Pitch everybody you know on your idea, and have the smartest people you know poke as many holes as possible, and address them. Then do it over again.
Had I walked in totally chill and hang loose I feel I would have at least done better. The more you can convince yourself that you really dont care the more relaxed you will likely be. At least that is my $.02.
In retrospect, I'm so thankful YC gave us an opportunity to interview though. The lessons learned on pitching to an investor were so valuable, as was the advice. Even not being funded by YC it is a win-win in one way or another.
Of course I agree with you, but it should hopefully go without saying.
Perhaps YC should start with a 1 minute elevator pitch. The point isn't to show off the project, but to get interviewees to understand you won't have more than a minute of the stage and things will be dominated by questions. If I understand it, the 1 minute should be a good test of how you can present at demo day with none of their coaching.
This really upset me at the time... We spent a lot of time answering the question "We're going to compare you to yahoo.com" and were rejected because they didn't see how we solved the essential design problem of our product.
The solid "yes" or "no" is a major feature of the YC process though.
Put it this way -- are you a magician? If you present a demo that effectively says "pick a card, any card," you'd better be. If you're demoing Google (the site it is today) for YC, it's probably not the best sell to just let someone type anything they want into it.
When you only have ten minutes, you need to retain as much control as possible over your message. Present the article to sell the magazine.
Of course the people with the money make the rules; it's not a question of power, but every process can be improved.
A 10-minute interview tests presentation skills more than anything related to the idea, technology, or founders' other abilities. This skews in favor of quick impressions and existing YC bias (towards ideas they already were waiting for someone to present).
I wonder if the median application received even that much (10 minutes' worth) consideration. It seems like a continuous process would work better, with staggered starts; looking at 20 applications every week rather than 500 in one week would allow a more thorough review.
Then instead of going to the same 10 weekly dinners, it would be a sequence of 10; e.g. weeks 31-40. The only real trick would be Demo Day; but that could be rescheduled for every 3 months instead of every 6 months (or monthly, and investors who are vetted get a card that gets punched every time and on the 11th visit they get kicked out if they haven't invested in anyone yet).
Now the acceptees have a couple months before the "official" start; other people might be ready to apply now but it's a 6-month wait for the next batch to get in and 8 months to "start"! That's an eternity in Internet Time.
Another advantage would be allowing the continuous pool of acceptees to relocate to the #1 spot (Silicon Valley) regardless of the time of year. It's a little silly that the Bostonians have to move there afterward anyway -- it's like a conflict of interest for YC, "Help the individual start-ups by putting them in the #1 spot" or "Try to help the #2 start-up scene play catch-up via imports".
Of course I could be wrong about everything, it's up to you.
We would not have guessed you could pick startups in 10 minutes either, at first. We realized this when interviews used to be longer. We noticed that we usually knew after 10 minutes whether we wanted to fund a group, and that in cases where we didn't we used to just have to sit there talking for another 10 or 20 minutes even though we'd already made up our minds.
I'm not saying we're always right; far from it; but more time would make surprisingly little difference.
Any "X" can work REALLY well if you do it "right". Doing it right, now there's the trick...