PG, just outsource the entire selection process to Mechanical Turk. It could be an experiment similar to that proposed in the finance classic "A Random Walk Down Wall Street" in which a blindfolded monkey throwing darts at a stock page out-invests stock-picking experts. If Mechanical Turk picks a more successful crop of startups than the YCombinator average, you will have succeeded in making yourself completely redundant -- a worthy goal of any enlightened manager.
That only works until enough people start doing it. If you think that blind faith in quantitative factors beats a qualitative analysis of the effects of such faith, I can probably rustle up a few CDOs to sell you.
If the applicants have a demo-ready B2C product, MTs [1] might even be consistently better at picking winners than YC, as the ultimate success depends on what MTs think, since MTs[2] approx= common people approx= customers.
[1] Given enough MTs
[2] Provided you restrict allowed MT demographic to match your target market
Yes, which is why I said "If [they] have a demo-ready B2C product". YC is more early stage than what you can meaningfully show MT.
Also: If users love your product (not just the idea) and pay, you have already demonstrated that you are a good team, and it is pretty easy to persevere when you have an increasing number of paying customers.
