I don't think the latvian gambit is dead. What about
1.e4 e5
2.Nf3 f5
3.Nxe5 Nf6?
I don't see it played that often and it might be good. One way to play it could be
4.Bc4 b5.
5.Nf7 Qe7
6.Nxh8 bxc4
7.0-0 fxe4
I also think the standard way of playing the latvian gambit is good.
1.e4 e5
2.Nf3 f5
3.Nxe5 Qf6
1.e4 e5 2.Nf3 f5 3.Nxe5 Qf6 4.Nc4 fxe4 5.Nc3 Qf7 6.Ne3 c6 7.d3 exd3 8.Bxd3 d5 9.0-0 Bc5 10.b4 Bd6 11.Re1 Ne7 12.Nexd5 cxd5 13.Nb5 Bxb4 14.Rb1! seems much better than the moves silman gave in his articles, the idea found in chesspub.
This move seems to roast black pretty badly. I guess 9...Bc5 is refuted.
Yes, 9. 0-0 Bc5 is refuted. In a challenge game Dave Taylor Vs Ken Reinhart Internet Exhibition game 2004 continued: 9. 0-0 Bd6 10. Re1 Ne7 11. Nexd5! cxd5 12. Nb5 0-0! 13. Nxd6 Qxf2+ 14. Kh1 Bg4 15. Qd2 Qh4 16. b4! Nbc6
17. Bb2 Rf2 18. Qc3 Qh6 19. Rf1 Raf8 20. Rxf2 Rxf2 21. Qe1 Rf4 22. Ne8! Rf7 23. b5 Nd8 24. Qg3 Bf5 25. Qd6 Qxd6 26. Nxd6 Rf8 27. Nxf5 Nxf5 28. Ba3 Rf6 29. Rf1 Ne3 30. Rxf6 gxf6 31. Bc5 Nc4 32. Bxa7 Ne6 33. a4 Nd6 34. Bb8 Nc4 35. Bxc4 dxc4 36. a5 1-0
Never played it.
Still, in my database of corr. games it achieves very, very strong results for black, better than many common openings. I think it should be in the "fundamentally-sound-for-black-but-(s)he-has-to-be-really-accurate" category.