Wikipedia describes it as "distinctly old-fashioned and the least popular of the defences at Black's third move," albeit adding that "it remains one of the least explored variations of the Ruy Lopez."
Nevertheless, the Cozio always had a dedicated following among those wanting to play the black side of the Spanish, while at the same time avoiding the Exchange Variation, 3...a6 4.Bxc6.
But in recent times it fell further from grace, partly due to the rise of the Berlin Defence, 3...Nf6, but also due to the popularity of a variation that came to be seen as more-or-less a refutation.
However, it is too early to read the last rites over the Cozio, because it seems to be making a comeback.
Just one person played it against me in 2023, and no one tried it the following year, but in 2025 I faced the Cozio twice.
Those are small samples, but already this year I have had it played against me twice, the second time being in round one of the Thuringia seniors' championships, where I was White against Stefan Krämer (1544).
 |
| Position after 3...Nge7 |
*****
*****
*****
*****
Before getting to my game, I want to mention Andrew Soltis's 1994 Chess Digest book, Beating The Ruy Lopez With The Fianchetto Variation.
The first three chapters cover, as the title suggests, 3...g6, but Soltis freely admits there are problems with fianchettoing so early.
He explains: "Because of the tactical problems presented by 4.c3 and 4.d4 exd4 5.Bg5, many strong players who like to play the Fianchetto Defence have adopted a different move-order."
The new move-order meant first developing the king's knight to e7, and then playing ...g6 and ...Bg7.
Soltis calls it the Improved Fianchetto, rather than the Cozio, albeit admitting the system "has been played for decades," which is arguably understating matters since 3...Nge7 goes back to at least 1836.
In that first game, White replied with 4.Nc3, but the knight move was quickly overtaken in popularity by 4.c3, 4.d4 and 4.0-0.
Even so, 4.Nc3 was still seen, with games often continuing 4...g6 5.d4 exd4 6.Nxd4 Bg7, when White probably has the first player's usual slight advantage, but Black has nothing to complain about.
That changed when players started appreciating the strength of 6.Nd5, delaying recapturing on d4 in favour of threatening mate on the move!
Tartakower met this threat with 6...Nxd5, but after 7.exd5 Qe7+ (the point behind Black's knight capture) 8.Kf1, White is winning, according to Stockfish17.1 and Dragon1, eg 8...Ne5 9.Qxd4 Bg7 10.Bf4 f6 11.Re1.
Probably a better try for Black is completing the fianchetto with 6...Bg7, but, after 7.Bg5, Black's only decent continuation is 7...h6 (7...0-0 loses to 8.Bxc6), when 8.Bf6 Bxf6 (best) 9.Nxf6+ is at least slightly better for White (Stockfish17.1), or perhaps gives White the upper hand (Dragon1).
One person to lose to it was Alexey Dreev, in 1989, so it was interesting to see what he came up with in his 2014 Chess Stars book, Anti-Spanish: The Cozio Defence.
Dreev points out that meeting 4.Nc3 with 4...Ng6, which was recommended by, among others, Alexey Suetin, allows 5.d4 exd4 6.Nxd4, where White "enters positions which are similar in their pawn-structure to the Scotch Game, but in a slightly better version because Black's knight on g6 is a bit passive."
Not surprisingly, bearing in mind his experience as Black, Dreev is not keen on meeting 4.Nc3 with 4...g6, as 5.d4 exd4 6.Nd5 (Dreev gives this move an exclamation mark) "fights for the initiative."
Not surprisingly, bearing in mind his experience as Black, Dreev is not keen on meeting 4.Nc3 with 4...g6, as 5.d4 exd4 6.Nd5 (Dreev gives this move an exclamation mark) "fights for the initiative."
He instead highlights the "more modest and not so risky" 4...d6, which he describes as a "fashionable" way to transpose to the Steinitz Defence, 1.e4 e5 2.Nf3 Nc6 3.Bb5 d6, in which Nc3 has been played "too early."
The message seems to be getting through - I have played 4.Nc3 against the Cozio eight times since 2018, including at Weimar, and no one has replied 4...g6.
But half the games, including the one at Weimar, saw a reply not mentioned so far, namely 4...a6.
This has been played by Nakamura, and seems very logical - after all, a major point of 3...Nge7 is to take any possible sting out of the Exchange Variation.
Naturally, replying 5.Bxc6 is somewhat silly, although there are 49 games with this continuation in ChessBase's 2026 Mega database.
However, as Dreev points out, the usual Spanish retreat, 5.Ba4, is not forced, and he recommends 5.Bc4, which is what I played.
Dreev adds: "Following 5...b5 6.Bb3 White, in view of the threat Ng5 and his more harmoniously deployed pieces, preserves a stable positional advantage."
Alas, I met 5...b5 by capturing on f7, which gives White a pawn and a check, but that is nowhere near enough for a bishop.
The engines award White the upper hand after 6.Bb3, so, instead of 5...b5, they suggest 5...d6, and if 6.Ng5, which features in nine of 20 games to reach the position in Mega26, they give 6...d5 7.exd5 Na5, claiming White has at best the usual slight edge of the first player (and Stockfish17.1 is not even sure of that).
All in all, it seems the Cozio Defence should once again be considered a viable proposition, remaining, as Wikipedia puts it, "one of the least explored variations of the Ruy Lopez."
White players of the Spanish should make sure they are not caught by surprise.
No comments:
Post a Comment