Sure they could manage a dozen good characters. Easiest way to do that is get rid of a couple of guilds. They could also improve the AI and give each NPC better routines. Best way to do that is get rid of a bunch of them. BGS games have always been quantity > quality. The more choices you give the player, the shallower every choice will be. There's no weird dichotomy at all: the quests are linear because it's a huge world with a million things in it. You can't make every one of them super. More time spent on one NPC means less time on another. Same for quests. I don't really know where people think all this extra development time is going to come from. They confuse real limitations with laziness and ineptness. Could the characters and stories be better? Sure. Just have fewer of them. Then you'll have more time to spend on each one. Of course, making decisions for the player beforehand can be really handy, too, because it prevents them from making other decisions that you'd have to incorporate.
I'm not asking for Bioware levels of depth with all the characters here. Doubling or tripling the lines of the main characters wouldn't remotely achieve that. Having better writers for the lines that do exist wouldn't achieve that. It would make the story a lot better though. It also wouldn't be that expensive compared to how much is spent on the game. Anyhow, it isn't about fewer people or cutting down the world. They don't manage the resources they have well. They keep whatever mediocre talent they have rather than getting better or demanding better. They just don't care much about the story being good, so they don't have their resources on it spend effectively. It isn't like what little we get is of good quality; by and large it isn't.
I think you are confusing AI with scripting. Skyrim has crappy, crappy, crappy, crappy AI. Most of it is heavily scripted, which is resource-intensive and means anything not scripted for cannot be reacted to. If they made a decent AI, then guards and others could react more to novel things and allow more types of interaction with the player. Say taking some guidelines from psychology and what can drive people and tossing some stats on everyone representing greed, altruism, concern for family, loyalty, etc, then have an AI interpret those stats to determine behavior (say, how they'd react to a bribe to look the other way while you kill a neighbor or betray a friend). That would give them rich dividends in the long run, since they'd save a lot of time writing specialized scripts for everything under the sun. They could toss some group-level AI fro bandits, city guards, and the like to determine how they move, grow, recruit, and so forth. Quests could then be generated automatically based on what is dynamically happening in the world based on how these groups behave. It's isn't like AI for group decisions or individuals hasn't been done before. Bethesda just doesn't even try to do it really, and so their world is pretty bland.
Heck, the little work they do make they don't carry on with in the future. The disposition system in Oblivion? That could have been further expanded into different types of feelings regarding your character or others and been given depth in any number of ways. Instead they scrapped it and replaced it with a like/neutral/dislike system that is just awful. The same is true of the random conversations in Oblivion, that also could have been improved and tied into other advances. They've made the game more shallow, not less. They scrap advances rather than build on them.
So I can't say I find any of your arguments particularly compelling. They could do more and they don't. They have the resources for it and it would pay off in the long run. Then again, it would be change and it would be work. The market buys what they do, so why bother to do more? So while I am frustrated by how they toss aside good ideas and don't build up systems and advances each game so each one is better than the last, I can't really blame them.