I think because 3E is a more coherent game than 2nd ed AD&D.
When I read 2nd ed material, it's full of remarks like "Don't let the rules trump the fun." This implies that the rules, as written, aren't apt to produce fun of the sort the game is aiming at. Which is to imply that they are bad rules.
2nd ed material also has a strong tendency towards railroading by the GM. This is linked to the point above - the rules aren't apt to produce the intended result in play, and instead of new rules that allow the players to make choices that will deliver that result, the GM is encouraged to impose it, despite the rules, in a unilateral fashion.
I think 3E was a success mostly because its rules (i) were more apt to produce what was wanted in play (eg clearer encounter-buidling guidelines, more access to healing etc) and (ii) transferred power from the GM to the players (eg much more intricate character-building rules, which in turn provided input into mouch more intricate and robust action-resolution mechanics).
To the extent that 2nd ed supported a playstyle, it seemed to be a type of high-concept simulationism (the GM fiating and railroading to ensure the delivery of an appropriate high fantasy adventure) with mechanics that pushed in quite a different direction (and therefore had to be frequently ignored).
To the extent that 3rd ed supports a playstyle, it seems to be a curious hybrid of purist-for-system simulationism (let the mechanics tell us what the world looks like) and gamism (let's push the mechanics as far as we can go!) - the inherent incompatibility of these two playstyles is expressed in the conflicts we see on messageboards between those who love the flexibility and variety of all the classes, races, feats, spells etc, and those who call themselve optimisters and are often labelled by others as munchkins.
