Actually, the comments you're referring to weren't made by me (judge Gamma) as I didn't judge In A Grove. But I'm willing to weigh in with my own opinions.
The game's deconstruction of role playing is a bit abstruse, perhaps a bit too much so to be informative. I think I get that each segment of play is supposed to represent a player's "individual imagined space" in a typical role playing game, while the written list of facts represents the shared imagined space of a typical role playing game. But when the game is played, the narration of each segment of play is, of course, actually a shared imagined space. Each player participates in creating and comprehending it. Therefore it's easier to regard the narration of a segment as a shared imagined space that later gets tossed out and redone (which doesn't represent anything that happens in a typical role playing game) than as an exhibitiion of the "current player's" "individual imagined space." A game in which narrated information that is not immediately overruled or contradicted does not become part of the shared imagined space is so different from the normal "process of role playing" that it's hard to read it as a deconstruction of it.
Partly, how what the game says about role playing would depend on how much, and in what ways, the individual segments end up diverging. And that's an area where the game system might not be suffificent to get the results you had in mind -- depending on what you did have in mind. If the players start out with Rashomon in mind, then they might constrain themselves enough to generate Rashomon-like results. But the game system doesn't assure this, so to the extent that you want the game system to assure it, it's incomplete. There are two important missing pieces.
First, in Rashomon we see different characters portraying different versions of the same event. But this doesn't happen at random or for no reason. There's a lot more to the differences than, say, bad memory or poor observational skills can account for. It's pretty easy to interpret, from the differences we see, the various self-interests and self-images that motivate each character to report the "facts" in the way he or she does. The bandit wants to appear blameless, the samurai wants to appear heroic, and so forth. Now, you seem to be taking for granted that the players of In The Grove will take on these or similar motivations on behalf of their characters. But why should they do so? For instance, why should the players, or even the samurai player in particular, care whether the samurai appears heroic in his or any other character's narration? They might do so in order to deliberately imitate the story or the film, but the game text doesn't instruct them to do so and doesn't appear to reward them in any way for doing so either.
Second, the handful of established facts through the course of the game doesn't appear sufficient to prevent the different versions from being either wildly different, or practically identical. In the film (I haven't read the story), the similarities between the different versions are just as important as the differences. All four stories follow the same basic outline, incorporating the same or similar sequences of events, such as who arrives on the scene when, and who comes into conflict. If this were not so, then instead of seeing the recountings as four differently distorted persepctives of something that actually happened, we'd be forced to see it as four different fantasies, delusions, or outright lies, which is much less interesting.
The game rules don't enforce any such parallelism. The established facts nudge in that direction, but they're very limited in number and in scope. There's nothing to prevent the bandit from inventing a whole army of walk-on characters to blame for the murder and then, unless certain particular facts are chosen, all the other characters not mentioning those characters at all. There's also nothing to stop them all from agreeing on a single tale that meets all the motivations they're playing on, such as a murderous attack by a powerful outside enemy that leaves the bandit looking innocent, the wife loving and dutiful, and the samurai heroic. Again, the players might refrain from these extremes and spin tales that are both appropriately similar and appropriately divergent in an attempt to imitate the story or film. In the game text you appear to be taking for granted that the players will do this. But there's nothing in the game rules themselves to assure their doing so, or to guide them to doing so if they're not familiar with the source works.
Of course, I can't tell if these are the same reasons why other judges might have regarded the game as incomplete or insufficient to accomplish its apparent goals. But they're some of the concerns I have, on reading the game.