<13...Nd8> This line might be inferior to <13...Na5 14.b3 Bd7 15.Nf1 Rfb8 ⩲> as in the game <Triantafyllos Siaperas (2385) - Ion Balanel, Moscow Olympiads (USSR) 1956 (Men), qualif. Group D, round 7 - Greece vs Romania, 0-1>, though it did not exist in 1924...

<15.Bg5 h6> Trading this bishop on f6 doesn't seem to be in the spirit of the Ruy Lopez, in which the bishop pair is important (hence I could not find it played in an earlier game, nor in any serious game since). More commendable would be <15.b3> in order to answer <15...c4> with <16.b4 ⩲> denying the black knight access to c5. <Viktor Moseev (2370) - W. Massing, Kecskemet Open (Hungary) 1990, 1-0>.

Instead of <15...h6>, after which White could have retreated the bishop instead of trading it with <16.Bxf6?! Bxf6▢=> (16.Bd2 Bd7 17.Ng3 ⩲ Stockfish 8 - 64 bits POPCNT), Black could also have played <15...c4 16.Qe2 Bd7=/ ⩲> followed with Nc5. Stockfish 8 - 64 bits POPCNT.

<17...Bd7=> Stronger seems to be <17...c4! ⩱> to immediately get the c5 square for the knight. Stockfish 8 - 64 bits POPCNT.

<18.g4?! g6 ⩱> Again, better is <18.b3=> Stockfish 8 - 64 bits POPCNT.

<19...Bg7?!=> Seems logical, but this bishop would be better on e7, thus <19...Kg7 ⩱> Stockfish 8 - 64 bits POPCNT.

<20.Qd2 Nc5 21.Rag1 a5=/ ⩱> More precise seems <20.Qe2 Rae8=> Stockfish 8 - 64 bits POPCNT.

<23.Kf1?> White's plan cannot be clearer! But here, this is a faulty execution. This is the first serious mistake by the older teenager. It was better to target the ♙/c4: <23.Qe2 Qc8=/ ⩱> Stockfish 8 - 64 bits POPCNT.

<23...Rab8?! 24.a3 ⩱> Young Botvinnik missed a tactical opportunity here: <23...b4! 24.Bb1> (24.a3 bxa3! 25.bxa3 Rab8 ∓) <Rab8 ∓> Stockfish 8 - 64 bits POPCNT.

<23...Rab8? 24.a3?!> I worsened my evaluation of Black's 23rd move. As for White: stick to the plan! Do not let the opponent distract you from it! <24.h4! b4 25.h5! g5 26.Kg2! Qc8 27.Nh2=> Stockfish 8 - 64 bits POPCNT.

<24...Na6?! 25.h4 b4 26.h5 g5=> Apparently played so the ♙/c4 will be protected by the queen when b5-b4 will be played. Still <24...Rb7 25.h4! b4! 26.axb4 axb4 27.cxb4! Nd3! 28.Bxd3 cxd3 ⩱> Stockfish 8 - 64 bits POPCNT.

Position after 27...Nf5?? Black wins!

click for larger view

<27.Nf5??> Incredible, this apparently plausible move loses the game! White could have played, amongst other moves:

A- <27.Kg2 b3 28.Bb1▢> (this bishop is guarding d3 and e4) <Nc5=> Stockfish 8 - 64 bits POPCNT;

B- <27.Qe2 b3 28.Bb1> (28.Bd1 Nc5! (Beware of opening the a6-f1 diagonal leading to the white king!) 29.Nf5! Kh7=) <Nc5=> Stockfish 8 - 64 bits POPCNT.

After 27.Nf5?? Botvinnik was now winning and played <27...Bxf5?? 28.gxf5!=/ ⩱> to eliminate this threatening knight. Maybe there was also a zeitnot (time pressure to reach the time control) involved, as we are getting close to move 30.

What should Botvinnik have played is the question! <27...bxa3! 28.bxa3 Rb2!> And here:

A- <29.Kg2! Bxf5! 30.gxf5 Nc5-+>;

B- <29.Ne7+? Kh8! 30.Qc1 Rfb8-+>;

C- <29.Qe2 Rfb8 30.Kg2 Nc5! 31.Rb1 Bxf5! 32.gxf5 Nd3-+>;

D- <29.Ne3 Ra2! 30.Kg2 Nc5▢-+> followed with Nxe4. All lines validated with Stockfish 8 - 64 bits POPCNT.

<28...f6> Botvinnik is giving up on winning this game. <28...Kh7! 29.Kg2 b3! 30.Bd1! Nc5=/ ⩱> Stockfish 8 - 64 bits POPCNT.

<29...Rb6?!> Better to complete the blockade at this point: <29...b3! 30.Bd1> (or 30.Bb1 a4=) <Nc5!=> Stockfish 8 - 64 bits POPCNT.

<30.Rc1?! bxa3 31.bxa3▢ Nc5=> Stronger is <30.axb4! Rfb8!> (if 30...axb4 31.Ra1! b3 32.Bd1▢ Nc5! 33.Qe3 ±) <31.Ra1 axb4▢ 32.Qe2! ⩲/ ±> Stockfish 8 - 64 bits POPCNT.

<35...Be7?!> Better was <35...Nxd2 36.Qxd2> (or 36.Rxb6 Rxb6 37.Qxd2 Be7=) <Be7=> Stockfish 8 - 64 bits POPCNT.

<36.Rb2?! Nxd2! 37.Rxb6▢ Rxb6 38.Qxd2=> Another missed opportunity: <36.Bxb3! cxb3 37.Rb2! ±/ ⩲> Stockfish 8 - 64 bits POPCNT.