KEG: Maroczy missed his chance to win this game.
Showalter's 14...g5 was criticized by the Tournament Book, but it could have led to approximate equality if properly followed up.
Although Showalter overreached a bit with 16...g4 and 17...Nc4, after Maroczy's 20. Qd2 (20. Nh5 was better) both sides had about equal chances.
But here Showalter erred with 20...NxB (instead of 20...Qh4).
The Tournament Book faults Maroczy's 22. Ne4 (claiming 22. Rad1 was better) and 23. RxQ (contending that the intermediate move 23. Nf6+ was best), but so far as I can discern Maroczy's play was sufficient to maintain a winning edge.
In a bad position, Showalter gamely sacrificed the exchange for a pawn with 23...RxN. This certainly appears to have been Showalter's best chance, but Maroczy could have (and on a good day no doubt would have) nursed his advantage to victory.
Both Fritz and the Tournament Book claim that Maroczy threw away his win with 28. Rfe1 and that the similar looking 28. Rae1 was correct. The abandonment of the f-file does indeed appear to have been a mistake, and after Maroczy's very weak 30. b3 (instead of the much better 30. Rad1) his winning chances, despite his nominal material advantage seem to have evaporated.
Fritz (or at least my version of Fritz) thinks that Showalter's 33...axb4 followed by the exchange of Rooks was a losing blunder. But this is one of those end-game instances in which the judgment of a flesh and blood chess-player proves to be superior to our silicon friends. Although Fritz gives Maroczy a winning edge after 33...axb4 (a move Fritz condemns), Maroczy could make no headway in the resulting Rook and five pawns against Bishop and six pawns ending, and Fritz--though refusing to concede the win was gone- was unable to come up with any winning plans for the remaining more than twenty moves of the game.
Marozcy's error came on move 28 (and to a lesser extent on move 30). After that, Showalter's accurate play kept the draw in hand, Fritz' insistence to the contrary notwithstanding.