FWIW, based on some of my experiences both inside and outside of EA, I quite strongly agree with your analysis. (But several caveats, e.g., ~all of my experience is from outside the US. I’d also potentially change my mind a lot if I saw outcome statistics from previous fellowships based on sound metrics—i.e., ones that get at more than “number of people who completed a fellowship” and shed light on how this affected the participants.)
I’d love to see more experiments with the kind of alternative models you’re proposing (but haven’t thought particularly hard about whether these in particular are the best experiments one could try).
I’d be very happy to give feedback on plans if that is ever helpful.
FWIW, based on some of my experiences both inside and outside of EA, I quite strongly agree with your analysis. (But several caveats, e.g., ~all of my experience is from outside the US. I’d also potentially change my mind a lot if I saw outcome statistics from previous fellowships based on sound metrics—i.e., ones that get at more than “number of people who completed a fellowship” and shed light on how this affected the participants.)
I’d love to see more experiments with the kind of alternative models you’re proposing (but haven’t thought particularly hard about whether these in particular are the best experiments one could try).
I’d be very happy to give feedback on plans if that is ever helpful.