Time, obviously. There's just four people in YC. pg has written about this on numerous occasions. Space is also a concern...the Anybots office holds about 20 companies, and then begins to get cramped and stuffy. Regardless, if they feel good about your prospects they will accept you, regardless of how many others have been accepted (though there is an upper bound, I suspect the pressure is applied during the written application phase rather than the interview phase...and there is an upper bound to the number of interviews they can reasonably handle in one weekend).
Most other Angel, VC, and incubator programs run a "rolling admissions" process. That would seem to take some of the time pressure out of the equation, but "time" as a limiting factor sounds more like the residue of inadequate productivity or know-how, or a function of the processes and structures the YC founders have put in place.
In other businesses, for example, if the number of startups that could take part in the program and offer a possibility of profit/upside or what have you for the investment of time and $ the business would find a way to address at least a majority of the demand (or unmet need). If YC had been funded by YC I would suspect they would be encouraging themselves to expand in a way that continued to create a compelling experience and a reasonable expectation of profit. Perhaps they are, since they talk about taking on more startups each time. I was struck by this article by George Gedron on the critical need for entrepreneurial education in the real world in Inc. http://www.inc.com/magazine/20071001/guest-speaker-the-real-...
It seems that YC is one solution (and if they make a nickel while doing so, more power to them). But Gedron isn't limited by time (any more than any of us are) because he is building a scalable repeatable model targeted at undergraduate education. I am not trying to play devil's advocate, I see a need for a factor of 100 to 1,000 scale up in what YC is doing (defined more broadly as entrepreneurial education, not just aimed at web applications).
Batching "admissions" is net more efficient. Plus the whole structure of YC requires it.
We do keep expanding. Every batch has had more startups than the one before. And we're always looking for ways to scale. Just this fall I wrote a bunch of software to keep track of and communicate with all the startups. We'd be lost otherwise, now that the total is up to 80.
I think there are many aspects of the model that can't be scaled via technology. I'm pretty certain that face to face time is really important, and I'm absolutely certain that the casual friendliness of the dinners is vital. Somewhere between the size of a dinner and the size of a Demo Day or party crowd the interactions break down into mingling rather than having actual conversations. I don't know where that size is, but there is definitely a cutoff. The dinner I attended in WFP08 seemed a wee bit too crowded. The noise level was a bit uncomfortable...only towards the end of the evening did I find I was able to carry on a normal conversation at normal levels. I think that makes a very real difference.
Of course, given that I consider the other founders and that intimate dinner experience every week the most valuable part of the program (with pg and Co. being merely the enabler of those gatherings)...one could theoretically split it into two weekly dinner groups (or three, or four, as YC patience and stamina allows), but then you run into the problem of getting good speakers in such abundance. pg would then be running a restaurant rather than a startup starter, though. If a few more good success stories come out of YC, the speaker pool could be bolstered by former YC founder appearances. I enjoyed Sam's talk as much as most (Joe Kraus is still my favorite of the lot).
So far the limiting constraint has been the number of good applicants. But time is a close second, and may become the first constraint unless we can think of clever ways to make ourselves more efficient.
