LostSoul
Adventurer
Isn't having degrees of success available in a challenge one of those holy grails of dungeon mastery?
Yeah, it's a good thing. I'm not talking about degrees of success, though, I'm talking about the DM not taking into account what the players do.
In any case, the problem is that you want the players to be able to outsmart encounters that haven't yet been written, and you want them to do so by happily metagaming the situation and trying to intuit DM motives, but you want the Dungeon Master NOT to do any of those things. You have D&D backwards.
What I think is good play is when the DM takes into account the goals of the players and gives them opportunities to make meaningful choices. Future encounters are affected by earlier ones. The way in which the future encounters are affected depends on what's important to the players and the choices they made.
For the lieutenant example:
We have a group of players who likes to overcome challenges through smart play.
They decide to kill the lieutenant because they think he'll make future encounters more difficult. This is their whole reason for killing him.
In future encounters, even if they are not written up yet, the DM should make allowances for the absence of the lieutenant - and in a meaningful way*, to reflect the success of the PCs earlier on, to show them that their choices have an impact in how the game unfolds.
(Since the players want to overcome challenges through smart play, meaningful - for them, in this example - probably means that the encounter is going to be easier.)
If the player's choices did not affect the future in any meaningful way to them - maybe the lieutenant isn't there, but they could care less; all they care about is the difficulty of the encounter and not the specific NPC - I don't think the game will be as rewarding. What I think is bad is when you pretend that the future encounter was changed in the way they wanted it to, when you pretend that their choices had the effect they wanted.
I say be upfront about it.
Let's change the example slightly and give the players different motivations. They want to kill the lieutenant because, even though he's a spy on their side, he's been a real dick about things to the PCs. He's hurt them personally and so they make the choice to kill him, even if it means letting the town burn because they don't have access to his intel any more.
I don't think these players would care if the future encounter is just as tough as had been planned. But what would bother them?
You make up an NPC who is the real spy on the spot so that the intel gets to the town and the PCs and they can make use of it. (Illusionism - when you tell them the lieutenant was never the real spy, that you planned this out the whole way.)
In both cases, the player's choices have no meaning; in the first case, the encounter was no harder or easier given the death of the lieutenant, and in the second, the moral weight of the choice is taken away because the get the intel no matter what.