The fundamental question is, is whatever mechanic we're devising actually fun.
And you can address that in a lot of ways, but I like to drill down to the fundamentals. Repetitive actions should be inherently fun on their own, even when stripped of their narrative context.
Rolling dice, on its own, is inherently fun as a tactile experience, and getting the numbers you want, based on which ones the game defines as valuable, is part and parcel to that.
But just rolling is pretty simplistic, and doesn't provide any depth, so then we start layering aesthetics and secondary mechanics, in order to create a more dynamic mechanic. And thats where we want to be when we're discussing if a mechanic is actually fun, as opposed to worrying about step counts.
Rolling to hit is ostensibly fun, as there's an inherent excitement from dice rolling when you hit the target number. In practice, however, only certain kinds of games actually maintain that fun in the context of their aesthetics, and RPGs are generally not one of them. While rolling a hit can be fun, and regularly is, we have to weigh that against how often it isn't fun to miss, and what the game does to mitigate that issue.
Thats when we start running up against ideas like failing forward or even simply removing to-hit mechanics.
But, I personally think those arent satisfactory either, as they both still rely on target numbers which are what I think is the unfun element in that experience. PBTA style moves ostensibly remove them in how they work, but in practice, the aesthetics of success versus failure still produces that unfun feeling, especially if as is common in those games, missing TNs just leads to a lot of genre-straining drama rather than a narrative intuitive outcome.
Removing TNs entirely, and retooling the aesthetics of the results, would fully remove the unfun element. That doesn't mean what then gets implemented is automatically fun, but thats why game design isn't easy and takes a lot of iteration, testing, and analysis.