I'll do my best - the expansion really means answering your other question:
I think skill challenges have two modes. (Maybe more, but I know of two.) One is essentially the "complex skill check" - to open this lock will take 4 successes, to decipher this cryptic text will take 4 successes, etc. That use of skill challenges is not (or at least need not be) metagame heavy. But it also generally won't bring out the claimed virtues of skill challenges, either - for example, if one of the PCs is weak at lock picking then s/he will just stand back and let the others go at the lock while s/he keeps watch, or shines her boots, or whatever.
For the skill challenge to achieve the virtue of engaging everyone, then the GM has to frame the situation, and adjudicate it, in a metagaming fashion - for example, s/he has to narrate consequences for checks (both successes and failures) in such a way that they suck the other PCs, and thereby the other players, into the action.
And once you are doing this, the other metagame dimension occurs - which is that you have to handle your adjudication and narration of consequences in such a way that, when the final roll is made, a conclusion results.
A simple example is in that in the Essentials Rules Compendium and DM's Kit. In that example, at a certain point in the challenge the PCs meet, argue with and irritate some NPCs. Then they go off and do other stuff. And then the 3rd failure occurs. And in response to that, the GM narrates that the irritated NPCs turn up again, ready for a fight. Now, the fact that those NPCs turn up is not an ingame consequence of the failed check. It's not as if, by making some mistake, the PCs have caused these NPCs to arrive at that particular moment (contrast: failing a pick locks roll does, in the fiction, cause the trap on the lock to activiate). The reason the NPCs turn up is because the GM has, at the metagame level, decided that that is a dramatic way to tie together the events of the skill challenge, and thus have its culmination be aesthetically satisfying (even if a failure rather than a success).
There are many inadequacies in the way that the various GM guides for 4e talk about these issues, but one of the most glaring ones is that no where in that example do the designers actually explain what the GM is doing, or point out that the consequences for failure have been adjudicated by the GM on the basis of metagame (aesthetic/dramatic) considerations rather than simply reasoning out the likely ingame causal consequences of failing a skill check. Contrast this with, for example, Burning Wheel, which has detailed discussion about the importance of both intent and task in action resolution, and emphasises adjudicating failure on the basis of intent rather than task (so if I fail my Athletics check trying to climb up the cliff to stop the fleeing bandit, it's not that I fall to my doom, but rather that I don't make it in time). I'm not saying that BW is necessarily a better system than a more task-based way of adjudicating failure - my point is that it gives clear guidance to the GM on how the designers intend it to be run. Whereas the 4e books leave it all up to guesswork, intuition, and inference from unanalysed examples.
A further, final point about metagame, which also answers the question - why bother with skill challenges that aren't just complex skill checks? My answer - and it's not one I would expect everyone to share - is that they create "space". Because the GM can't narrate the resolution of things until the requisite number of checks have been made, the GM is obliged to keep introducing new complications, to keep pouring on the pressure, to provoke the players into having their PCs do things - take new risks, offer new compromises, whatever it might be - and out of these moments unexpected but interesting things happen in the story. Running a skill challenge in this way is, on the GM's side, again all about metagaming - at each moment, the GM is doing the same metagaming thinking that in the WotC example leads to the GM resolving the skill challenge based on metagame considerations.
And if a GM or a play group is not interested in running skill challenges in something like this fashion, then I think the mechanic has got nothing to offer other than as a complex skill check mechanic (and for complex skill checks I would not normally go above complexity 2 ie 6/3). But nowhere do the WotC books talk about skill challenges in this way, or how to run them like this. I learned how to do it by reading the rulebooks for Maelstrom Storytelling and HeroWars/Quest - both systems that have comparable resolution mechanics, and were available and well known when 4e was written. I think the WotC designers should be trying to do a better job of explaining what their mechanics are for, and how they can be used to advantage.
