Well, as a player, you could conjecture what your PC would do in the face of an almost unstoppable humanoid opponent of this type. Sure, you are not referencing information which you, the player, possess, but you could still at least try to imagine what the PC would do.
<snip>
Surely there is something going on which is indistinguishable in one sense with meta-gaming, yet in another sense it might also be compliant with what the player would choose the character to do even if he didn't know about trolls.
This is all true.
What's going on with Ron Edwards and The Forge in respect of stance? Well, they're not really trying to analyse the nuances of constructing a character while playing B2 - from their point of view, that's been covered under Pawn Stance with maybe a bit of Author Stance. There's an ongoing discussion on the various RPG forums/newsgroups which has coined this idea of stance, and Edwards is trying to develop the idea and use it coherently to analyse the play that he is interested in.
At that time - ie 2001 - as far as "mainstream" games are concerned they're trying to work out what's going on in, and what are the variations occurring in, such systems as Champions, RuneQuest, WW/storyteller, and the AD&D 2nd ed settings. A game like Over the Edge is obviously big on their radar, but if you read OtE the GM advice and overall tenor is a strange mix of player-driven characters being shoehorned into GM-driven setting with no very coherent account of how this is meant to work. (That's not to say that Jonathan Tweet was running OtE as a railroad; but there was no established way of expressing these techniques in a RPG rulebook.)
I would say that, at the heart of what they're trying to distinguish
being true to the character in action declaration (actor stance) and
using the character to drive play in a certain direction by way of action declaration (author stance). This is why there's the attempt to clear away underbrush that crops up so often in discussions about player-side RPGing (like 1st person/3rd person, or IC/OOC, which is still the first thing that will come up on a thread on these boards about player-side roleplaying techniques).
In a game in which story doesn't even matter except perhaps as a byproduct, and in which the idea is to win - and B2 would be a paradigm of that - then I think it's easy enough to say that most action declaration will be directed at winning, which is clearly a player priority, hence pawn stance and we're done. The odd bit of actor stance (eg the elf playing pranks on the dwarf when nothing else is at stake) is simply not that significant to the overall analysis.
And once we get to "story"-focusd D&D play of the post-DL, 2nd ed era variety, then I think the assumption is that the GM will establish the key player motivations (by setting backstory, policing alignment, all the standard techniques) and players are expected to adopt actor stance within that context. I think this is borne out by the AD&D 2nd ed text that I quoted a little bit upthread.
For the troll example to fit neatly into this conception, either the GM tells the players that their PCs know about trolls, or tells them that the PCs are ignorant. Then the player plays his/her PC as appropriate (perhaps with a significant degree of awkwardness or frustration if s/he knows the answer but has to pretend not to). There is no expectation that this sort of play will produce what [MENTION=6696971]Manbearcat[/MENTION] called "discovery" - as opposed to fidelity to the motivational scheme established by the GM.