It's tough on the DM because how do they know if a particular challenge is "very difficult" or "hard?"
It's tough on both the player and the DM because the odds of them seeing the difficulties the same are almost nil. Then it's tough because the player defaults to the DM's judgment (which the game encourages, as it should), and the DM makes the call which likely isn't what the player had in mind, likely causing a reconsideration of the action.
A lot of this has to do with Hussar's point about people not doing the math properly in a snap judgment; I can't tell you how many times I've seen this manifest since I first read him warning about it in games. Caught myself doing it wrong, other GMs, people in non-RPG scenarios...it goes on. Stunts often seem not worth it. In a game like 13th Age which uses the "fail forward" mechanic, it's less of an issue, but that's not D&D.
What 5e could have done is provide examples of skill checks that yield advantages commensurate with how difficult the roll required was. Assuming the person providing the example doesn't fall prey to the bad mathematical thinking Hussar outlines, it would have been really helpful and given players an idea of what to expect.
Another issue that makes judgments difficult is the existence of spells that enhance normal human activity. So if Jump allows X, and Jump is magical, then surely a non-magical human cannot achieve that result. Magic ends up defining what the non-magical can do. It's understandable why DMs would resort to this; for ad hoc situations, find references elsewhere in the game, etc. It can still steal some fun, though.
It's tough on both the player and the DM because the odds of them seeing the difficulties the same are almost nil. Then it's tough because the player defaults to the DM's judgment (which the game encourages, as it should), and the DM makes the call which likely isn't what the player had in mind, likely causing a reconsideration of the action.
A lot of this has to do with Hussar's point about people not doing the math properly in a snap judgment; I can't tell you how many times I've seen this manifest since I first read him warning about it in games. Caught myself doing it wrong, other GMs, people in non-RPG scenarios...it goes on. Stunts often seem not worth it. In a game like 13th Age which uses the "fail forward" mechanic, it's less of an issue, but that's not D&D.
What 5e could have done is provide examples of skill checks that yield advantages commensurate with how difficult the roll required was. Assuming the person providing the example doesn't fall prey to the bad mathematical thinking Hussar outlines, it would have been really helpful and given players an idea of what to expect.
Another issue that makes judgments difficult is the existence of spells that enhance normal human activity. So if Jump allows X, and Jump is magical, then surely a non-magical human cannot achieve that result. Magic ends up defining what the non-magical can do. It's understandable why DMs would resort to this; for ad hoc situations, find references elsewhere in the game, etc. It can still steal some fun, though.