Earlier quoted context omitted.
This is an interesting point on the timeline for YC in that their apps increased 40%! PG says they now have 3 silos, like 3 mini YC's reviewing. But if you have 1400 alums reviewing apps, by definition you are going to have scoring that is biased against finding black swans, regardless of how many silos you create. As you say, the problem is too many apps and not enough time for a smaller amount of alums to review th…
"PG says they now have 3 silos, like 3 mini YC's reviewing." Can you point me to where PG said this? "But if you have 1400 alums reviewing apps, by definition you are going to have scoring that is biased against finding black swans, regardless of how many silos you create." Why do you think the scoring is biased against black swans?
As far as the black swans, I could be wrong on this, but my understanding is that YC has maybe 10 partners all giving a grade on the apps coming in (apparently started this after they missed SendGrid). Im not sure if the 1400 alums initially flag or what, but even with the 10 partners grading, their odds of finding black swans get worse because black swans don't look like winners to most people - they are essentially counter intuitive, crazy ideas that don't appeal to the majority in the beginning (or maybe the founders are not Stanford drop outs or ex Googlers, which further sways the majority to give higher grade to what is more obvious. Thus pushing down more potential black swans below the interview cut off line.
So the more partners you have grading a particular startup, the more partners you have trying to cover their asses and not fuck up. So the safer choices will get more interviews - and safe choices don't usually result in black swans.
Not sure if I'm making sense here, but maybe the proof is in the pudding in that some of their biggest wins are Dropbox, Reddit, and Airbnb...all relatively early in YC batches when there were fewer partners making choices.