I think we're talking past each other.
How strongly you define a character at the outset doesn't in itself make the game hard or easy to minmax. You can make suboptimal choices later on just as well, you haven't tried all the endgame spells and abilities before you get them after all, or seen what kind of monsters you'll be facing later. Rerolling your party 2 hours in vs. 10 hours in, which is better and why?
I don't object to "minmax" depending on what the term means. Here is what I think is great gameplay: being confronted with a long-term goal and making strategic choices for how to achieve it; in turn making tactical choices for how to overcome the particular obstacles your strategic plan faces at any given time. Within that framework, I think resource management (which means how to sacrifice the least to achieve the most,
i.e., how to minmax) is absolutely critical or the choices don't matter. Here is what I think would be my GOAT RPG: one that has that kind of gameplay married to compelling narrative. I'm not sure any RPG has ever had strategic choices (AOD, maybe FO, I dunno), but some have had good tactical choices married to compelling narrative (PST, AOD, FO, etc., etc.).
I generally like complicated character creation systems while I'm fiddling around with them -- precisely because they are often the only moment in RPGs where you get to do fun stuff defining your character that goes beyond just incremental advancements within an already set framework. But I don't think complicated character creation systems are "great gameplay" in what I'm calling to Mustawd's chagrin "narrative RPGs" (which is to say, RPGs that are designed to be played start to finish and won on the first character build, often aren't ever played again, and reward mostly through the arc of the story rather than the powering up of the character/party). The reason is that while they entail the kind of resource management decisions and neat choices that
could be part of great gameplay, those choices are being made in a vacuum -- the player has no idea what the long-term goal is, or even what the short-term obstacles are. It is true that if you've been around the block enough time you know that the long-term goal is to overcome some final boss and that the obstacles are things like popamole fights and dialogues that can be won by picking the option with the highest diplomacy requirement. But that is to say, character creation "works" only when the gameplay is stupidly predictable and bland.
Because an RPG can't start (IMO) without the player having
some kind of character, I think that some level of character creation at the outset is a good thing. You basically give the player a basic narrative hook, and then choose an archetype (like the outset of AOD). But I don't see a good reason why all of the minmax fiddling should occur at the start (in a vacuum) rather than as you go along. Deploying your resources in respect to obstacles is tactics, deploying your resources in pursuit of goals is strategy. Deploying your resources in a vacuum is just dress-up. I get that a lot of people enjoy the dress-up aspect of RPGs, it's just not great gameplay IMO.
Anyway, I think that learn-by-doing [EDIT: by this I mean "improve skill scores by doing"; "learn by doing" in terms of the
player's experience is great!] is terrible (this is off-topic), and AOD showed that people don't particularly like having character-choices-in-response-to-obstacles accomplished by point hoarding, so you'd have to think about what the best mechanisms would be. IMO part of the answer is that it's dumb to define 8000 variants of "talky" character on a statistical level, rather than treating the character's talky-ness as a given, and make the variation happen in response to environmental choices (alliances forged, information gathered, gear equipped, etc.). In that way, you can achieve min-maxing in the face of obstacles without having point hoarding.
If you want to make a game easy, include different difficulty settings. Then a player won't need to reroll their build even if it's shit, when playing or normal (casual) difficulty. Then even the casual player, if they really like the game, can replay it on a higher difficulty if they really liked the game, and put their newly founded understanding of the game into use in making a better build than last time.
This is why I think we're talking past each other. What I'm saying has nothing to do with ease/difficulty of play, and has to do with where you put the moment of resource deployment.
If you push all the character design from the start until later though, a consequence is that the early game gets much more samey for replaying. Much more fun if you can tackle the first dungeon already with a new party, and it actually plays differently than last time with another party.
This is a fair point, and it is why I think in games that focus on replaying (like crawls, rogue-likes, etc.), the analysis is different. But I also think that this problem is best addressed [EDIT: in narrative RPGs] as AOD and Dragon Age (!) did, by having the player making archetype choices at the outset that affect the early part of the game, rather than having the player walk through the same content with marginally different treatment: "This time you talk your way past the bandits at the bridge rather than blowing it up, sneaking past them, or killing them! What variety!"
