visayanbraindoctor: I have been putting myself in Capablanca's shoes. This game almost defies belief. If I were to play 43. f4 at move 43, in a position that carries no risk at all to me and that most likely would have ended in a draw without any further fireworks, I had better be sure that the ensuing fire would not engulf my own position. That means I would have to calculate and see clearly in my chess eye the position at 49.. Rxc6.
But in this position White only has a Queen and Knight against R + B + N + three pawns! I would have to find a way to regain some material quickly. Easy enough to calculate if I am already at the position at move 49 over the board, 50. Qxg7 Kh5 51. Qh7 Kg5 52. Qxe4 regains some material. But remember that I am still in move 43, and move 49 is still a non existent vision in my chess eye.
Furthermore, the position after 49.. Rxc6 isn't the only potential position after 43. f4. There are dozens of others that can reasonably be expected to occur, and I have to analyze and judge all these potential positions and every other reasonable position from move 43 to 49. The analysis tree branches resemble a dense unnavigable forest. All of these potential branches have to be pictured accurately in my chess eye.
I would not play 43. f4 at all if I thought I would be risking a loss in the ensuing complications. If Capablanca did so, then he must have seen the exact position at 49.. Rxc6, and correctly analyzed and evaluated it as winning for White. And he must have seen at least dozens of other positions and correctly evaluated them as not losing for White at the very least.
The number of chess positions that flashed in and out of Capa's mind must have been staggeringly huge, and he had to correctly evaluate each one of them.
A further scenario. Suppose a computer is limited to analyzing accurately only up to 7 moves. How will it evaluate a position in which it only has a Q and N against R + B + N + three pawns, which is the position at move 49? Even a 'slow' computer might not have seen this line as winning for White. The problem is similar to what computers sometimes experience in the Capablanca vs Tartakover 1924 ending. Some computers initially assess Tartakover's position to be better than Capablanca's, simply because Capa had given away too much material with the win still far in the invisible horizon.
The above considerations are what I have in mind when I said that this game from White's perspective might genuinely have been played at a higher level than what most humans can hope to achieve.