To me, that's because if the goal is to produce a narrative during play, the game mechanics should allow us to play, effectively, in real time. That is, you should be able to get through a two-hour movie's worth of game in about two hours. Of course, a bit of extra time is allowed for food, bio breaks, a bit of table talk, decision making, etc. But if it takes you four six, ten hours or more because of things like detailed conflict resolution mechanics or combat mechanics, the game utterly fails as a narrative game. If the line "30 minutes of fun packed into four hours" can be applied to the game, it's clearly not a narrative game.
In his "story now" essay, Edwards notes thatI'm going to offer up a different take.
A narrative game is one that will naturally produce a narrative when played. Here I'm using the standard, real-world, writers' definition of a narrative. It's a narrative game if, when you play it, you end up with a story that could be reformatted slightly to be a novel, a comic book, a TV series, a play, or a movie. If you have to add in tons of things like plot, theme, coherent story structure, etc...then it's not a narrative game. If you have to remove tons of things like redundant scenes, superfluous characters, dead air, side quests, dead ends, etc...then it's not a narrative game.
The two best examples we can look at are Record of Lodoss War and Critical Role. Compare the RoLW replays with the manga, anime, novels, etc that were produced later. Also compare the CR actual play with the comics, cartoon, novels, etc that were produced later. They both had to add in a whole lot and remove a whole lot more to beat their games into something like a story or narrative.
Things like Dimension 20 and Worlds Beyond Number likely come much, much closer to producing a narrative just from play, but that's down to the people involved all being professional improvisers and storytellers. They're professional storytellers who happen to be using an RPG to tell a story rather than gamers who are trying to produce a story from their game.
To the best of my knowledge, there are no games we'd recognize as tabletop RPGs that do that. Some games produce a nearly endless string of complications that must be dealt with, for example games that are PbtA and FitD. But that's not all you need for a narrative. Games like Fiasco try with a more formal scene and act structure, but it generally produces a separate narrative-like thing for each player rather than a single, unified narrative from the whole experience.
the key to un-premeditated artistry of this sort (cutup fiction, splatter painting, cinema verite) is to know what to throw out, and role-playing does not include that option, at least not very easily.
The lack of editing is a big issue in RPGing.
But I have had RPG sessions that get close to you "2 hours of story in 2 hours of play" desideratum. Cthulhu Dark and Wuthering Heights one-shots; Prince Valiant; and almost, but perhaps not quite Burning Wheel; and falling even a little bit further short, Torchbearer and Classic Traveller. What helps is both relatively speedy resolution (eg versus tests in BW and Prince Valiant) and having consequence narration and scene framing tightly integrated (which all of these systems have) - this latter means that a lot of time is not spent working out details of consequences that are relatively insignificant from a thematic point of view (eg damage tracking), and likewise a lot of time is not spent getting everyone onto the next "page"/"scene" of action.
RPGs that have no hope of doing it, in my experience, include D&D (any version - 4e is weird here because skill challenges belong in my previous paragraph, but its combat doesn't), Rolemaster, and Cortex+ Heroic (I love the system, but resolution takes too long to achieve your goal).
Have you played A Penny for My Thoughts? It's not a RPG. I have always called it a "storytelling game". I think it is closer to what you are calling an actual narrative game.Most designers have things by the wrong end, I think. They're trying to add story elements to games, rather than adding game elements to creating stories. For me, doing that latter would get you far, far closer to making an actual narrative game.