Except 3.5 is also Pathfinder 1e which was also the best thing since slice bread. It is the quantum game, both the best and worst game depending on who is holding it.
If one is looking for a game where there are rules (and plenty of them), but an absolutely enormous space of
possible combinations with a variety of distinct textures, it's hard to beat 3.5e. It has epic play, half a dozen alternative magic systems, several dozen classes, zillions of feats, enormous piles of prestige classes. It can be an actually fun challenge trying to find a build that achieves your goals--yet, because there
are so many rules, it's generally possible (but difficult!) to also evaluate whether a particular build is better or worse at that goal.
Pathfinder reduplicated much of this in its own way, largely axing PrCs but creating even
more feats and (especially if you allow 3PP supplements) base classes, but also by creating its Archetype system, which sort of streamlines the process of doing the PrC "thing": you have to choose between often-mutually-exclusive options, which can achieve an enormous variety of specific ends by making sacrifices. (Or, if we're being honest, by making the smallest possible sacrifices for those goals, because PF, like 3.5e, does passively encourage min-maxing through the many options that trade fluff abilities for significant power increases.)
But, and this is a huge sticking point...there's also a lot of people that got really really burnt out on or frustrated by that very build-o-rama nature of 3rd edition. It might be the smaller group, but not by too much. (That's the unfortunate thing with sales--if 40% of the base loves X and 60% absolutely loathes X, all you see is that your sales have crashed. Even if it were 80/20, you'd see problems.) Because, if we're being really brutally honest, that huge possibility space I mentioned before? It's
mostly full of garbage. Like, nearly all of it. The vast majority of PrCs, feats, and base classes just, frankly, suck. Moreover, they suck in ways that are really hard to see unless you've sunk your teeth pretty deep into the system. See, frex, all the hand-wringing about how "powerful" Monks were because of their pile of special features, despite the class being pretty obviously
awful in hindsight.
And then, of the ones that remain, a pretty significant handful are stupidly overpowered in various ways, or more commonly, so incredibly good that no character that qualifies would ever ignore them. Natural Spell is of course a go-to example there, but so is the Pounce ability--if you can acquire it somehow,
no matter how much you torture the rules to do it, you ABSOLUTELY should, because it's an enormous boost.
4e said, "Let's try to give the players a game where the vast majority is NOT garbage, yet also is NOT overpowered or must-have. And where the players can see reasonably quickly how different parts interact. Oh, and where they just DO their cool things right away, without having to wait for a long time to qualify first." So,
by the standard listed above, this would be the worst game imaginable. There's by definition far less variety because, y'know, brand-new game can't compete with one that's got multiple years of
heavy splatbook support. Further, cutting down on
bad variety was an explicit goal, but since that's easy to re-frame as
demonizing variety when you're already upset about the lack thereof, what should be a selling point instead looks like an openly-recognized flaw.
Then, with every class bringing multiple Cool Things you can do from level 1, it can look an awful lot like the game has killed the character-building minigame dead; no one will need to discuss builds when it's so obvious what works and what doesn't, right? (Obviously this wasn't true. But the active pursuit of clear, direct rules contributes to that
feeling on first reading even if it's ultimately wrong.) And when everything is just good, what does it matter whether you choose Good Thing A or Good Thing B? Even though that choice has plenty of reasons to be more meaningful than choosing between Good Thing A and Bad Thing B, it's easy to re-frame that as choosing between
indistinguishable things, and thus (again) turning what should be a selling point (your choices should always be meaningful because both options have value) into an openly-admitted flaw (your choices "don't matter" because they're "the same" either way).
3rd edition D&D and Pathfinder 1st edition will almost certainly retain a fanbase well into the foreseeable future. It's going to be almost impossible to match the perfect storm those games offer. No game can ever come out on day 1 with a similar level of variety and valid available content. Further, trying to do what those games did but
without the cruft and crap options is going to be incredibly hard to communicate to an audience that is primed to reject alternatives
and so easily turns "we're trying to fix Problem X" (caster/martial disparity, for example) into "we're trying to destroy Beloved Thing Y" ("magic that feels magical"/"magic that actually does things" etc.)
Edit: This, incidentally,
is one of the few ways in which 5e truly is a compromise edition, rather than "copy 3e, but with most of the worst holes patched up." It reasonably successfully captures some of the PF-style "Archetype" build game, as one can see from all the hand-wringing about things like Hexblade Paladin multiclasses. But, if you saw off the first two levels and do a fair amount of work on the back end as DM, you can finagle a game that reasonably gets people straight to the "you're good at X" and doesn't have
too many bizarro exceptions or annoying rules. Especially if you're the kind of DM who would refer to that sort of thing as "building
my game with a toolkit" rather than "papering over the holes." (I doubt it's going to surprise anyone that the latter is how I view it.) You get
some of the 4e "streamlining" as 5e's designers put it, but
some of 3.x's "what do you want to build for?"
