to sum up, White have won 90% of the games ! ( I myself have lost four ).

Instead, if Black try on trasposing to "our" line with 8..Nc6 then he must to face well-known by Latvian players 9.Nb5! Bg4 10.Qc3! already pointed in Kosten's book.

In the year 2006 I myself started on analyzing deeply unexplored move 7..Be7!? ( played by Argentinian IM Perez Pietronave at the end of '90 decade ) and since then, most Latvians funs chose this move as Main Line against 4.d4; 6.Nc3 variation. Usually game continues 8.fxe4 and now 8..Nh6!? ( or 8..Nc6 and 9..Nh6!? ) in the spirit of the Kjell Krantz's idea 7..Nf6 8.fxe4 Be7 9.e5 Ng4; the point in "my" pawn sac line is similar: 10..0-0 combined with ..Bg5-Bh4+/Ng4. Many games has been played so far, but now I've seen a bit of inconvenience ( but's it's a secret ! ) and past year I started ( again ! ) my last improvement !,
7..Nc6!? ( only three games are known ). If 8.fxe4 Be7 often we trasposes to 7..Be7 lines, but in better conditions; against both 9.Be3 or 9.Nc3, always 9..Nh6!?. Finally if logical 8.d5 Ne5 9.Nxe5 dxe5 10.Nxe4 ( 10.fxe4 a6 ) 10..Bf5 11.Bb5+ ( 11.Qe2 0-0-0 ) 11..c6! etc. was a nice Black surprise in Van Willigen-Jove, corr. ICCF thema, TT/4/98/4, 1998-99

I think we need the inspiration of great players as Buecker or our Director Tony Kosten on iluminiting us !!.

Stefan Buecker wrote on 07/10/11 at 19:13:18:


Actually I still think that that, or some closely related idea, makes good sense. But it presupposes no chess expertise on my part, nor indeed do I have very much.

Regarding Tartakover's dictum, it might be worth observing that it was uttered in an altogether different epoch of the game.

Markovich wrote on 07/10/11 at 17:06:12:

You are too modest, e. g. you proposed an improved set of evaluation symbols http://www.chesspub.com/cgi-bin/yabb2/YaBB.pl?num=1266527927/0 . Excerpt:

=b   Scant winning chances on either side, Black's play is easier
b    Some winning chances for Black, scant ones for White
B    Good winning chances for Black, scant ones for White

You seem perfectly qualified to refine Cordel's thumb rule. By the way, in our context another quote by Tartakower seems quite relevant: "If an opening is said to be incorrect, it is playable."