I guess my preferred solution to the probability_get_any_job issue is generally to just effectively set it to 1. That way you can answer the question “impact of a grantmaker” rather than “impact of trying to be a grantmaker,” and the counterfactual impact of trying to become a grantmaker is P(become a grantmaker)*EV(grantmaking) with opportunity cost depending on your circumstances.
I guess my preferred solution to the probability_get_any_job issue is generally to just effectively set it to 1. That way you can answer the question “impact of a grantmaker” rather than “impact of trying to be a grantmaker,” and the counterfactual impact of trying to become a grantmaker is P(become a grantmaker)*EV(grantmaking) with opportunity cost depending on your circumstances.
Makes sense! Would add another 0.5 orders of magnitude vs. the unconditional estimate, so ~2 x-risk basis points.