Why? What if you have a game that is entirely player skill based, with no character stats, but the gameplay is so deep and varied, that it would take the player years to master it all. In that case, customization can be done by working on your (player) skills, instead of developing character stats.
Then its not an RPG. Its an action game.
Mind you, im not reducing the RPGs to stats alone.
Most games some kind of "stats" of the character you play. Or a "role". Only in RPGs is that role made more tangible. By limiting gameplay options though character abilities that the player cannot directly override, but can shape and evolve. As i already said...
An RPG is a game where character abilities affect tactical in game options, while the player can affect only meta and strategical options.
Its not a hard specific boundary, which is why there is a whole spectrum of various RPG games.
I understand your perspective, it's a fairly popular one here on the Codex, but my point is, it's a very arbitrary definition. It's based on the historical definition of RPGs, which comes out of PnP games. In those, you HAD to restrict roles by stats or explicit roles, because otherwise PnP games would grow repetitive and stale. Roles forced player to react differently and led to different situations.
But in the context of modern games and hardware, I would argue that definition is outdated. Developers ought to be aiming for deep simulations of various RL systems, which will create and support roles in a different way. Imagine a game that has a combat system similar to real world fencing, and a stealth system similar to Thief but even deeper, and other systems. So players would have to specialize, because mastering each one at the same time would not be practical. At the same time, because they are player skill based, the gameplay would be a lot more interesting than the typical stat based RPG gameplay (stat-check, win/lose, etc).
I suppose what you are talking about would be theoretically possible. The thing is, there should be, as I mentioned above, some kind of system that allows for the formation and expression of distinct and discrete character configurations. In this case, the system would presumably be grafted into the gameplay engine itself, instead of being superimposed on it as a kind of abstract layer in the way of traditional character systems. I also suppose it would be a learn-by-doing system where in-game actions are the means not only of expressing but also of building and developing your character. The important point here would be that of character configurations being
discrete. The system that would have to undergird this sort of game would have to include some provision that prevents the character from becoming a master of all trades.
In real life, we cannot specialize on everything because limited time and stamina do not allow it. The development of character proficiencies in games is much faster than it is in real life, so in order for the kind of simulation that you are proposing to work, the pace of the game would have to be slowed down considerably or some kind of systemic limitation would have to be introduced. In the first case it could lead to a very boring and tedious game. In the second case, the limitations could end up becoming so artificial that they would end up returning to elements of the traditional system of barriers based on an abstract layer. A case in point is Skyrim, a game that does not use a class system and implements a learn-by-doing system. In that game, the structures in place that limit the master-of-all-trades syndrome are a (simple) attribute system, a point-purchase perk system, and level-scaling. The first two are classic elements of traditional abstract character systems and the last presupposes level, which is also a feature of these last, not to mention it is aggressively anti-simulationist.
Ultimately, the question is, how would you propose should a game of this sort be designed without either resorting to traditional character systems in at least some way or slowing down the pace of the game in an unacceptable way?
I wouldn't say they are what creates quality either, since there is plenty of rpg games where exploration, combat and the story are done badly.
Making a very good quality RPG or any game is another specific issue on its own.
I would say this is where we disagree, then. The RPG template/triad that I mentioned above is not derived analytically from the definition of RPGs, but historically from the design style of the old school of RPGs, of which the clearest representatives are early Wizardry and the Gold Box games. It is true that historical crystallizations of a thing do not define it, but they provide us with the empirical data that allows us to understand how the ideas and concepts that we are discussing translate into practice and are implemented, and thus what works and what doesn't. I suppose it then becomes a matter of interpretation of history.
The way I see it, tactical RPGs and dungeon crawlers represent the epitome of RPG gameplay that is complex, engaging, challenging, and stimulating, and the genre declined from there. The cRPG Renaissance was a transitional period in which quality games were still made, but only because the lessons of the Golden Age had not yet been forgotten, and combat and dungeon design were still competently executed. The way I see it, games like Fallout and Baldur's Gate II are heavily indebted to the legacy of the classics for much of what was good about them, and the biggest difference between them and later titles like the Troika and Obsidian games, is that the foundation of quality gameplay had been eroded beyond repair in the case of the latter. This state of affairs was a result of the cRPGs Renaissance's emphasis on story and reactivity that eventually resulted in a neglect of the crafts that used to provide the gameplay foundation of RPGs.
The template may well be historically-contingent and not essential to the nature of RPGs, but it still furnishes the best extant example of quality RPG gameplay, and should thus be the standard by which RPGs are generally judged. This does not mean that I am against games that try something different - like AoD, Morrowind, or Witcher for example - but my view is that the paradigmatic cases of an RPG that should be considered the golden standard are tactical RPGs and dungeon crawlers, and when these are no longer considered as such, the rest of the industry suffers.
EDIT: One might argue that there doesn't have to be a single paradigm, and I would agree, but there has to be a
reigning paradigm, and make no mistake, there always is a
de facto ruling paradigm. There have been at least three reigning paradigms in Western RPG history. The first was triad of exploration-combat-customization based on early DnD and represented by Wizardry and the Gold Box games. The second was the narrative and reactivity-based paradigm of games like Fallout and Planescape: Torment. The third was the Biowarian paradigm of cinematic RPGs. We might well have already entered the era of the open world RPG thanks to the success of Skyrim, but I think Witcher 3 shows that the Biowarian impulse is still strong. In any case, of all of these paradigms it seems to me the most healthy and universal one was the first, and that is why I consider it the exemplar for RPGs.