Will proliferating seed companies start looking like micro sub-prime mortgage traders? They operate on the similar assumptions: pooling small, high-risk investments hoping that these individual bets are unrelated, thus driving down the variance when bundled together. This is probably only true for the seed company that gets to cherry pick first, which seems to be YC as of now.
Assuming an initial chance of success of 1/100 for the groups that are selected by YC, the chance that every single one from a group of 100 fails is about .36. That means that on an investment of about 1.5 million dollars (100 * 15,000) they have a 64 percent chance of a success, which I loosely define as a big acquisition or buyout. Also, YC companies seems to prop each other up and drive up success rates through intelligent nurturing and beneficial network effects. I'd guess that the payout probabilities are much higher over a three-four year period.<p>Now I am highly skeptical that the same math extends to other seed companies. Do they get applicants of the same caliber when there are so many second tier seed companies to choose from for a prospective applicant? People are willing to uproot their youthful lives and move for YC, but will they do the same for Techstars or LaunchBox? Probably not, at least not for 15,000 dollars and little else. If I throw in the assumption of a power-law distribution for startups, the quality goes downhill rather rapidly. And additionally, if YC happens to be decently good at identifying truly promising candidates and taking them all into their fold as they claim to, it the median success probability for the rest is going to be rather low. Assuming an order of magnitude drop-off, that gives an payout probability of .09
( 1-(1-1/1000)^100 if they are operating on same scales as used for YC). The number of seed companies sprouting is surely going to be limited by such math...