And netcode is a small factor in whether games stay alive, I agree with you. SCV fizzled out in weeks, KoF made it about a year, SG packed it up in a few months, etc. I don't buy that argument. But I can't really single out what was most influential in the downturn for KoF.
I wouldn't say it's a
small factor, I'd say nowadays netcode is a
requirement for your game to settle in. A good netcode won't guarantee you anything, but without it your game has all the reasons to get dropped hard and quick, unless you have something really special going for you.
The way I see it is...
* MvC3's netcode is not THAT crappy: not good, but not terribad neither. It's an "old days" netcode, and it suffers for it, but it doesn't make the game "unplayable". MvC3 online is a different game than MvC3 offline, but it's still playable. It got a huge player base because it's the sequel of MvC2, and a pretty good sequel.
* SCV died quickly because the game in itself is not very good. The netcode is good, but the game is quite meh, so there goes...
* KOF is just the opposite: the game is hella good, but the netcode is probably one of the worst there is. I'd say for this particular game, the bad netcode played a big part in why it never really caught up, along with its reputation of being complicated (and it is, to a certain extent: the 4 different types of jumps can be a bit daunting).
* BB/P4A both have an amazing netcode, while being pretty good games (well, depending on the version for BB), but suffer the infamous reputation of the "animey air dash fighters" in the west.
* TTT2 is a good game with a good netcode, so it gets people playing it. Simple as that.
* SF4 is Street Fighter so it gets a free pass whatever it does. This is probably the only fighting licence whose attendance is pretty much guaranted.