So the goal is to figure out the direction of the Dawnmote using the approach of observing the environment and making deductions. Assuming there's an uncertain outcome and a meaningful consequence for failure, this could be an ability check.
Yes, that is the gist of it. I didn't go deeper than that because (a) this is a quick hypothetical for conversation and (b) I figured the procedural elements were intuitive.
The player has no ability to declare the location of the Dawnmote in D&D 5e. That is for the DM to decide. If I did not previously establish in some way where the Dawnmote was, then I'm usually inclined to just say "Yes, and..." But if I'm setting up a specific challenge, I will have framed this already in a way that the player's action declaration would not be appropriate as it would contradict existing fiction.
So there are multiple components to this.
1) If this was a reasonably high resolution, map-driven hexcrawl, then the Dawnmote would be a preconceived point on the Feywild map. Therefore, the default answer to the question outlined (regarding the Survival check) would already be answered prior to the player's action declaration. So the GM would have an answer to the question the player proposed. If the answer is possibly "yes", then the test would be about (a) can the survivalist suss out the correct direction to the Dawnmote at all (pass/fail result) or (b) can the survivalist suss out the correct direction to the Dawnmote without triggering some form of obstacle (due to time expended or being a presence in a dangerous area or something else).
2) If this was a sandbox with a very low resolution map and this Dawnmote thing was just recently adlibbed (meaning the GM was sorting out, but not firm on, the precise location of the Dawnmote in this Winter Fey domain the PCs have emerged in) then, effectively, the player's input into the situation is a form of vetoable content introduction. In 5e, the GM can "say yes", "say no", or "disclaim authority to the dice."
As we should always do in these cases, we're taking for granted that the outcome of neither (1) nor (2) contradicts prior established fiction.
I'm not against it per se, but I don't need someone to fail a check to make that thing happen. It's a little more disconnected from the action declaration than I would prefer in this game.
Of course you "don't need someone to fail a check to make that thing happen." That isn't the point of the exercise.
The point of the exercise is "this moment of play happens....what now?"
I'm curious about your thoughts on it being disconnected in 5e though. This is always an interesting one (which is why I asked people about their thoughts on the complication).
This touches upon my 3 year old thread (very interestingly and unfortunately destroyed due to the board's disintegration and deus ex machina) "DC 30 <etc>" where we discussed the nature of GMing 5e, genre logic vs internal causality logic, etc etc.
When I look at that hypothetical play excerpt above, I think its pretty mild in terms of the tension between genre logic vs internal causality logic and should satisfy both.
1) Sussing out a thing takes time (even if a marginal amount). Time on the ice means the danger of the substrate becoming unable to load bear the PCs and, therefore, it possibly failing.
2) Time + the presence of new environmental inputs (the PCs and the sounds of their interaction with the ice) = local denizens (benign or predatory) might be attracted.
By my estimation, both forms of logic are satisfied.
Which do you think isn't satisfied and why?
What would be more satisfying (or "connected" as you put it)?