The difference from what you describe, though, is that the need for the narration to establish immersion is not the same as what you were referring to in the post to which I replied.
In "story now", the principal source of immersion should be the fact that the situation is "thematically compelling". That is, the GM - in establishing the framing - is drawing on already-available stuff that everyone at the table is committed to. So the need to build a "word picture" by reference to sensation is less. The description of the setting provides a context for action rather than itself being the engine of immersion.
In Cortex+ Heroic this is formalised via the mechanic of "Scene Distinctions" - the GM may declare up to 3 scene distinctions (and may spend GM-side resources for more if desired) at the start of an action scene. In the session I GMed yesterday, the first scene involved the PCs - who had just crossed a frozen mountain lake - commencing their final ascent into the northern mountains. The scene distinctions were Chill Winds, Narrow Defile Between the Peaks, and Unpassable Snow. This, together with the description of their adversaries (a flight of wyverns, one with a rider; and the chieftain of the mountain folk), sets the scene.
Later on, after the PCs left the village of the mountain folk, them were confronted by the Earth Giant (as they knew they would be). The scene distinctions were Boulders Aplenty, Terrible Drops (which had an attached mechanic increasing the risk of PCs falling down them should they or the giant edge towards them, or break away their edges) and Clear Skies (the PCs had climbed so high there were no more clouds about them).
I have to say, this absolutely boggles my mind: that the mechanics of a game system
limit what the DM can describe in a scene!
Now in fairness these particular scenes you reference might not need much more to get the point across. That said, "Chill Winds" doesn't tell me if it's snowing (reduced visibility?), or bright sunshine (snow blindness?), or what...which means I'd have to ask. "Narrow Defile..." needs a direction, which when coupled with the time of day (particularly if it's sunny) would tell me whether the defile is well-lit or is in deep shadow at the moment. Little things like this - if you describe them up front players don't have to ask about them; and yes I'm saying it's usually better to describe in too much detail rather than too little.
These distinctions establish a context in which the PCs (as directed by their players) take action. They can also be operated upon - eg the PC sorcerer used his magic to dismiss the Unpassable Snow, and again to create eldritch walls and nets to neutralise the threat of the Terrible Drops. In an earlier session, a different PC was able to rescue villagers in need of rescuing by succeeding on actions to eliminate a Frightened Villagers scene distinction.
Though any of this could have (and likely would have) been done without the Formalized Mechanical scene descriptors, simply as part of the run of play.
Other systems handle this differently: but in 4e, for instance, one way to convey the really salient elements of a situation is via mechanical specification using the rules for traps, hazards, terrain powers and the like. The setting comes to life through its role in resolving the action rather than simply via description.
I'll give it this much: 4e does terrain well.
But, let's try an example. The party enters a study in a castle; they're here looking for a map and have decided that if the castle has a study that's the first place they'll look...and so they either explore until they get there or are framed straight there (no difference for these purposes). In either case, were I to go into detail my narration of the place might go something like:
"You've found what appears to be - or have been - a study. It's a small room - maybe 15' on a side - with stone walls, rug-covered floor and plastered-over ceiling; there are no other obvious exits other than the door you are in, and no obvious occupants. A leaded-glass window across the room from you looks out north across the lawn toward the gate house, and allows enough light in that vision here is not really a problem; the room is otherwise unlit. The place clearly isn't used often - dusty gray sheets cover most of the furniture, some of the shapes hinting at two chairs and a table beneath - and everything is covered with a thick layer of dust, slightly stirred up by your arrival. There are but two pieces of furniture not covered by sheets: an overstuffed armchair beneath the window whose sheet - on the floor nexxt to it - has clearly fallen off at some point, and a solid-looking wooden desk just to the right of the door. On this desk are a small box of some sort, an inkwell with what's left of a quill sticking out of it, an empty wine glass, and what might be some papers - it's hard to tell under the dust. The desk also has a couple of closed wide shallow drawers just below its top. A large tall sheet-covered piece of furniture against the west wall might be a bookshelf or a shallow wardrobe - again, hard to tell. What do you do?
So, no mechanics here, just a description in enough detail to forestall some obvious questions and provide lots of things to interact with. Would Cortex+ Heroic allow this, in this wording?
Once the source of "drive" or momentum in "story now" is appreciated, you can see the error of description in your (2). The PC tries because s/he has hope! And the player, rolling the dice, has the same hope. So there is not disparity at all, but rather congruence!
That's my (1). In (2) the player, while still having the hope, also has the meta-knowledge that a good roll guarantees success; which the player in (1) - just like the PC in the fiction - doesn't have.
Lanefan