What about holding the additional $70k per startup in escrow and dividing it among those startups that deserve to stick around?
Basically, if you take 60 startups, give each $80k and put the $4.2 million (60 * $70k) in the bank. When that batch of YC startups gets to $20k left in the bank on average, figure out which ones show promise and which ones should be deadpooled. Assuming a 2/3 deadpool rate, that leaves 20 startups to divide the $4.2k among, or approximately $210k in convertible debt per startup to reach the next milestone.
This is more inline with your essay about the future of funding, in which you speculate that funding will become more continuous.
If we delayed the investment, the investors would lose money. The investors are counting on indexing to generate returns. If they invest at the start of the YC cycle, all the startups take it, including the one that will later turn out to be the star of the batch. By Demo Day, though, there is a good chance that the star will have started to emerge and already have enough money. So this strategy would cause the money to go to the medium-good startups, which is not good in a domain with a power-law distribution of outcomes.
This would all but put a nail in the coffin of those who are "lagging" behind. The first question an investor will ask is whether you received the additional $70k. No investor wants to institutionalize that 30% of their companies will receive bad signaling by default.
Basically, if you take 60 startups, give each $80k and put the $4.2 million (60 * $70k) in the bank. When that batch of YC startups gets to $20k left in the bank on average, figure out which ones show promise and which ones should be deadpooled. Assuming a 2/3 deadpool rate, that leaves 20 startups to divide the $4.2k among, or approximately $210k in convertible debt per startup to reach the next milestone.
This is more inline with your essay about the future of funding, in which you speculate that funding will become more continuous.
http://www.paulgraham.com/future.html