In my mind there is no question that the limiting factor is Paul's time. The amount of time Paul spends with his portfolio of companies is the integral of the number of companies he funds with 50% decay to account for companies that quit. There is obviously a physical limit at 24 hrs/day.
Right now it's the number of good applications. That number appears to be increasing fairly rapidly, though. Then it will be our attention.
So there are basically two factors then:
1) Smart people within YC
2) Smart people applying to YC
So basically you have to find the optimal ratio and then grow both in proportion. :-)
That sounds like it puts you in approximately the same boat as traditional VCs.
Paul, there is nothing to the amount of attention given to applicants after acceptance?
Out of curiosity, what do you think of the work done at the MIT Media Lab? Would you fund those sorts of things?
Amount of money available to invest?
How much attention you can pay to the people you fund?
The number of applications that don't suck?
And given an unlimited supply of the current limiting factor, how much bigger could you get before running up against one of the other two?