it needs at least some active opposition and some player-resource-management.
There
is resource management, namely, of powers and rituals that affect or substitute for skill checks:
From the PHB p 259:
In a skill challenge, your goal is to accumulate a certain number of successful skill checks before rolling too many failures. Powers you use might give you bonuses on your checks, make some checks unnecessary, or otherwise help you through the challenge. . .
You can use a wide variety of skills, from Acrobatics and Athletics to Nature and Stealth. You might also use combat powers and ability checks.
I think it's pretty unambiguous that resources can be brought to bear. Some of the examples I linked to above have illustrations of this. The player of the sorcerer in my game, in particular, uses his attack powers for all sorts of funky effects in skill chalenges.
As for active opposition, that is all in GM narration - as is the case in any "player's roll all the dice" system. If a GM is running a skill challenge, but is not pouring on pressure so that the players can see why their PCs need to engage the situation, then no wonder it is boring! That's like a "combat" encounter in which the NPCs all stand around and don't draw their weapons, but just let the PCs cut them down.
What does a skill challenge do that you can't do without it? Assume the party needs to sneak into a town undetected. What does the skill challenge add to the game that a simpler and more flexible "What do you do?", followed by adjucating actions and reactions, doesn't?
I've indicated this above. It creates a minimum and a maximum narrative heft at the table. It introduces complications, because the players have to have their PCs do stuff, and the GM has to resolve checks and evolve the situation to create the framework for new stuff.
[MENTION=386]LostSoul[/MENTION], a year or so ago (?), gave a nice example of a social skill challenge: the PC coulndn't lose (succeeded on a 1), but had to make 4 checks. And in the course of making those checks, offered a compromise to the NPC (why? because, as part of the resolution of an earlier check, LostSoul had the NPC ask for a compromise, and the player, in making the next check, agreed to it). A single-check system, or a system that focuses only success or failure without focusing on the fiction that generates it, won't give you those sorts of outcomes.
I linked to some examples from my own game above, and have described some others in this post, which also illustrate use of the skill challenge framework to generate complications and outcomes that wouldn't arise otherwise, out of a simple focus on task resolution without any structure to (i) mandate introduction of complications by the GM, or (ii) entitle the GM to call for checks by the players without being vulnerable to accusations of check-mongering or "mother may I?" (because it is within an overall framework).
Absolutely nothing. A skill challenge does not do anything that you couldn't do without.
What it does do though, is provide a nice neat framework from which you can work instead of trying to play amateur game designer all day long. Is it a panacea that cures all ills? Nope. Is it the best way to do things? Certainly not.
But, having that framework makes the job of the DM SOOO much easier.
As my reply to Fenes indicates I don't entirely agree with this. I agree that skill challenges aren't a panacea, and they aren't necessarily the best form of extended contest. But I do think that the history of RPG design since the mid-90s shows that one of the best ways to get dramatic, non-combat game play which is player-driven and centred on the fiction is to have some form of structured extended-contest mechanic.
Without the structure, the GM has no framework to regulate his/her pouring of pressure onto the players - it just becomes arbitrary adversariaism - and the players have no rationale for continuing to make checks - and so you won't get the nuance of compromises and other tweaks in the fiction that add player-driven subtle content to the fiction.
(Imagine if, instead of generating a compromise the way LostSoul did, you said on a success by a margin of 5 or less, you must compromise - how would you work out what the compromise is? You'd have to fiat it. Whereas, the back-and-forth structure of the skill challenge produces a compromis as a natural outcome of the resolution, that reflects the choices the players made in engaging the fiction.)
I do believe that SC is a bland mechanic that rely heavily on DM fiat to succeed
For the reasons that I've given I don't agree with "DM fiat" - but yes, skill challenges are heavily reliant on the GM's adjudication and narration to make them work. That's a general feature of an extended contest mechanic - otherwise it just turns into meaningless dice rolls with no connection to the fiction. Unfortunately the 4e books (DMG, DMG2, Essentials) don't give any advice on how to do this. Luckily for those who are familiar with them, though, there are many other excellent GM resources available which do give extensive advice on this (eg Burning Wheel's Adventure Burner, HeroQuest revised edition, The Dying Earth RPG, etc).
what does a frame help me, unless I plan to press and railroad the players into it?
<snip>
Skill challenges are far too rigid, and do not add anything but traps that turn the game into a dice rolling exercise.
I think this is a bit like asking, how do combat rules help me, unless I plan to railroad players into combats?
If you tell the players that their PCs are attacked by assassins, they may want to fight back. You would then use the combat rules, which set a framework and parameters for adjudication. Some of that you've determined at the start of the combat (eg the assassins have AC, hp etc). Some of that you determine in the course of resolution (eg which assassin moves where to flank with which other assassin).
Suppose instead the players decide that their PCs run away. That's the sort of thing you can resolve as a skill challenge. Some of it you would settle at the start of the challenge (eg how many successes are required). Some of the challenge you determine in the course of resolution (eg a player uses Arcane Gate to give the PCs a 20 square headstart, and you then narrate some salient complication or narrative twist or turn in response to that).
As to being nothing but a trap that turns the game into a dice-rolling exercise, have you read the actual play examples I've posted and linked to? Have you played games like Maelstrom Storytelling or HeroWars/Quest that rely heavily on extended contest mechanics?
It's one thing not to like a mechanic - nothing wrong with that! I don't particularly care for many of the mechanical features of classic D&D. But in this case I think you are seriously misdescribing it. Not everyone likes HeroQuest as a system, but it's rarely accused of being nothing but a dice-rolling exercise.
in many extended skill tests, without a framework, players will fail more often than succeed. Which, in turn, leads players to adopt a "Kill everything" approach because it has a higher chance of success than trying to beat the rigged game that the DM is unwittingly presenting.
Then the DM comes on En World and bitches about how his players always try to kill everything and they won't role play. And a bunch of other En World DM's pat him on the head, tell him that his players suck, and he should keep doing what he's doing.
Ouch! (But true.)