Strangely enough, a lot of the most successful startups we've funded applied at the last minute. E.g. David Rusenko of Weebly told me he applied 30 min before the deadline.
That could be because a) anyone who can put together a coherent pitch in 30 minutes already knows the problem space and solution very well, and/or b) there isn't time to lie or exaggerate in 30 minutes, so if you fund them, they are actually as good as they said they were.
Mostly a). The people who apply at the last minute don't usually have the idea at the last minute. They apply with something they've already started working on, or at least thinking about.
Or c) a person applying at the last moment has enough confidence in his or her product that applying to YC seems optional rather than urgent.
That said, it would be strange if all the people who applied at the last moment, even all the successful YC alumni who did, did so for the same primary reason. I'd wager that it's more that there's a cluster of desirable founder traits that correlates with applying late, of which in any particular good founder a different one can dominate.
anyone who can put together a coherent pitch in 30 minutes already knows the problem space and solution very well
I'd guess most of the successful YC companies knew the problem space very well when they applied. That's probably a universal prerequisite to making something people want.
Is there any data available on how much forethought and planning went into the decision to apply and the app/vid themselves?
From what I've learned talking to YC startups and reading about others it seems pretty mixed, that some spent many hours planning and some made the decision at the last minute.
I'd argue that the unpolished application videos are the best. I really like the video that the guys from DirectedEdge submitted (http://www.youtube.com/watch?v=3nTFBO5lvGc). This video demonstrates that there is chemistry between the two founders, identifies the problem/opportunity, and clearly explains their solution. I wasn't in the room, but this video appears to be spontaneous.
I can't speak to the timing of it all. But, given the data points available I'd say rough videos and applications free of marketing speak are the best.
Any ideas which?