If you have heavy obscurement, you can hide, otherwise, without an exception, you can't.
The question is not "When can you hide?" but "Are the requirements for hiding (transitioning from un-hidden to hidden) different from the requirements for staying hidden?" Since I think the answer is, "No, they are not different," I'm not going to try and argue the contrary position. But there are people who do argue the contrary, and so far as I can tell they're quite sincere about it based on their reading of the rules. Same thing with the wild elf question, which you didn't address.
What Vic Ferrari says seems to me obviously wrong, because there is the notion of "distraction", which permits a creature to remain unperceived, and therefore to remain hidden, even if not obscured.
There is also the passage in the Exploration rules that says you can remain stealthy as long as you don't come out into the open, which might be read as implying that any cover or obscurement is sufficient to remain hidden (as per the 4e stealth rules).
For me, the most natural reading of the rules is along the lines Dausuul has sketched, but I aslo agree with him 100% that those rules are poorly written and presented. (For instance, I have had posters argue that a wood elf rogue hiding behind a wall, in fog, becomes
harder to detect if the wall is disintegrated because now the fog imposes Disadvantage on the relevant Perception check. It seems to me that nothing but poorly-written rules could lead to that conclusion, which surely has no grounding at all in the fiction of the situation.)
Mike Mealrs: "Our rules for stealth, which may sound like a funny example. But having worked on 3rd and 4th edition, creating a set of rules for hiding from other people and monsters that run without a DM, is crazy. You always end up with a situation where you’re standing right in front of the monster but he can’t see you, because there’s a loophole in the rules."
"So we just came out and said you know what, let the DM decide.
These are separate issues and you keep conflating them. Ambiguity is often intentional, and I gave you an example from Mearls where he explains the stealth rules are intentionally written that way.
<snip>
But take for instance the Grappler feat in 5e. The third part of it says, "Creatures that are one size larger than you don’t automatically succeed on checks to escape your grapple." But, that's referencing an old rule from the playtest which doesn't exist in the game at all anymore. It's an error (or, as you put it, a poorly written rule). It's not ambiguous, it's just erroneous.
<snip>
So you're saying it was not intended to be left to interpretation, like the stealth rule. OK. So who is saying "I can interpret it therefore it's not broken"? In my experience, that response is reserved for rules where it's intended to be up to the DM, like the stealth rule, which you complained about earlier.
How am I meant to know which rules are deliberately ambiguous, which are errors (like Grappler) and which are just poorly written (like the Magic Missile example)?
The Stealth rules certainy don't come out and say "You know what, let the DM decide". In fact there are mutiple column inches of rules text, which turns on technical mechanical notions like heavy vs light obscurment, disadvantage to checks, etc.
For me, the clear contrast is with the Hermit background's insight feature, which I regard as both a good rules and a clearly-written rule. It
does come out and say that a decision abou the meaning and implications of the feature have to be worked out between player and GM.
The Stealth rules could have been written that way, but weren't.