Entertaining as this is, I think the class/skill dichotomy kind of misses the point. Or rather, there are lots of distinct issues here and lumping them all into classed vs classless tends to obsure more than it illuminates.
I think we can all agree that the real problem in so many RPGs is the lack of reactivity to player choice, particularly choice in character building. You could have a Fallout/GURPS style classless skill based system, but if the game frequently and dramatically reacted to your tagged skills as though they were an important part of your character that would make for an awesome RPG.
Bloodlines is great because the game reacts to your clan at nearly every turn. But that really has nothing to do with class based vs skill based. There’s no reason you couldn’t do the same thing with skills, it would just require substantially more work for the writers. Maybe not worth the effort, but it’s not incompatible with a skill based system.
Have any of you played the Stygian demo? Rather than classes, you pick archetypes (criminal, investigator, con man, aristocrat, occultist, academic, soldier, explorer etc...) and these determine which of your skills get tagged. So it’s a skill based system that’s somewhat restricted by class, and the demo does a pretty good job of reacting to your character build choices. We’ll have to see how that actually turns out, but the model seems solid.
Maybe the best term to use is just role. We want RPGs that make you pick a role and then react to it via more than a handful meaningful dialogue checks and a bunch of insignificant ones, right?
So the problem with, say, Pillars is that while it has plenty of stat checks, as well as race checks and class checks and skill checks (I admit the skill system was pretty pointless), it’s all kind of scattershot. The world doesn’t react to you consistently in a way that makes you feel like you have a particular role, other than Watcher, which is not a choice and also pretty meh.
But it doesn’t necessarily have to be class that the game reacts to consistently to make you feel like you’re playing a role. An RPG set in a world full of racial violence could make your race the defining feature, or you could have a world of religious war where your faith was the defining characteristic. Maybe I’m a fighter or a cleric or a rogue, but what really matters to NPCs is that I’m a follower of Helm.
Too be fair to the much maligned Tyranny, this is something the game does extremely well with faction because they make you choose your faction early. So you can role play as a rebel, disfavored, scarlet chorus, or anarchist and the world very much cares about that choice.
This is why NWN and NWN2 drove me nuts: great, I can pick all this stuff about my background and my patron deity and countless prestige classes, but no one seems to give a flying fuck that I’m a member of the Neverwinter Nine or a worshipper of the Red Knight. It’s mostly LARPing and I agree, the class stuff is mainly useful for min-maxing.
But anyway, a great RPG needs to be constantly reactive to some characteristic that the player has chosen. When it reacts to every characteristic intermittently, you end up losing the feeling of role playing.
You're getting close, but I'd say there is a missing aspect of this discussion that can only be appreciated from a narrative and development perspective. In principle, there's no difference between a Warrior, and a guy who specializes in Armor + Weapons + Strength + Constitution + martial talents. But a Warrior is a fantasy archetype, while a guy who just happens to have the same skills, is not. The lack of a label makes it significantly harder for a developer to interpret your character building choices in relation to a genre archetype, and therefore much harder for the game to react in an appropriate fashion.
To address this issue, developers give you other labels - backgrounds and narrative specific roles - to make sense of your identity, motivations, etc. This is all well and fine, until you realize that the roles and backgrounds the developers come up with aren't very compelling. An important reason why medieval fantasy has such staying power is because people actually do fantasize about being a Warrior, Wizard, Rogue, etc. - archetypes they've read about through fantasy novels, or watching films, or playing other games, while they don't particularly fantasize about a former Criminal who learns how to use Bows.
You can attempt to use faction as a replacement, but the two are not equivalent. A faction is an organization you
join; it's not your
identity. Most games do not treat factions as the central aspect of your character. They are, rather, narrative devices that you take on - or choose not to take on - as a matter of convenience and expediency. Being a member of Caesar's Legion determines what quests you complete and whose side of the conflict you're on. But it doesn't define your character; it doesn't establish your archetype.
The reason games like Assassin's Creed are popular is because they exploit the power of established archetypes. The concept of an assassin is edgy, it's cool, it's attractive to the masses. Change that to a different archetype, and the game collapses. There's a reason why we don't have Scholar's Creed or Peasant's Creed. People want to play assassins, and they want that experience to match the archetype they've always known. You can subvert the archetype, to a degree, but you can't fundamentally change it, and you definitely cannot ignore it. Indeed, that whole franchise is built around fulfilling people's expectations, and the same applies to most action and adventure games.
This brings us back to the need for structure. Not just any structure, but
compelling structure. The archetype classes can be thought of as tried and proven roles that we
know work, and which players can identify with instantly, as they already know them from other mediums or previous experiences. With respect to new roles that we ourselves create, whether they are a new class, or labels that we associate the player with in a system without classes, they are neither tried nor proven, and so their success is not remotely guaranteed. It is entirely possible to create a new role that people love - the Witcher franchise is an excellent example - but it's also entirely possible to fail. The price of failure, in this case, is a lack of immersion and apathy. When players don't like their roles, they don't care about them, and when they don't care about their roles in a CRPG, they don't care about the game.