The first rule of fudging to make things more exciting is you do not let the players know when you're doing it. Even if they know, in a general sense, that you do it, you never ever want them to realize that it's happening in the moment. It's the difference between knowing that magicians use illusion and misdirection, and seeing the strings on stage. Instead of creating excitement, you kill it dead.
Intervening to deny a long rest, when the party has every reason to believe they can rest safely, fails that requirement hard.
To me, this reads as if you've misunderstood what
@Manbearcat is envisaging. He's not thinking of a GM saying, flat out,
Sorry, no rest allowed. He's imaging the players setting up their rest and then the GM using his/her control over the fiction to obviate that (eg they get woken up; a scout from the BBEG turns up and dispels their Rope Trick; or even, as has come out in some of the more recent posts, the GM lets the rest take place but adjusts the planned follow-up encounter to reflect the fact of the rest).
So, is having Strahd prepare while the PCs are resting negating the long rest? If they've done intelligence-gathering, is his ability to prepare previously established?
<snip>
Changing the facts out from under the players' feet is kinda dirty pool. Giving Strahd resources the players would have noticed (but didn't) seems like that.
<snip>
If you never undid any of their achievements, you were honoring skilled play, as well as the story.
one effect of taking a long rest in such a campaign would be "the bad guy also gets an extra 8 hours to prepare." It's a cost/risk of the long rest. This gives the dm a lot of room to adjust the final encounter to dial up the difficulty.
I think that notion of
giving the GM a lot of room is pretty crucial. What use, by the GM, of that sort of room is fair? Conversely, what sort of use of it counts as "changing the facts out from under the players' feet" or "undoing players' achievements"?
I think 13th Age (p 171 of the rulebook) has a fairly elegant answer to this: if the players have done 4 encounters before their recharge, then the GM is not to do very much changing; whereas if the players have done 3 or fewer encounters, they have to suck up a "campaign loss" ie the GM can change quite a bit, and quite adversely to the players and their PCs!
My impression of 5e is that it deliberately eschews any rule of this sort, thus opening the way for exactly the question that
@Manbearcat asks in the OP.
My recounting of this scenario (prompted by the mention of Strahd) was largely about the fact that in my mind, no matter what they did, this was going to be a difficult encounter. There was a dial, and I was willing and able to adjust it, and do so invisibly.
For example, they had the Sunsword. So Strahd had minions on hand not susceptible to the Sunsword's power, and they swarmed the fighter who wielded it, and tried to take it away from him. I did this to make the encounter more difficult. However, would I have added those minions if the players had not managed to find the Sunsword? Probably not.
Is that kind of encounter design consideration contradictory to anything suggested by the books?
<snip>
What rules would be broken by "doing what sees like it will be most exciting"?
I like your posts on this. As well as the actual play reportage, they also point out that the GM has multiple dimensions of decision-making about the fiction available to him/her - not just
what happens while the PCs rest but (eg, and as in your example)
who do the NPCs have as allies/minions?
If you don't mind me getting inspired by the direction of your thoughts, I'd like to tweak it a little to add a "third solution" that perhaps we should be considering. How does this play out if the DM simply says to the players, as one hobbiest to another, "I think this will be more fun if you push on".
I think I'd take that advice. Though I would be salty if we then died.
if I have a cool encounter that wouldn't work if the players have full resources, then I'll probably just ask them to not take a long rest.
When I was GMing a 1st to 30th 4e D&D campaign, I adopted a variant of this approach: if the players were thinking of taking an extended rest before their resources were really spent, I would make fun of them. Normally the shame would make them carry on! I would also use my control over the fiction to soft-force them to go on.
Because of the way 4e distributes resources across short and extended rest, the risk of the saltiness Blue mentions is less than in some other versions of D&D. For the same reason (ie much less player-side nova capacity), it's also easier (I think) to ensure that a 4e encounter is not anti-climactic by way of ad hoc adjustment to reflect the fact of an extended rest.
All of the above does mean that extended resting ceased to be part of "skilled play" in my approach to 4e, and became more about how the players and GM - each using their various powers over the fiction - would choose to shape the player resource pool for what comes next. The skilled play happened primarily within the framework of encounters. And this also helps make the "cooperative"/"consensus" approach to extended rests work - because the players feel so pleased with themselves when their in-encounter skill allows them to defeat an above level challenge while sitting on a small number of dailies and one healing surge per party member, they don't object to me as GM controlling the fiction so as to generate those sorts of situations!
My impression is that 5e is less often approached in this sort of "scene-framing" style, which again makes
@Manbearcat's question come to the fore.