But for it to be a "game," the player needs to be able to make informed decisions, learn from them, and apply that learning going forward (whether "do better" after doing poorly, or "keep it up" after doing well.) Inconsistency is one of several ways that you prevent that from being possible. When the rules are inconsistent--when the rules change without the player's knowledge, so the player can't reason from past experience to future events (within the limits of probability, of course)--you aren't actually playing a game anymore.
So, I'm coming to believe that engaging with TTRPGs as games, in the sense you're using the word here, is not, even in the context of classic wargaming norms, a given. Bear with me on a lengthy tangent to illustrate my point:
I've just gotten the kickstarter for Stefan Feld's reprinted city game series delivered and I'm looking forward to trying out Amsterdam and Hamburg later this week. The essential loop of Feld's designs in getting handed a pool of largely random resources each turn, and then allocating them to various different actions on a board, some of which provide immediate returns in victory points, some of which position the player to receive more resources in later turns, and some of which build engines to increase the efficiency of taking other actions. Player actions change the game (slightly, Feld doesn't make particularly interactive games) between turns, such that the emergent board state needs to be evaluated after each decision, to try and best optimize the outcome. Feld is a very classic German designer, so the goal of the games is to get as many victory points as possible.
So, the activity I'll be doing when I play those games is trying to make the best possible decision to acquire the most points, balancing the tactical concerns about the current board state, predictions about what my opponents will do and weighing the benefits of laying groundwork and building an enginge for future turns. The number of actions I can take is fairly small and I understand going in the approximate value of those actions (i.e. zoo cards in Hamburg offer no bonus resources, but have a better return on costs/point, or the building action costs generally 3-5 marks to perform), but what I don't know is precisely what the best choice to make each turn is in advance. Both because I don't have a perfect evaluation of value (how many marks can I expect to have around turn 3?), and because the board state is dynamic and will be different round to round. What I do have is perfect information about the resolution of each action (i.e. the build action's costs are laid out in the manual ahead of time, so I could in theory plan to have the cards/money I need to create a specific building at a specific point).
The whole point of this experience, the value I will draw from playing Hamburg, is that I will make a series of decisions and through the course of the game, get feedback on how effectively those decisions achieved the game's goal, acquiring the most points, and be delighted in watching my choices unfold, both as an act of personal expression in how I approached the game, a challenging optimization problem to solve, and a little bit of gambling joy in the uncertainty (albeit quite limited in such a eurogame). I will, after completing the game, look back at my decisions and note points that I should have chosen otherwise, and likely have some sort of post-mortem discussion with my friends, where we'll conclude that taking a zoo card before turn 4 isn't worth it or that one of us really had the right idea by playing to the cleric tokens or whatever we conclude. That series of decisions is what we're talking about here as gameplay.
TTRPGs as games (setting aside their storytelling/improvisational aspects, which have their own separate merits) can engage with exactly that same sense of "game" and the same attendant joy. In fact, they offer one further degree of freedom that can make it even more appealing: instead of the fixed goal "acquire the most victory points" I am given the opportunity to set my own goal, usually determined by the narrative/storytelling functions from earlier. Even better, these games are unbounded so that I can set subsequent goals or continue playing them after failing to achieve one.
D&D's gameplay (again, in the very specific sense I'm deploying the term here) is, generally speaking, not very good. Decision making is often trivial, in that the optimization case for any given decision is so clear that the decision isn't particularly interesting. Other times, decision making is too opaque, because the results/mechanisms of resolution for any given decision aren't known, such that the player cannot choose well. Even worse than that, in some cases (particularly situations where spells can't be brought to bear) the game produces no decision making at all, and characters are essentially just rolling reactive skill checks to events as they unfold.
However, I'm coming to believe this loop, is not actually desirable to a chunk of people playing D&D. Or, perhaps more accurately, it's significantly less important than the improvisational/narrative elements, or, that it is less interesting to them than engagement witht he fiction itself. If your primary engagement has nothing to do with making interesting decisions (in again, the specific gameplay sense) but rather in the narrative outcomes of those decisions, then it doesn't really matter how the rules to resolve the situation work. Or, it might not be a question of gameplay at all, the "play the world, not the rules" concept in action, or the sorts of tables where players are generally discouraged from knowing the broader resolution mechanics. The players are not encouraged to do gameplay in the sense I used it above, but to make decisions without information, striving to emulate how people actually do this in real life all the time (albeit in a more risk-free environment).
The thing is, it's not a strict hierarchy between those three things I mentioned above. I actually care quite a bit about the "narrative" elements I just talked about stripping away and I've spent a lot of time thinking about how rules can use that gameplay loop to harmonize the decision making of a person in a world and a player playing a game, something I think is generally desirable in TTRPG mechanics. The point remains though, that if you want gameplay as a thing, with the attendent benefits of learning, crunchy choices to chew on and mull over, lines of play that can be considered in retrospect and analyzed for alternatives, then you need rules everyone can agree on. Otherwise, the line of play you analyze after the fact will always be "what could I have said to make the DM do something differently," or, as Ezekiel said, you're not playing a game, and there's actually nothing you as a player could have done to do better.
