Other than the obvious problem of people using that information to try and "game" the application process, if I was accepted I would prefer to have YCombinator's full attention trained on helping the startups out and less on providing feedback to those who didn't make it.

EDIT:

I realize that sounds greedy. What I meant was that I didn't get accepted and I'd rather see YC spend their energy helping those who did get accepted.

I was impressed to learn that Paul personally speaks to each group that makes it to the interview, but is not accepted into the program. He does just what you're asking: explains why the group wasn't chosen and what they can do to improve the odds in the future or with other investors.

And, as another response to your post mentioned, there is no benefit to YC to enabling groups to game the application process. Everything, everything, everything you need to know about the process has been documented by Paul and many of the groups who've been through it (both accepted and declined).

But I'll sum up:

Build something people want. You need to prove you're building something people want. A "proxy for demand" is necessary if you're inventing something new, or evidence that users want what you're building rather than your competition if it already exists.

Be capable of building what you say you want to build. Link to a website or product that's in progress, or at least a website or product that you've built in the past. I get the impression from my brief exposure to the process that one of the more frustrating things for YC is seeing a smart group with a good idea, but without the obvious ability to execute. This is why they've discontinued requiring an "idea" on the application. Just prove that you and your team can build great things. If you don't have that evidence...why would they choose you when there are dozens of groups that do have evidence of building great things. This isn't school where each project stands alone. Show them your track record. There's space for it on the application.

That's pretty much it for the early process. Then you've gotta pull off a demo for the interview day, and show extreme passion for building a business. Very few no-demo groups make it into the program, and I believe the percentage is shrinking (and I suspect the ones that do get in without a demo have a long and fancy track record--I bet Zenter would have been in no matter what because they have a kick ass track record of building things people want, but they also had a killer demo that they'd built in the weeks prior to the interview). It can be partly mocked up, or carefully orchestrated to avoid the bugs, but they'll see through wholly fictional demos so don't try to fool them--these aren't your typical non-technical investors.

This comment -- http://news.ycombinator.com/comments?id=14882 -- in response to this article -- http://redeye.firstround.com/2007/04/american_idol.html -- might give you an idea why they won't provide more specific feedback.

But I agree with you: a more specific feedback would be better.