Okay, I've done a bit of organizing thinking on this, and here's the rub: If you're going to use goal and approach as a method, you must present a game that offers handles to the players to propose goals and approaches. In short, yes, it's part and parcel of the method that you must change how you present situations. If you're only going to call for rolls for things that are uncertain and have a risk of failure, then it's incumbent on the DM to present uncertain situations with consequences of failure. This doesn't work if you just have hallways that may or may not be trapped, as what happens is that players are now asked to do repetitive goal and approach declarations and this gets old fast. It's easier handled in an ask-for-roll approach as the entire exercise in the fiction is abstracted and pushed off onto the mechanics to get past this repetitive play and move to the bits with heft.
To me, this takes a very good method of play - goal and approach - and dials it up to a point where it takes all sorts of equally-good other methods of play that could happily co-exist with G'n'A in the same game, and tosses them out.
Goal and approach require that the DM change the presentation of the game.
No it doesn't. The DM should present the game exactly as she did before*, giving no information that the PCs wouldn't have and remaining neutral.
* - assuming the presentation of the situation is already of worthwhile enough quality for the players/PCs to interact with; if not, this needs improvement regardless of playstyle.
What's required is a player-side change, where they take the elements of both their PC and the described situation and incorporate them into how they describe what they're doing next. Put another way, the players need to take a bit more time and use a few more words to describe both what they want to do and how they're going about doing it.
Situation: party's trying to break someone out of prison and know what cell he's in.
"We sneak down the hall to the prison cell" might become (after some in-character planning either now or earlier)
"The prison cell is at the end of the hall, and we have to assume there's guards about even if we haven't seen any yet. This means absolute stealth, so we spend a moment and tie bits of cloth around our boots to muffle the sound of footfalls, we run on the minimum possible light (or better yet, none at all), and to avoid getting separated we run a long piece of string between us that we each hold in our left hands"
You have to present challenges that prompt the players into action. This is different, as most games just have the DM present the description of the room and have other information gated behind the obligatory skill checks. You either gain the information or you do not, and this affects the actions your take and if the things you did not notice affect you and now call for new checks or if you engage what you have noticed via other checks. Goal and approach, though, doesn't work at all with this presentation -- you must provide a handle on the action for the players.
Sure it works with this presentation, provided you're not hung up on the idea of the PCs/players acting on faulty or incomplete information. If you don't know something, or miss noticing something, you can't very well act on it, can you?
And sometimes people do miss noticing things even when they're right in front of their faces.
As such, it requires a form of framing more akin to more narrative-style games where you present a dynamic situation with a clear call to action and then say, "what do you do?"
Yes, this method misses some of the things that the ask-for-rolls does -- they are completely different styles of play. What's missed, though, are the things that no longer make sense in terms of goal and approach play. I don't miss that my players ask for rolls, fail, and give me the opportunity to create new fiction to describe their failures because my method does this well, just in a different context. My method creates consequence based on what the players express rather then what I, as DM, think. I find this preferable. I have to do a bit more work on the front end -- I have to provide a clear call to action in my scene framing and this isn't trivial -- but I offload a lot of work on the backend as I'm now just reacting to the players and following their lead through the scene. This is very different from the much more DM mediated experience of asking-for-rolls and using rolls to gate information and provide tension. Both are very valid ways to play. Neither can recreate the experiences of the other. That's actually a big selling point for me -- most of my pain points with D&D came from the heavy DM load and I find goal and approach lightens that considerably and presents play that I enjoy very much. YMMV, and that's part of the coolness of this hobby.
Nothing says these things can't be combined.
All it needs is that your scene framing be based on the level of PC knowledge given by their rolls, rather than just assuming they notice everything. For example if the PCs enter a room in an old castle that has various furniture, a half-open closet, a curtain-covered window, and an assassin hiding behind the couch your framing and description would - I hope! - change slightly based on whether the PCs noticed the actual assassin, or just noticed signs of his presence, or didn't notice anything amiss at all. It's a safe bet the presence or absence of this information is going to have a lot to say in determining the PCs'/players' next goal(s)!
