korger: <Honza Cervenka> I'd like to offer an explanation to those perplexing moves you pointed out. The authoritative source for this game seems to be the Book of the Second American Chess Congress, available on-line at https://www.google.com/books/editio..., where this game is listed as Game No 30. White's and Black's 16th moves are given, in descriptive notation of course, as16. R-KB5 Castles
Suppose that Black actually castled long, which is the obvious move, but they forgot to record this fact. With this change Black's 16th and White's 17th moves are no longer blunders, and the game progresses like given here, up until move 21, which was given as
21. P-QR3 QR-B
In the version having 16...O-O this was transcribed as 21...Rac8, but with 16...O-O-O this also makes sense, and now it becomes 21...Rdf8. The game then continues with the same moves, although the position is different.
The full score with this change is then
1. e4 e5 2. Nf3 Nc6 3. d4 exd4 4. Ng5 Nh6 5. Bc4 Bc5 6. Qh5 Qf6 7. O-O d6 8. h3 Ne5 9. Bb3 Qg6 10. Qh4 f6 11. f4 fxg5 12. fxg5 Nhf7 13. Kh2 Be6 14. Ba4+ Nd7 15. c3 Bc4 16. Rf5 O-O-O 17. Bd1 Nfe5 18. b3 Bd3 19. Bb2 dxc3 20. Nxc3 c6 21. a3 Rdf8 22. b4 Bb6 23. Bg4 Nxg4+ 24. hxg4 Bc7 25. Qf2 d5+ 26. Kh1 dxe4 27. Qxa7 Rxf5 28. gxf5 Qh5+ 0-1,
which otherwise agrees with the tournament book. I've also checked this with a computer, and it makes perfect sense--in fact, Black's play was impeccable in this version.
Could we accept this reasoning as evidence that the score had a transcription error, and accept the proposed change as the correct version?