Till recently the two main bottlenecks in YC were my time and the size of our space. Which is why we recently hired Harj and expanded the orange room by a third. I'm not sure what the next bottleneck will turn out to be.
As I said in another comment on this thread, we approach scaling YC the same way we approach scaling software. You can never predict what the bottleneck will be till you hit it, so you just fix them as you hit them. If we did eventually hit a bottleneck that we couldn't fix, in the sense that if we continued to expand, the startups we funded would start to do worse, we'd stop expanding. But we clearly haven't hit such a bottleneck yet, and my experience of scaling stuff makes me cautious about predicting exactly where it might occur.
I thought about it a bit more, especially after seeing that article talking about how %74 of your startups from this round have taken funding or are profitable already -- which I saw as having only %26 of them not be successful (whether among users or VCs) in the super-short-term.
I think you could afford to increase that rate significantly -- if anything, it's too low! Do you have a better metric to gauge your expansion by?
How much of that tranche of un-profitable and un-funded startups has historically been just in continual stealth ramen mode, and how much is actual failure? It'd be really interesting if you could put up some anonymized statistics about the 207 startups from the perspective of the founders. I'd visualize it as a series of images for each quarter, with a grid of venn diagrams of none/dead/acquired/funded/profitable for each YC round up to that point. Put it in a slideshow so you can scrub back and forth in history. Would also work in table form with YC rounds on one axis and time (or time since YC) on the other. I know you're rightly hesitant to talk about YC startups that didn't do well, but I'm not really interested in them specifically, just the collective attrition/success rates over time.
http://ycombinator.com/party.html
Till recently the two main bottlenecks in YC were my time and the size of our space. Which is why we recently hired Harj and expanded the orange room by a third. I'm not sure what the next bottleneck will turn out to be.
As I said in another comment on this thread, we approach scaling YC the same way we approach scaling software. You can never predict what the bottleneck will be till you hit it, so you just fix them as you hit them. If we did eventually hit a bottleneck that we couldn't fix, in the sense that if we continued to expand, the startups we funded would start to do worse, we'd stop expanding. But we clearly haven't hit such a bottleneck yet, and my experience of scaling stuff makes me cautious about predicting exactly where it might occur.