You can still find it on the Wayback:

https://web.archive.org/web/2001103...

<It is from like 2005 and my pages on GeoCities were closed over 10 years ago.>

I am thinking 10.Nd5 but the engine gives the line 10..exd5 11.Bb5+ Kd8 12.Bxa6 bxa6 13.Bxf6 Qxf6 14.Qxd5+ Kc7 and now of course 15.Qxa8 Bb4+ does not work. So instead 15.0-0-0 Qc6 16.Qxf7+ Kb6 17.Qb3+ and a draw.

An interesting remaneuver to the King's side is 9. Bg5 Qc7 10. Bxf6 gxf6 11. a4 Bg7/Bh6 12. Ra3 Na6 13. Ne2 Nc5 14. Qh3 Ne6

click for larger view

After 9...Na6 10.Bxa6 bxa6 11.O-O-O, 11...Qb4 relieves some of the pressure. Although Black would nonetheless be saddled with four pawn islands after an upcoming Bxf6 gxf6, retaining the queens might give White more opportunity to exploit Black's broken pawns.

This suggests [9...Na6 10.Bxa6 bxa6] 11.Qa4 Qb7 12.Bxf6 gxf6 13.Nd5

click for larger view

I perversely ran this game through Stockfish, which gave Morphy a 97.9% accuracy, and estimated rating of 2500.

What moves did Stockfish find fault with?

4. dxe5: Stockfish thought this was Third Best, and wanted 4. Nc3.

6. Bc4: Stockfish thought this was Second Best, and wanted an immediate 6. Qb3

8. Nc3: Stockfish thought this was Second Best, and wanted 8. Qxb7

That third one I can kind of go along with. That's the question beginners always ask about this game. Why didn't White play 8. Qxb7? And they usually get some nebulous answer about well, er, he wanted more. So, in other words, White avoided the endgame and <gambled> that Black would give him more opportunities in the middlegame. Okay, but what was the actual best <move> in this position? 8. Qxb7, you say? Yeah, okay.

The second one is iffier. On general principle, developing a piece with 6. Bc4 must be better than moving the Queen a second time, right? Yet Stockfish says +1.42 after 6. Qb3, and +1.23 after 6. Bc4. Is there really any way that 6. Qb3 is better than 7. Qb3, even slightly? Well, maybe. Black pretty much has to sacrifice the b pawn after 6. Qb3, because 6...Qc8, and 6...b6 are both bad. Stockfish thinks 6... Nd7 is the best, which I agree with, but how likely was Black to play it? Knowing how cavalier 19th century players were about giving up pawns, maybe very likely. But still, how could 6. Qb3 be objectively better than 6. Bc4? The only thing I can think of is that 6. Bc4 gives Black the opportunity of 6...Qd7 and 7...c6, protecting both b and f pawns. 6. Qb3 puts the question to the b pawn quicker. So, all right, I'll maybe concede that 6. Qb3 is slightly better for a high rated player <who knows when you can get away with violating opening principles, and when you can't>. For most people I'd still recommend 6. Bc4.

On the first one, 4. Nc3, no. 3...Bg4 is a bad move that has to be punished. 4. dxe5 forces Black to either give up a pawn or give up Bishop for Knight in a wide open position. This is the move you should be playing. What does Stockfish say? +1.00 after 4. Nc3, +0.98 after 4. Be3, and +0.96 after 4. dxe5. I'm not sure if it means anything to say that Nc3 is 1/25 of a pawn better than dxe5, but even if it does, that's assuming perfect play on both sides, which you can never do in a real game. In a real game if two moves are effectively equal, but Move A gives your opponent 5 chances to go wrong, while Move B gives him 20 chances to go wrong, you should play Move B.