But no. 5th edition has to compete with Incantatrix and Mystic Ranger and Erudite. Frankly, 3rd is more fun.
Here's a real world example:
Borderlands was fun.
Borderlands 2 is more fun.
Why? More choices. More adventure paths. and MORE GUNS.
I think this is actually a bad analogy. Or, possibly a counterexample, even. I thought you were saying that later editions/iterations should include at least the "greatest hits" of earlier editions. So, 4E or 5E should start, pretty much from the outset, with say Incantatrix, Mystic Ranger, and Erudite. Or, at least stuff like them.
But, BL2 actually kind of proves the opposite point. Sure, BL2 has a Siren, but there's no sense that Lilith and Maya have the same power set -- their action skills are quite different, and while there's some overlap there's no way anyone would call Maya just Lilith 2.0. Ditto Axton and Roland. Salvador and Brick are similar-ish, except, not really at all other than being beefy.
Instead, the BL2 example simply shows, maybe, that a greater number of options is, according to Captnq, better. It doesn't really tell you anything about "repeats" from the previous editions.
On the other hand, though, BL2 probably doesn't even show that. BL2 was a great game prior to the Mechromancer and the Psycho, the only 2 classes it added to the game. Setting them aside, it's not like it had all that much more content that we'd label "character creation" as compared to its predecessor. It still had a handful of classes, I think the skill trees all had the same amount of numbers to it, and it had a whole 1 new elemental damage type (slag). Maybe it had more guns, but when the number is nigh infinite to begin with, that's a trivial difference. Perhaps there was a marked uptick in the number of legendaries in the game, I don't know.
Now that we're making the comparison, BL2 just shows that you need some minimal level of options -- you need all the classes to have multiple, engaging, viable builds, supported by equipment or whatever tools the game gives you. And, that those options just should be
better than what came before. BL2 is, at most, a slight increase in variety of options, from a raw numbers perspective, than BL1. But, it made a great many of those options much more engaging and awesome.
So, if I'm going to take the move from BL1 to BL2 as a leading example, I'd say this. Say 3E D&D has 1000 options. Successor editions don't need to have nearly that many. But, if they have some critical number, like 50 or whatever, really good options, then that would easily fit the BL2 model.
My point here is that I don't think more = good is necessarily supported by this example. And, as Raineh Daze points out, there are costs to vast amounts of stuff -- decision paralysis, cumbersomeness, game balance, etc.
One place where I do find it jarring, and something that D&D in particular has been kind of bad about is if there was an archetype that was reasonably well-supported in previous iterations that is absent from the new one. If I could make a cool Swashbuckler (or shapeshifter or blaster or assassin ...) in edition X, and if that archetype is a fairly classic or obvious one, I get jarred when I can't make one in edition X+1.
P.S.: the game(s) could really be organized in a way that is much more user-friendly to veterans and new players alike.