There's some interesting stuff in this thread, that I'm still catching up on.
To me, this seems highly contextual. If the construct is simply a device for introducing a certain proposition into the fiction as something for the PCs to entertain (so the functional equivalent of eg finding a diary entry, or a carving) then the idea that there is nothing for the PC to discern seems plausible. Depending on context, I might expect an Insight check to be one way of working out that the construct is just reciting pre-established words.
But if the construct is itself an element in a social challenge, and the issue of its truth-telling matters to the resolution of that challenge, then I personally find a GM-fiat no a little railroad-y. (But at some tables perhaps constrcuts can't participate in social challenges, and are really just like Magic Mouth spells?)
This doesn't seem right. If the player doesn't declare any action for his/her PC that would suggest ascertaining the truthfulness (or otherwise) of the NPC, then how can the player infer that the NPC was telling the truth from the fact that no check was called for?
OK, but (a) don't they have to declare some action to trigger the skill check?
I have trouble following this - how does an Insight cjeck maintain mystery and doubt? Only if you don't tell the players whether or not the check succeeds - but in that case, what is the check adding to the game? I mean, the player can be uncertain if no check is made.
In my game, if there shopkeeper is telling the truth then I want this to become clear so that play moves on to something more interesting.
These posts make it seem like the function of the checks is to "skip the boring bits". Although, as per the discussion of Insight checks maintaining mystery, it's not clear exactly how this will work if players aren't told whether or not their checks succeed.
This reads like, or at least fairly similarly to, "say 'yes' or roll the dice". That's a methodology that used to be extremely controversial on these boards.
It's also my preferred way to deal with "boring bits", or bits where nothing significant is at stake. Let's cut to something everyone's interested in!
My own preference in action resolution - which goes with "say 'yes' or roll the dice" - is that on a success it's the player's narration that becomes part of the shared fiction, and on a failure it's the GM's narration of the consequences that becomes part of the shared fiction.
The overall idea is that (1) we establish something that both GM and player are invested in; (2) the check to find out what happens is framed and made; (3) on a success it goes as the player wants, on a failure as the GM thinks will step up the pressure.
In your example of player narration you have the player narrating the moss that his PC slipped on, but don't elaborate on what the consequences of failure are or who establishes those. If by "player narrating own failure" you're talking more about what form the immediate event of failing takes, rather than what flows from it, then I'd see that as a shared GM/player/table thing.
"These posts make it seem like the function of the checks is to "skip the boring bits". Although, as per the discussion of Insight checks maintaining mystery, it's not clear exactly how this will work if players aren't told whether or not their checks succeed."
An attempt is made, s roll is made (possibly with advantage if the effort warrants it by bringing in beneficial things other than what normally comes with the effort) and what happens as z redult is narrsted.
A Perhaps, you get a status quo, nothing changes.
Perhaps you get some positive gains with additional noticable drawbacks as well.
C Perhaps you get some positive gains without any noticable drawbacks.
Note that in my gsme you also get included in the narrative a degree of confidence worked in. That may or may not tie in with the ABC result.
A shows pretty much a fail.
B shows a partial success partial success with hidden setback or a success outright.
C shows either a fail with progress and hidden setback or an outright success.
But as you can see, of the three cases two give you something new to work with, a positive change and possibly a negative change. The good news is, even a negative chsange can lead to positive results or even a fail.
Consider, if an attempt to treat a disease fails in spite of a high confidence effort, the player knows and the character knows "oh crap, I need to go another route." (Just like a player missing on a to-hit roll of 19 realizes he has an attack roll problem and shifts to other options.)
So, maybe they bring up the spells for cures.
If my failed interrogations leads to still being suspicious, but not conclusive, maybe I move to searches or leverage? Maybe my failed intimidation leads z shopkerp to call the attempt in and thugs come to us... clues-with-feet.
As I often tell my players, often as not, what is bring contested is control of events. More often when the pcs succeed they dictate results and to varying degrees the aftermath. When they fail, others do - often in ways not too favorable.
