Good thread, I didn't remember all the details anymore (and I'm not sure I ever had enough information on temporal AA anyway).
Anyway, I'm OK with FXAA. As far as I've seen, it doesn't really make the image too blurry but it does eliminate the jaggies well enough. Of course I prefer better methods if I have the processing power, but I'd much rather use the extra power for more noticeable things than better AA. Whenever given the exact choice, I rarely go above 2xAA because it seems to be the biggest help and I don't care about further improvements enough to bother playing with the settings to see if I can go higher. Of course with older games, the choice is easy: I can go higher, period.
SMAA is subjectively better than FXAA, and is only a slight bit slower.
I recommend giving it a shot.
Both methods have almost zero impact on PCs (consoles too, but they don't often get a say in the matter).
Temporal AA is cool, but is has too many downsides in the implementations I have seen.
Hopefully some day in the future we can get a smarter, more accurate method of doing it.
Developers could probably devise a method that uses it only in slower scenes, or on select portions of the screen.
Adaptive Temporal Anti-Aliasing, sounds pretty cool.
I'm not sure I have much choice usually, so there's not much 'giving a shot' to do. I'll try to keep it in mind, though, in case I do get the choice some time.
Adaptive temporal AA crossed my mind as well... It sounds neat and not very difficult, but I'm not sure about performance. It sounds like a really simple idea so if it really isn't common, I'm going to assume low performance is the culprit.