I know this is a parenthetical unrelated to the topic at hand, but I see this a lot and it's really not as bad as the Big Scary Numbers make it seem, at least for 4e. 3e I'll absolutely grant because the vast majority of 3e feats were either garbage or solid platinum, and either (almost) completely generic or so hyperspecialized that there was no reason to consider them.(I'd argue the number of feats in 3.X and 4e are not within reason)
I don't have numbers in front of me right now, but let's say that 4e had 5000 feats by the time it was done. It might be more or less but that's a ballpark. Now, let's say 10% of those feats are truly generic, like multiclass, power-swap, skill training, weapon/armor proficiency (huge list there just by itself), etc. That leaves 4500 non-generic feats. Now, these feats are not all class- or race-specific, but many narrowly pick out a few classes and/or races. I'm going to make the simplifying assumption that 2/3 are perfectly class-specific and the remainder are perfectly race-specific; this is false, but there only to illustrate my overall point. From here, we see there are 25 classes and something like 20ish typical races (ignoring very niche options like bladeling).
So, for any given single character, you have 3000/25+1500/20 = 195 specific feats that are worth considering, and a mass of 500 generic feats, most of which you will never pay attention to because you don't care about being able to wield wands as a Fighter or Ki Focuses as a Swordmage or whatever, and which get heavily reduced the instant you take your first multiclass feat. (Given almost every class had multiple MC feats, I have if anything underestimated the number of generic feats, which would drive down the number of class-specific feats.)
From there, we split it up by tier. 195/3=65 feats.
So, at any given level, you are typically only considering at absolute most about 65 potential feats, most of which won't actually be all that interesting and you'll know they aren't that interesting (e.g. they affect domains or powers you don't have or apply to options you aren't using). The only reason the Scary Big Number looks big and scary is because of the sheer number of races, classes, builds/subclasses, and interactions between these things.
None of this is to say that 4e couldn't do better. It totally could've. There are still a lot of very meh feats, or boring workhorse ones like Expertise. But that shows just how adding a lot of distinct options can actually be way better than the "simple" condensed ones. The Essentials Expertise fears were really good, not because they were necessarily more powerful than ordinary Expertise options, but because they did flavorful or interesting things other than "attack goes up by 1 per tier." Yet the only way to do that was to de-genericize, to make the feats narrower and more focused.
And from that, bringing things back to the topic at hand, we can conclude that yes, having generic but interacting options can be good (as with Background and Subclass), but moving away from generic options and in so doing proliferating them can in fact also be good for depth and richness of play. There is no magic formula. Simplicity uber alles does not get success in this regard, nor does complexity, nor generality, nor specificity. We must really think and, more importantly, test our thoughts against metrics to see whether they achieve the ends we seek. Which is why I almost immediately knew 5e was going to disappoint me the moment Mearls said "math is easy, flavor is hard." Because he's half-wrong. Both things are hard, and they're hard in different ways...but most people who do game design don't have the training to work through the hard math part so they kludge it and slap bandaids on as necessary and call it "easy."