They must be grateful that Topalov declined to participate.

Here's another one that at least rivals the 2005 game vs. Anand:

Topalov vs Kramnik, 2008

I think the widespread dislike of Topalov has nothing to do with his playing style (which I, too, absolutely love), but rather such incidents as his Danialov-"inspired" (*) protests at the 2006 world championship match. [(*)-Here, Conrad's phrase, the "demon of perverse inspiration" comes to mind.]

I also think that most of the bad press is pertaining to a few controversial past events and it is time we moved beyond.

The WCC match with Anand that went to the wire on Topalov's home turf has proved beyond doubt that he plays a fair game and is a modern day legend. Anand won much more convincingly against Kramnik.

Moreover I have yet to see an engine that can play games like this one or the one against Kramnik that you highlight. Or even a draw like this

Anand vs Topalov, 2005

As regards choosing the best game, to me the Kramnik game was visually more appealing though this one is no less spectacular. I'd be happy with either.

1. What happens if 15... or 16...Bxd5?

2. What's the point of 20...b5?

From one chess blind fellow to another:

<1. What happens if 15... or 16...Bxd5?> I think the reason black didn't take basically has to do with black's queen being on the same file as white's d1 rook combined with the fact that black's king isn't overly safe in that position. A couple example lines I'm coming up with on the spot (which means they've probably got gaping flaws in them, but oh well): 16...Bxd5; Bc4, Bxc4; Qxc4+, Kf8 (Ke8, Re1+); and then something like Ng6+ or Bf4

16...Bxd5; Bc4, Be7; Bf4 and the bishop on d5 is awkwardly double pinned.

If nothing else, taking the pawn would seem to be risky if not suicidal.

<2. What's the point of 20...b5?>

b5 deflects white's bishop away from the protection of d5, which is supporting white's strong rook on e6. Notice how black captures the d5 pawn shortly after playing this.

<3. What happens if 25...Qxh8?>

Probably 26.Qe4, threatening both Qxa8 and Qf4+.

Qe4,Rb8; Qf4+ wins the black rook anyway.
Qe4,Bf5; Qd4+, Kg6 (Kg5; h4+, Kg6 or else it looks like black could be mated); Qd6+ picks up the black knight and now black's bishop is loose, king is still exposed, and queen is out of play in the corner.

FWIW, here is a summary of a few key features of this game that I posted earlier today at page 30 of this thread: Topalov vs Ponomariov, 2005:

Topalov’s sacrifice in [the Anand] game (<14. Nxf7!?>) in this position:

click for larger view

was of a decidedly speculative character. He was probably counting on the fact that he had heavily analyzed the resulting position at home, whereas his opponent would be needing to find his way over-the-board, but to play such a move against a great defender like Anand invites comparison with the first well-known essay of the Marshall Gambit: Capablanca vs Marshall, 1918.

Topalov fared better against Anand at Sofia 2005 than Marshall against Capablanca at New York 1918 when Vishy went astray with <21. … Be7??> (Better would have been <21. … Kg8>.) If two moves later Topalov had found the best continuation (<23. Re5!> putting Black’s pieces in an inextricable tangle from which major material losses could not have been avoided) instead of playing the much weaker <23. Rxe7?!>, the game would have ended much sooner. As the game actually went, Topalov won a beautiful ending coming down to a final position in which Anand, in the position where he resigned:

click for larger view