I haven't had a chance to do a thorough math analysis on it yet, but I can give you some highlights.
First of all, WOTC definitely curbed many of the problems:
1) Higher Complexities are always harder.
2) Parties have good win rates.
3) Each person has a good chance to succeed at any individual skill check.
4) Adding off the wall skills are now medium Dc in general , not hard.
5) Aid Another should now be used infrequently.
There are now three issues that still remain, one is old, and the other is a creation of the new system.
The old one is variance. If we look at each player having a 90% chance to succeed with their best skills (and with the new DCs that is highly likely). You get about a 98% win rate at complexity 1 and a 84% win rate at complexity 5. That's actually really good for such a wide complexity range!
However, let's drop each person's check to 85% (basically a +1 to the DC). The complexity 1 win is now 95.27%, meaning the win rate hasn't dropped that much....that's also very good! However, when we look at complexity 5, the rate is now 64.79%. that's a difference of 20% from the original complexity 5, and a 30% difference between the complexities at that DC. That's a big difference, and it gets worse as you drop further down.
This is a fundamental property of the success/failure model, you cannot get rid of it. Much of my work on my original system was creating mechanics to curb it, but that's all you can do. WOTC has removed some of the variance of its original system, but much of it remains.
The next problem is that the ceiling has been made too low. At 1st level, a player will generally have +9's to his good skills, and could easily have +10-12 if he specializes a bit more. Those are automatic successes on medium DCs. And of course, when you throw in utility powers you can easily get players autowinning skill challenges. I think it was good they lowered the DCs, but I think they went too far. HOWEVER!! If you take away the standard assumption that players are supposed to use their best skills, this changes. If similar to Obsidian, a DM says that these skills are the right ones for the job then the low DCs might actually be a blessing for characters who aren't as trained in those skills.
The third problem is that they still have disincentives for rolling. In fact, now they've taken out the assumption that all players participate in a challenge...which to me is the worst problem. To me, that was the whole point of the skill challenge system, to get players to accomplish a skill encounter as a team. Now the best way to go is your best skill guy just rolls and everyone stands back.
Now...all of that aside, I will say the math is a lot more solid than the original version, and because of that, this system is worthy of playtesting. I have said many times that I don't think math should be the final judge of any system. The reason the original system broke that rule was because its math was so blatantly bad. If this system had been the one I was originally handed, you would not have gotten all of the crazy math analysis I have done. I would have noted some math oddities, but I would have taken faith and tried it out a lot before making any judgements. But now I have Obsidian, my group likes it, and I doubt we will change from that any time soon.
