Mainly because we're not finished yet reading all the applications. When someone gets an exploding termsheet, we can all read their application immediately, but we can't do that for everyone.
Also, Robert flies in for the real interviews. When we interview people early, we have to do it without him.
Also, purely curious question: when you're looking through applications, is there a bar of "this is how many people we'll interview"? Or do you set a bar for quality-of-app, then take as many people as place above that bar?
There's a numeric bar for how many people we'll interview, based on how many groups we can physically talk to in 3 days. We usually interview 50 to 60 groups; it's hard to talk to more than 20 per day. At interviews, though, we care more about quality than numbers. We wouldn't reject a good group because we were over some number.
Well, people who interview when there's a full moon tend to get in, but of course it all depends on what pg had for breakfast that particular day.
Joking aside, it's probably best not to publish that kind of thing so that people inadvertently worry about something they can't control. The only reason to do those sorts of stats might be for pg & co. to check and make sure they don't have some sort of biases, but it doesn't seem likely.
The first part regd. time of the day was asked since PG mentioned (not sure if it was an essay or a comment on HN) that he used to get mundane stuff done during the day, and then code from 9 pm to 3 am.
This is not to say, PG ignores the early interviews and kicks it into high gear for the evening ones; just that if you look at the results, are there any insights or patterns.
I agree that YC should not publish these stats, but if they did learn anything interesting that could be shared with everyone, it would make a good read.
Anecdotally, I feel like we made a mistake choosing to be the first interview on Day 1 last season. You only get 10 minutes and they (as you'd expect) weren't quite yet "in the zone" regarding the daily schedule and interview rhythm, so our interview probably wasn't as focused as the later interviews got. We had lots of fun talking with them, though, and the "first interview of the season" could have easily swung the other way in terms of being a good time to present.
But, maybe because they hadn't gotten warmed-up yet, we didn't even get to talk about the startup idea we came to talk about. We shouldn't have mentioned our existing site, but they seemed so excited about it that the 10 minutes was over in a flash without getting to the business idea we submitted.
No matter when you interview, though, know that you don't get those 10 minutes back. Focus on what you think has the highest percentage chance to get you accepted. For us, we started off with an exciting idea that just happened not to be aggressive enough for that particular round.
This year might be better off because you've already submitted a video that gives you a headstart, but make the most of your interview time, regardless of timeslot.
Not YC specific, but from what I've heard from friends that have pitched VCs fairly often, earlier is better because that way you're not the n-th pitch they've heard that day.
I know when I've been at startup events where there's a lot of groups pitching I'm much more attentive for the first ones. Pitch #24 has to really be something for me to even remember what they're doing.
Hopefully there's little to no bias based on scheduling, but even if there was, publishing the statistics would only exacerbate the problem.
Not only would PG et al be thinking about this while the interviewed people ("This doesn't really seem interesting, but is that just because they're the 18th pitch today?"), but it would also distract the startups, when all they should be focusing on showing YC how awesome they are.
As far as I can tell, the people at YCombinator are extremely qualified to pick the best companies out of the group, and they generally do a pretty great job of it.
