We haven't decided for sure, but we have to change something, because reading all the applications is becoming overwhelming.
Why not have someone create a simple, private site which makes salient application information: a screenshot which may lead to the site's demo, video of founders, answers to questions. Then certain trusted people, partners or alumni, can rate the applicants over time. Instead of going days without shaving and being exhausted, you could spend a few minutes each day flipping through applications. Surely over time those you are interested in for an interview will be apparent. I'm pretty sure there was a reason not to use this approach, but it may be worth it to bring it up again.
I don't think voting would be a good way to do this. Now, YC applications are different from scholarship applications or honourary degree nominations (which are the areas where I have the most experience judging); but if they're not too different, there are going to be some very good applications which most people don't immediately recognize as good. I often find that out of a large committee, some of the best applications are only noticed by one person -- but when he/she says "seriously, I think this one is really worth looking at", everybody else will look closely enough to recognize the application's merits. Simply having everybody vote yea/nay will result in obscure applications being lost when people default to nay.
To fight against that -- maybe you could make the total number of votes an application has gotten (it's score) invisible to the people doing the voting?
I think groupthink is far more of a problem when people see "hey, this has been voted up to 10" than when people hear a reasoned argument for why they should read an application which they previously ignored.
I wasn't talking about doing polls only. Each application would get a discussion thread, so there is room for someone to write a passionate essay or whatever.
I agree that there are issues with exposing voting. Maybe it would be interesting to shuffle the number of votes at first, see what happens when an application that was below the fold shoots to the top. (Maybe news.yc should do this too?)
The advantage of voting is that you can quickly flag wastes of time. In, e.g., fiction publishing, the ratio of acceptable to crap is 1 to 100 or worse, because any idiot can and does write a novel. You don't need to know everything as a reader to be able to eliminate the crap.
I think what you are really worried about is a situation where you have so many winners that there's no time to find the diamond in the rough. I don't think that's different from these startups' current situation though. They need to stand head and shoulders above the competition in some manner. They might as well start now.
Exactly. The startup would have to opt to be publicly vetted, but it would make a lot of sense. The worries of group-think are overrated. I think most would happily allow it and be grateful for the feedback and to see how they rate.
It takes some of the stress off of YC and lets a very smart community get involved. It's a win-win, really.
You can deal with that using some scarcity. Give each team 10 apps to rate. You could allow them to vote "good" or "bad" for all apps, but give them one "Drop everything and fund this" star that they can dole out. Ideally, distribute enough that each app gets rated by 2 or 3 teams and you could sort the list by star count first, and yes-to-no ratio second.
I don't think that this system necessarily would require a strict yes / no vote. I would imagine interesting commentary and feedback springing up around each idea. This system would give the entrepreneur a chance to defend his/her idea or iterate on some of the shortcomings others address. PG et al are betting on this iterative process occurring during YC anyway, why not accelerate it by making it part of the application process?
Maybe they can use http://alpha.crowdmind.com to facilitate the application review process. Going to release private group collaboration functionality soon, this could help when comparing the applications.
Why not have someone create a simple, private site which makes salient application information: a screenshot which may lead to the site's demo, video of founders, answers to questions. Then certain trusted people, partners or alumni, can rate the applicants over time. Instead of going days without shaving and being exhausted, you could spend a few minutes each day flipping through applications. Surely over time those you are interested in for an interview will be apparent. I'm pretty sure there was a reason not to use this approach, but it may be worth it to bring it up again.