A more broad reading of "complex plans and operations" would see them as a way of moving from point A of and adventure to some other point (be it B, Z or something inbetween). In combination with the general "yes, but" tenor of the advice in the 4e DMG, the real questin is "will the complex plan or operation take you from A to B as you desire, or will it instead land you in B with a penalty, or instead in C which isn't quite where yu wanted to be"? When you look at it this way, that's just what a skill challenge is for by your own account of it.
That's correct, though I believe PCs should always end up at B. We are telling them "Yes, you get to B....now let's just see how difficult it was to get to B".
My issue is that given two situations:
A) I've decided to use a skill challenge, the PCs must get 10 successes before 3 failures.
B) I'm just going to ask for skill checks where appropriate until the narrative arrives at the right place.
B creates an identical situation to A except the pacing amounts to "whatever sounds and feels good in the narrative" whereas A says "I've set a mechanical number that must be achieved regardless of the narrative".
The only real purpose of A is so that you can accurately measure HOW MUCH they succeeded. It's the same thing as a combat. You could roll one dice to determine whether the PCs win a combat. You could even roll one dice per enemy. Instead you roll a series of dice with a carefully chosen difficulty to see how well they defeat the monsters.
An audience with the duke, a mysterious set of sigils in a hidden chamber, finding your way through the Forest of Neverlight — all of these present challenges that test both the characters and the people who play them. . .
I think this is the real key here. Skill Challenges are precisely this. A way to test both the player AND the character. And it is a structure to determine difficulty so that you can give out XP for non-combat portions of the game.
Without skill challenges, many situations would be resolved in previous editions purely by description. You need to find your way through the Forest of Neverlight? Alright, what actions are you taking to make sure you find your way though? In 1e and 2e, this would often be resolved either entirely by DM fiat given the player's description or randomly determined by dice to see if you keep walking the right direction. This doesn't test the character at all. It only tests the player...or it might test neither of them if it is random. Also, it's impossible to know the difficulty of finding your way through the forest.
Other times, you might test the skill of the character but not the player. In 3.5e, situations like this would often be resolved by one or a couple of die rolls: "Alright, give me a Survival check once a day to stay on course". This doesn't test the player at all since all you are doing is rolling the character's skill checks. Also, since you might get lost for 20 days in a row before finding your way there, it's once again impossible to know the difficulty of finding your way through the forest. Especially if you set the DCs of the Survival checks based on modifiers in a chart or guessing.
Skill challenges were created out of a desire to mix the two: You are making your way though a forest, you are lost...what do you do? The player then has to decide what skills to use based on the situation. The number of successes and the DCs of those checks are set by the rules so they have a fairly accurately predictable difficulty. This allows you to say "Succeeding on rolling high enough this many times in a row is X difficulty...so it is worth Y XP" in the same way that using 5 monsters of level X is worth Y XP. The difficulty of the checks required to get out can be modified slightly by both the player(Choose one of the skills with an easy DC and have an easier time) and by the PC(choose a skill that your PC is good at and have an easier time).
This isn't just about "avoiding penalties": it's about "success or failure" of a complex ingame situation with "changing conditions" (ie requiring more than just "one roll to resolve") being "uncertain" and having "serious consequence". To me, that seems to cover (indeed, to describe) complex plans and operations.
The primer on Skill Challenges that was sent to me by Chris Tulach back in the day as well as the panel we got with Mike Mearls both kind of said the same thing about WOTC's philosophy on skill challenges: The "changing conditions" should come from the DM in order to make there a narrative reason for continually changing die rolls. The first ever publicly released skill challenge(the intro adventure I ran at D&D XP before 4e came out) was basically:
"You are attempting to flee the city without getting caught by guards. You are running down the street but there's a cart in your way. The guards are gaining on you, what do you do?"
"I use an athletics check to jump over the cart. I succeed."
"I use athletics to toss the cart down when I get to the other side. I succeed."
"Alright, you manage to gain on the guards, but you find yourself at a dead end with a fence blocking the way. What do you do?"
"I use acrobatics to climb the fence. I succeed."
"I use stealth to hide. I succeed."
"The guards pass you by and you leave the dead end. The guards are about to turn around. What do you do?"
"I use Diplomacy to start talking to some nearby people so I'll blend in. I succeed."
"Alright, that's 5 successes. The guards pass you by and you are able to make your way out of the city."
Failure meant losing a healing surge as the guards caught up to you and got into a fight with you that caused you to lose a healing surge before you beat them and still managed to escape.
That was the other key to skill challenges. They must not prevent the story from continuing. If the story requires the PCs to escape(which this one did), the PCs escape. They just fail to get away unharmed. Plus, they lose the XP since they didn't successfully defeat the challenge of rolling high enough.
I think you different approach from mine (and [MENTION=87792]Neonchameleon[/MENTION]'s, I'm guessing) might be because you're running pre-written scenarios (eg LFR) so that the desired outcome of events has been pre-written in: the only uncertainty is the mechanical penalty the players will take on the way through (eg how many surges do they lose). This seems implicit in your reference to minor story penalties.
From what I can tell most of the people at WOTC run prewritten adventures, even in their home games. They write out the adventure for the next couple of sessions in advance and write out the primary and secondary skills for a skill challenge as well as the consequences of failure long before the PCs start the encounter.
The DMG definitely appears to be written from this point of view.
Loss of surges isn't the only penalty I've seen bandied around. I've also seen "since you are spotted as you are leaving, the enemies are able to better prepare for your attack in the next encounter, add 3 more enemies to the battle with no more XP." or "The PCs don't make it though the forest quick enough, so they will have a -2 to all skill checks to stop the doomsday device at the end of the adventure since the process of charging up is further along."
But the key benefit of success appears to be XP in most cases. Whenever you don't want to hand out XP, you don't use skill challenges. Especially if you want to keep up a certain pace of advancement. Part of the reason for skill challenges to exist at all appear to be the idea that if you need to fight X combats in order to get the PCs to the next level but don't want to have that many combats, you can instead replace combats with skill challenges to keep the same pace of advancement.
LFR definitely has a different style of doing the game, certainly.
Though I suspect that the lion's share of actual D&D play is somewhere between the LFR and pemerton sides of the spectrum.
I still believe that LFR was doing it entirely as was designed by the authors. They made sure to send us(the admins of LFR) a bunch of information about skill challenges and advice on how to properly design and run them in advance of LFR launching to make sure that LFR was an accurate portrayal of the way 4e was supposed to run.
They spent a lot of time telling us how we were supposed to be the ambassadors of D&D for the world. People would come to games days having never played before and likely would come face to face with an adventure we wrote. That's why they wanted to make sure we understood how to do it "properly". Chris Tulach spent a lot of time making sure we had the resources required to make LFR a success since he took the RPGA's job as the "advertising" wing of WOTC rather seriously.
Though, I still found that Skill Challenges were kind of a waste of time since they tended to add more confusion and awkwardness to games most of the time. Both when the DM didn't know how to run them properly or couldn't come up with a proper improvisation for what the PCs were attempting or when the players didn't properly understand them and stared at each other having absolutely no idea what to do.