Please do. I agree that some of the mechanics of how an RPG story gets woven is different from how a book or movie script gets written, but good storytelling is good storytelling, no?
Well, yes and no. Good storytelling in a movie or novel is, I think, most of the time* quite different than good storytelling in an RPG. In a movie or novel there's a predefined plot, plot-protected key characters who only do anything if it serves the story, and almost invariably the heroes or sympathetic characters win in the end; a end which is predetermined by the author. Very static.
In an RPG there's often little or no predefined plot, key characters are not plot-protected (or shouldn't be), and the heroes/sympathetic characters (in this case the PCs) may very well lose. Not only is the endpoint always in motion (if ever defined at all), the next chapter and even the next sentence is always moving based on what happens right now...which isn't pre-determined* either. Nobody knows how good or bad the story - or the telling of it - is going to turn out until it's all looked at in hindsight. Very fluid.
* - the exception, of course, is those instances where a DM is running a hard-coded railroad in which the PCs just play out her mostly-pre-scripted scenes: a novel disguised as an RPG. Henceforth I'll ignore this type of game for purposes of this discussion.
So, how does any of this factor into a discussion on metagaming? Well, tangentially at best; when asking the question "does metagaming make for a better story?" or some variant thereof. My answer would be that sometimes it probably does and other times it probably doesn't; but even in those instances where it does any story benefits in hindsight are outweighed by the loss of enjoyment in the actual play of the game at the time.
Why does it have to be "most likely"? (And what does that even mean?) Shouldn't it be "most interesting thing they could plausibly do"?
Not necessarily. Another difference between a novel / movie and an RPG is that in an RPG you're telling the story from the inside, as it were...living it out from within using the perspective of your character rather than objectively observing and chronicling it from afar using a perspective of omniscience. As such you quite likely don't know what your "most interesting" next move might be but you probably do know the next move that flows most naturally out of both a)your last move and b) the surrounding environment, in a vaguely logical sequence: this is your "most likely" next move. An omniscient observer might know the most interesting move but the character is almost never omniscient and thus needs must use what limited information its own point of view gives it. Otherwise the character becomes a pawn.
And if you-as-player think there's more information to be had from your character's point of view than you've been given, ask the DM to either provide it or explain its absence; and such explaination can legitimately include "there's no way you could know that" and-or "you rolled to perceive that, and this <true or false> information is what you perceived".
Oh, sure, the player's decision might screw up a plan that the DM had and therefore might not lead to the best story. But we have to at least trust that everybody...players and DM...are trying to weave the best story they can.
Assuming that a "good story" is everyone's top-of-mind end goal. I know as a player mine often isn't...in character I'm just going to do or say whatever seems to make sense to that character in the moment, based on what it knows and how it views its world (alignment, established personality, etc.), and let the story - for better or worse - sort itself out later.
And maybe that's the key difference between the two viewpoints here: our assumptions about the other people at the table.
Or, perhaps, our own ideas as to what we want form the game and how we're willing to try and achieve it?
Lanefan