The big problem with the skill challenge mechanic is that it is very "swishy" (sensitive to small changes in skill level). But any multi-roll system is going to have that problem to some extent. Any small percentage advantage or penalty in a single roll will grow in significance as you make more rolls.
I happen to think the original core mechanic has a pretty good chance of meeting the requirements. By this core mechanic, I mean "the whole party makes skill checks from a limited set of skills trying to accumulate a certain number of successes before a certain number of failures."
I think any other multi-roll skill system is going to have similar problems with swishy-ness. I also think that the original (non-Errata) challenge system has some features to help swishy-ness that folks are not considering. In particular, by forcing every party member to participate, the skill level per roll is averaged over entire the party, which mitigates the presence of both Skilly McAwesome characters and Lamo McLamerson characters.
I think what the core system lacks is (a) solid base numbers and (b) additional tactical options to make the system interesting enough for long term play. There is nothing preventing us (or WotC) from adding these options to the existing challenge system.
In fact, many of the sample challenges and suggestions in the DMG do include such options, just not in a rigorous way. Examples include one-shot Easy skills, Information Skills that provide data on how best to beat a challenge, failures that cost your resources like healing surges or gold, etc.
My current thinking is to make those and other options for skill challenges more rigorous, assigning complexity modifiers to them when they are added to individual challenges. Each challenge would use the base challenge rules (which should be simple) plus a few challenge options to twist things up and keep it interesting.
This is why my suggestion (above, somewhere) was to allow everyone in the party to contribute, even Lamo McLamerson, but have their skill checks contribute modifiers to the "main skill roll."
So in a 5 member team, one of the members (likely with the highest modifier for the skill in question), will be the "point man" and make the main skill roll.
The other party members make skill rolls, if they want, to try to grant modifiers to the main skill roll (the result of the assisting skill rolls might be bonuses between -1 and +4 for example per PC, with a lot of +0's and +1's in there, with a -1 for skill fumbling)
This means that the lame skill PC's get a chance to do something, and if they fail it's not a "failure" it's just not giving a bonus to the main skill roll. If a lame skill PC rolls well, is clever, has other useful skills they use in good ways, they could contribute up to +4 or something to the main skill roll.
This means:
1) All the party is involved, even lame skill PC's. Giving a +1 or +2 to the main skill roll is a big deal.
2) Party cleverness is rewarded (if PC's use their lame skills in clever ways to help generate modifiers to the main skill).
3) DC's are much easier to determine and fix, because they will be based on the expected "high skill" bonus for the party.
So if I know a knowledge check will be required, I know the high/maximum range for this check throughout all levels. So I could set a DC allowing for 50% success if the main skill check gets no help. If all the party really is on fire, they might contribute +8 to this check, helping a lot. If the rest of the party isn't able to help, then they might fail. Or set it to 75%. Or whatever. It's a lot less swingy because you can control the maximum amount of "helping" skill modifiers - and assume that the party will always use the highest skill PC for the main roll. Which is what they do currently.