A post calling for more exploratory altruism that focuses on discovery costs associated with different potential interventions and the plausible ranges of impact of the associated intervention.
A public list that identified different unexplored, or underexplored, interventions could be really helpful.
I actually thought about this after listening to Spencer Greenbergâs podcast- his observation that we shouldnât think about personal interventions, like whether to try a new drug or adopt a habit, in terms of naive expected value, but rather in terms of variance in effect. Even if a drugâs average affect on someone is negative, if some people get a large benefit from it, it is worth testing to see if you are someone who benefits from it. If it really helps you, you can exploit it indefinitely, and if it hurts you, you can just stop and limit the bad effect.
Likewise a lot of really good interventions may have low ânaive EVâ; that is to say, if you were to irrevocably commit to funding it, it would be a poor choice. But the better question is, is this an intervention that could plausibly be high EV and have high room for exploitation? What are the costs associated with such discovery? With such an intervention, you could pay the discovery costs and exploit if it turns out to be high EV, and cut losses if it does not. It is worth considering that many promising interventions might look like bad bets at the outset, but still be worth discovery costs given the ability to capitalize on lottery winners.
I understand that there are many organizations like CEARCH and Rethink Priorities (and many others) that are involved in cause prioritization research. But I think if one of them, or another org, were to compile a list that focused on search costs and plausible ranges of impact, making this publicly available and easy for the public to convey thoughts and information, this could be a very useful tool to spot promising funding/âresearch/âexperimentation opportunities.
A post calling for more exploratory altruism that focuses on discovery costs associated with different potential interventions and the plausible ranges of impact of the associated intervention.
A public list that identified different unexplored, or underexplored, interventions could be really helpful.
I actually thought about this after listening to Spencer Greenbergâs podcast- his observation that we shouldnât think about personal interventions, like whether to try a new drug or adopt a habit, in terms of naive expected value, but rather in terms of variance in effect. Even if a drugâs average affect on someone is negative, if some people get a large benefit from it, it is worth testing to see if you are someone who benefits from it. If it really helps you, you can exploit it indefinitely, and if it hurts you, you can just stop and limit the bad effect.
Likewise a lot of really good interventions may have low ânaive EVâ; that is to say, if you were to irrevocably commit to funding it, it would be a poor choice. But the better question is, is this an intervention that could plausibly be high EV and have high room for exploitation? What are the costs associated with such discovery? With such an intervention, you could pay the discovery costs and exploit if it turns out to be high EV, and cut losses if it does not. It is worth considering that many promising interventions might look like bad bets at the outset, but still be worth discovery costs given the ability to capitalize on lottery winners.
I understand that there are many organizations like CEARCH and Rethink Priorities (and many others) that are involved in cause prioritization research. But I think if one of them, or another org, were to compile a list that focused on search costs and plausible ranges of impact, making this publicly available and easy for the public to convey thoughts and information, this could be a very useful tool to spot promising funding/âresearch/âexperimentation opportunities.