Well, the PC doesn't know anything about dice rolls. So that is already a player/PC divide in any system that uses dice to resolve action declarations.
In the end, I can only report my experience: rolling and hoping correlates strongly to searching and hoping. The player knows the action isn't futile (because success on the dice is possible); but the PC must at least believe that searching isn't futile, or else s/he wouldn't be doing it.
Not quite.
In (1) the PC has no idea if the search is futile or not and on failure still has no idea if the search was futile or just badly done; ditto for the player (assuming these rolls are hidden, which for just this reason they should be). In (2) what you say above is true, and that very player knowledge that the action isn't futile vs. their not knowing is exactly the difference I'm both getting at and saying is bad: it hauls the player out into the meta-game no matter how hard she tries to resist.
More description is permitted, but it will just be colour. Whereas the scene distinctions are not mere colour.
Well, it doesn't matter to resolution. If you think the Chill Winds are hampering your PC, you can declare as much (and earn a plot point). When you describe what is going on, you might refer to snow being driven by the Chill Winds, or to the winds themselves, as you feel fits your conception of the situation. No one else at the table is going to contradict you.
Alright.
What if a player (intentionally or otherwise) forces you to introduce a fourth element - say, she tries to climb one of the cliffs along the defile, so now you have to worry about a Steep Cliff issue. What then?
Again, this doesn't matter to resolution.
At the start of the encounter described in my earlier post, the berserker identified and established a defensive position for the PC seer and himself - he delcared that he was moving some rocks into place against the mountain wall (thus using his Godlike Strength as the biggest die in his pool). In my mind's eye, this was on the left looking at the wyverns flying in (because that fitted where those two players were seated at the table relative to me). I don't know how the player envisaged it in detail, but that didn't matter.
For tactical reasons it certainly would matter!

I've had characters die in the past due to just this: where a DM describes enough to let me imagine a scene or element, and I act based on my envisioning of what was described. Problem is, his envisioning is different and his descripton is just vague enough to allow either interpretation...
What follows is usually a fearsome argument.
This is why I use minis, so that everyone has roughly the same idea of how the various moving parts spatially correlate in situations like this. For this set-up I'd probably make each square represent 50' or so, and place the various minis in a representational manner; it would also allow me to more clearly describe the course the wyverns were taking as they flew in (such things
always get misinterpreted IME if just described or done TotM).
The number of RPG tables which worry about the location of the sun, and hence (eg) the difficulties of shooting arrows at backlit foes, or the chance of momentary blindness from looking into ths sun, is - I assert - very very small.
I've never seen a table that wouldn't ask about the level of lighting in the defile if not told, as if it's in deep shadow who knows what could be hiding in there.
Which leads me to this: I'm wondering now if I've got a different view of this scene in my mind as "player" than you do as "GM".
In Cortex+ Heroic, that risk is all subsumed into the Narrow Defile scene distinction.
Which, if "Narrow Deflie" doesn't have a clear definition somewhere in the system rules, is going to lead to a bunch of questions every time to draw out more specifics...at least, it would if I was playing.
Well, it discourages it.
I would say something more like:
You come into a small sunlit study. The scene distincitons are Stonewalled Room, Sheet-covered Furniture and Dust-covered Desk.
Oon this approach, if the players look for things on the desk - papers, boxes, whatever - then, given that we're talking about a hunt for something, that would (in mechanical terms) be about creating assets or resources. It probably wouldn't be built into the situation by the GM.
If the GM wants to make the box a feature, then an alternative would be:
You come into a small sunlit study. The scene distincitons are Sheet-covered Furniture, Dust-covered Desk and - on the desk - an Intriguing Box.
That would be followed with a boatload of questions from me.
Also, doesn't the system limit of so few distinctions - in a scene that might have many - tend to overmuch lead the PCs by the noses to where they need to go? For example, you only mentioning as distinctions the Furniture, Desk and Box immediately tells me-as-player I can ignore the rug, the papers on the desk, the small chandelier*, the fireplace*, and the faded portraits* on the walls as they've all just been defined as irrelevant. My PC, however, wouldn't know this.
* - not included in my original description but I throw them in now as things that could easily be in such a study
And another example of misinterpretation, in this case of a detailed description: in my narration I state the window looks out to the north, meaning that while the room is daylit it's unlikely to actually be sunlit unless it's early morning or late evening in the summer (and if such was the case I'd have amended the narration to suit). Sunlit vs. daylit makes a huge difference to the ambient light level in the room; only being daylit means there'll be some dark shadowy corners, and with all this dust if someone lights a torch or candle during their search...
Lan-"yeah, there's a reason I mention the dust three or four times in that narration: it's the room's hidden hazard"-efan