thomastonk: This is the stem game of the Guimard variation of the Tarrasch Defence (C04, btw), and a complex game without kibitzing.
G.H. Marten wrote in his Nimzowitsch biography (Verlag Das Schacharchiv, 1995) that he would like to include an analysis by Robert Huebner into the second edition. Well, I played Huebner in 1985, and so I feel free to add some comments here.
Black's opening concept didn't work, and after 8.. b6? he is already in trouble. Richmond (in his book on the Tarrasch from 1980) notes 12.d5! , and this is his only comment for the fragment. But 12.d5? is a mistake, because simplifications are in Black's favour. 12.0-0 or 12.♕e2 would have been . White could still have saved an significant advantage by 13.♗g5!?, but you can guess who this move suggested.
Both sides continue normally until 20.. ♖e8?, after which the exchange sac at the next move is forced and yields a lost position. Here it was possible to improve the knight on a5 by 20.. ♘c4 .
After winning the exchange, Spielmann prepares for more than 20 moves the final blow, because Black's blockade cannot be broken down by a fingersnip.
After the game he thought that he missed the win by 50.Rc8, but this move is okay. Instead 51.♖c7+? is the mistake, whereas 51.♖xc6 is still winning.
The final position is unclear: pawns are equal and Black is an exchange down. I played the position with my engine, and its plan is as follows: sac the h6-pawn, exchange a pair of rooks, sac the d4-pawn, and break through with the king via d4 into blacks camp. Very interesting, I think.