"Facts that need pointing out, although they are plain on inspection" is a not-too-inaccurate paraphrase of the definition "class NP problems" in computer science. You aren't describing a failure of rationality, but a very basic limitation of knowledge generally: it's harder to solve a problem (how can I toast marshmallows in my studio) than to verify that a proposed solution works (put them in the oven).

That's dangerous knowledge. It's kept secret for a reason. Now that you've let the cat out of the bag, campfires have lost some of their magical power.

Also, my family had a recipe for sweet potato casserole that included toasted marshmallows on top. But it was a special thing you could only cook Christmas and Thanksgiving.

Learned blankness about the possible use of everyday appliances, besides the explicitely stated?

Now you're morally obliged to find 3 creative use for a freezer...

This general problem is why I want these signs: they call out the "shouldn't need pointing out" things in public establishments, all of which are severe impediments if you don't know them.

Of course, there is the opposite problem to worry about, of adding so many bits of advice that no one bothers reading the list, and then discounts its value, leading to people not posting these ...

Though the point Sewing-Machine made about NP-completeness is completely apt, one should not take NP-completeness as a sign to stop thinking; it's too prevalent.

The original question implicitly gave an argument something like: Alicorn has (various enumerated) good reasons to believe that "ways to cook marshmallows" is within her expertise. A straightforward query yielded only one way, requiring an open fire. Therefore, she can conclude that it is difficult or infeasible to cook marshmallows without an open fire.

I vaguely recall a quote something like "If an expert believes a thing is possible, they are very likely correct. However, if an expert believes a thing is impossible, they are very likely mistaken."

Possibly we can take this as a recipe for judging / weighing specific kinds of arguments. I think Douglas Walton calls this kind of argument "argument from position to know", and we could add a checklist for judging argument of this sort, something like: Is the claim universal over a combinatorially large domain? Does the expert have a reason, other than their expertise, for their belief? If not, discount the expert's belief as the expert is likely to be overconfident.

I'd be willing to bet that if you had at some point found yourself with an active (and at least moderately strong) desire to have a toasted marshmallow, you would have sought and found a way (oven, toaster, etc) to toast one in the kitchen... mostly because once upon a time, I found myself with a bag of marshmallows and some chocolate, and wanted s'mores, and decided since "toasting" to me at that time mostly meant "torching," a candle would suffice. And it did.

Toasting a marshmallow without a campfire wasn't a difficult problem; it was just one you didn't consciously try to solve. Maybe this marshmallow incident you've related is as simple as a recommendation for us to more actively identify little questions like that in our daily lives. If those questions can be converted to some form of desire ("I want X," or "I wish that X..."), it seems that we'd be more likely to see the simple-but-not-obvious solutions.

Over the course of working in a research lab at first part-time in undergrad, and now as a full-time grad student, I've run into a lot of little mental connections like that that tend to make me want to slap myself when I hear them, so I've started to make it a game to notice when something's unreasonably difficult, finicky, or irritating and, try to change something about it rather than just grumbling quietly to myself.

(One good example: re-papering bench space where ethidium bromide is used. Gloves are mandatory when handling EtBr-contaminated material. Labeling tape sticks to latex gloves like CRAZY, making it difficult to tape the bench paper down. It's almost embarrassing how long it took me to think of just wetting the fingertips of my gloves a bit to keep the tape from sticking to them.)

I think I see where you're going with this, but to be more analytical than you intended, the answer to one of your questions is hypostatic abstraction. It's an immediate (and therefore deductive) logical inference that, in this case, goes like this:

1. You can toast marshmallows with an open flame.
2. There is some means/power/virtue/relationship by which you can toasts marshmallows with an open flame.

This relationship inferred is called the hypostatic object. It's a deductive inference, so if it seems like this hasn't actually added any information, you're right. But it is often a useful inference in that it brings out the hypostatic object which is latent in proposition (1) above, but isn't itself (yet) an object of thought. In this case, obviously, the hypostatic object is to be identified with heat. A lot of our scientific concepts were once nothing more than hypostatic objects: think of electrons (the whatever that produces electricity) and photons (the whatever that produces light); eventually more is learned about these objects, so that saying something like "electrons are responsible for electricity" is no longer tautological.

The point being, when you were associating roasting marshmallows with an open flame, you were thinking too concrete. So how do we stop and think about these daily, ordinary things, at a more abstract level? I don't know. There are probably all sorts of things like that which I'm hardly aware of. Maybe I should just go through a brainstorming sessions about ordinary things and try to think at a more abstract level.

bacon can be cooked in an oven. I never knew this until recently, I had always fried it. much less messy/time consuming.