Portfolio theory probably makes the most sense. If you think opportunities can't be compared across cause areas, otherwise, if you're committed to maximising expected choiceness, you should just do that By that, I mean strategy or approach where you committing dollars to buckets, presupposes some facts about the buckets themselves As in, it's not principle, first effective altruism per se There's no such thing as the marginal rate of substitution between two cause areas, only the rate of substitution between projects whatever cause areas they may be Private equity should be the right reference class for this.Because in private equity, you can't buy an index across a theme.You're only have particular opportunities, whatever in may be --- Debt and equity versus korea and donations, which is which which which which fits better and therefore, what can we learn from that like?What can we infer starting from that point Maybe career is debt because it can be redeployed faster, and you have higher like billings But donations have more of the property of what the money's there.It's kind of gone Not sure --- Side research project for someone if cause areas are asset classes.What's the correlation between them?Because that's really critical for portfolio theory E g long termism and animal welfare and maybe more correlated, because they're both exposed to aggregation Whereas for global health and long termism, I would expect them to be uncorrelated If global health proves to be "-un" wrong, I expect it'll be for reasons unrelated to long terrorism. Eg implementation But everything's correlated in of big enough disaster eg solipsism --- Securitisation of impact? Longtermists buy from short-termists Capital allocation Jacob and Esau? --- Reject thing that looks like MFT Boom, segue to MEC / MU --- - What looks like consumption and what looks like investment? --- And there should be nice symmetry between the invidual and the community cases: we each contain multitudes. # Other bits and pieces https://forum.effectivealtruism.org/posts/YhPWq784eRDr5999P/announcing-the-ea-donation-swap-system # Other notes - Does {EA forum post strategy} vary from standard aggregated portfolio? - Randomise preferences across N actors - Simulate game - What happens? - Are results stable / convergent? - Is there $Nth$-mover advantage? - Acknowledge but ignore (?) "cynical" strategies (per comments) But - uh oh - is risk-adjusted portfolio even what we want? - Moral Uncertainty: MFT - very legible / transferable - What are the analogies / disanalogies? Really(?) => optimise across {career, giving} - Maybe this is the only diversification that matters? Maybe build to an ITN-based approximation? Related to Rethink's cool tool?