Black's best strategy is to pound away at White's c-pawn with ...c5, ...b6, ...Ba6, and ...Nc6-a5. (Indeed, he can even try to dispense with ...c5.) To that end, Stockfish 17 says that Black's best moves are 5...Nc6, 5...b6, and 5...c5. JRG's 5...O-O is a slight inaccuracy and only Black's seventh-best move according to Stockfish 17. Indeed, White often plays a deferred Saemisch with 4.e3 O-O 5.a3 (as in Botvinnik vs Reshevsky, 1948), or 4.f3 O-O (4...c5!, 4...d5) 5.a3, arguing that the Saemisch is more effective after Black has castled.

6.f3! would have been the best response, intending to play e4 in one go. 6.e3 can also be considered an inaccuracy, at least by today's standards. Of course this game was played 64 years ago.

9...Ne8! looks odd to the uninitiated, but the knight gets out of the way of both e5 and the pin Bg5, and prepares to move his f-pawn (...f5 or ...f6) when needed to slow down White's offensive. Sometimes the knight can pop out to d6 to intensify the pressure on the c-pawn. The move was introduced in the aforementioned Botvinnik-Reshevsky, a famous game that JRG surely knew. 9...Ne8! is by far Black's most common move, Opening Explorer, and clearly best according to SF.

White usually responds with 10.O-O, when Black has equality after 10...Ba6. The second most common move is 10.Be3, when Black should ignore the threat to the c-pawn with 10...Ba6!, when 11.dxc5?! is met by 11...Ne5! 12.cxb6 axb6 and Black is slightly better.

Diez del Corral instead played the natural 14.f4?!, which was a mistake because of 14...f5!=, as JRG played. Now White should have played something like 15.Rad1 or 15.Rae1. After 15.d5, JRG should have played 15...exd5! 16.exd5 Bxd3 17.Qxd3.

click for larger view

Here a human way to play would be 17...fxe4 18.Nxe4 c4! 19.Qe2 and now 19...Nc7 or 19...Nf6 with equality. SF instead gives the immediate 17...Nf6! Now 18.e5 dxe5 19.fxe5 Qxd5 20.Qxf5 Nd7 leaves Black on the sunny side of equality, 18.Nxf5 Rae8! 19.Ng3 Qa4! is slightly better for Black, and 18.exf5? Qa4! leaves White struggling to stay alive.

JRG's 15...g6= was OK, but not the most incisive. Now DdC should have kept the central tension with something like 16.h4 intending h5. His 16.dxe6 Qxe6 17.exf5 gxf5 18.Rae1 would have left JRG slightly better after 18...Bxc4. JRG instead played a little too solidly with 18...Ng7. After 19.Nh5!, SF says that Black can maintain equality with 19...Qf7 20.Nxg7 Kxg7; 19...Nxh5; 19...Bxc4; or 19...Nxc4. Having done all this preparation to win the pawn on c4, Black should take it off!

Unfortunately, JRG blundered with 19...Rae8?, overlooking the strength of 20.Nxg7 Kxg7 21.Bd4+! Kf7▢ 22.Qf3! Bb7▢, and now White headed for a winning ending with 23.Rxe6.

After 25...cxd4, 26.Bxf5! was most accurate. Instead, DdC's 26.cxd4 left JRG with chances to resist after 26...Nb3! 27.Rh3 (as played) Kg6! JRG, probably in time pressure, played 27...Nxd4? and White won.

I didn't see a problem for White so I ran it through SF.

click for larger view

I didn't see a problem for White>

How does <fishie> rate 5....d6?

<....As we all know, it's much harder to actually play the game than to sit on the sidelines with the all-seeing engine and take potshots.>

Slightly behind the three I mentioned. At the moment (depth 41/66), it's giving 5...c5 -0.06, 5...b6 and 5...Nc6 0.00, 5...d6 +0.09, 5...h6 (which no one plays) +0.11, 5...O-O (now in sixth place) +0.20. These numbers are neither etched in stone nor very meaningful. If I let Stockfish run longer they would change, with all decent moves likely heading in the direction of 0.00. They can also change from test to test, and based on how many candidate moves you're asking SF to consider.

The other day, I asked SF about a position in the Albin Counter-Gambit. I asked it for Black's four best moves. At depth 44, it said 5...Nge7 was best (+0.59) and 5...g6 a hair behind (+0.61). I stopped that test and then asked for just Black's best two moves. Now it said, again at depth 44, that 5...g6 was best and was .06 better than 5...Nge7. Sometimes it will give two lines that transpose into each other, and claim that one is .04 better than the other! No doubt computer experts could explain these phenomena. I'm no computer expert.