In any case, in looking purely at the math of +9-11 vs DC 20, you're missing a lot of the options and depth that go into a skill challenge. First you have the potential modifiers to the math. Then you have the fact that a good skill challenge should always provide interesting options; it should be more than just a few flat rolls. And finally there's the fact that failing a skill challenge shouldn't be the end of the world. In many cases I assume that the PCs WON'T succeed at a skill challenge (remember as DM, I KNOW what their skills are when I'm designing the challenge); the issue is that the closer they get, the better.
What I find interesting is how skill challenges seem to confound the assumptions many brought to 4e. The skill challenge system does not banish DM fiat from social conflict; rather it seems to return DM fiat to a system that was a simple, boring, one-roll skill check. Once again, your best chance at success comes from entertaining the DM:
In running a challenge, I'm not looking for the PC to say "I'm using Diplomacy." I want him to roleplay the scene. How's he making his case? Is he drawing on anything specifically relevant to his target? While I like this for color, it's called out as something that SHOULD be rewarded. In providing advice to the DM, page 74 of the DMG specifically notes that you can choose to reward creative action (or penalize the opposite) by applying a -2 to +2 modifier to the check. In some cases, I've specifically set up encounters where the player can get an even higher bonus if he brings up the right thing...
I honestly don't consider this a bad thing, though I wonder at the utility of a dice rolls. I suppose the system does provide the new DM with a rigorous skeleton to hang new skill challenges from. And D&D just loves getting you to roll dice.
In any event, this is one topic that's clearly stirring the natives to restlessness:
As far as I can tell, skill challenges are a game in almost the same way that snakes & ladders is a game. Player choice doesn't matter -- it's all dice rolls, no strategy. The inclusion of skills that count as automatic failures would seem to mitigate this somewhat, but tactical play requires knowledge choice between meaningful options, and the book seems to suggest you can only find out which skills autofail by using them and suffering an autofail.
For a game that prides itself on a lot of interesting tactical choice, this is bad design at the foundational level. Adjusting DCs won't make them work -- the system needs total replacement. There is nothing "X successes before Y failures" does that "one success on X skill" doesn't do, except obfuscate your chances of success.
UPDATE: Mike Mearls gives us the word from the Coast:
We had a meeting about skill challenges on (cue creepy music) Friday the 13th. We came to a few conclusions on what happened, what our intent is, and what we're going to do about it.
The system went through several permutations as we worked on it, and I think there are some disconnects between the final text, our intentions, and how playtesters and internal designers use skill challenges.
So, we've been listening and reading threads and figuring out some stuff on our end.