The SBF example is a poor one that’s obfuscatory of the basic point because you don’t address the hard question of whether his fraud-funded donations were or weren’t worth the moral and reputational damage, which is debatable and a separate interesting topic I haven’t seen hard analysis of
I’ve wondered about this as well. Scott Alexander’s mistake #57 seems like a relevant starting point:
57: (12/4/23) In In Continued Defense Of Effective Altruism, I said of EA’s failures (primarily SBF) that “I’m not sure they cancel out the effect of saving one life, let alone 200,000”. A friend convinced me that this was an unfair exaggeration of the point I wanted to make. There are purported exchange rates between money and lives, destroying $5 - $10 billion in value is pretty bad by all of them, and there are knock-on effects on social trust from fraud that suggest its negative effects should be valued even higher. I regret this sentence and no longer stand by it.
One guess as to how Scott’s link (US VSL) might underestimate the value destruction is that in Nov 2021, GiveWell aimed to direct ~$1bn annually by 2025, and the year after they revised downward their future funding projections due in part to their main donor Open Phil revising downward its planned 2022 allocation to GW due to a ~40% reduction in their asset base since the end of last year from “the recent stock market decline” changing their portfolio allocation, in particular more than proportionally reducing their GW allocation (partly offset by GW overachieving by ~40% RFMF-wise in finding cost-effective opportunities). My low-confidence guess is that “the recent stock market decline” has quite a bit to do with FTX. It seems likely to me that the NPV of the projected funding reduction to GW is >$1bn over the next say decade at ~5k per life saved (~1,000x the US VSL Scott linked to), or >200k lives that could have been saved but weren’t, most of them children under 5. (This galls me, to be honest, so I’d like to be told my reasoning is wrong or something.)
That’s one guess; I’m sure there are more I’m missing that’s still BOTEC-able, let alone the knock-on effects on social trust from fraud Scott mentioned.
I’ve wondered about this as well. Scott Alexander’s mistake #57 seems like a relevant starting point:
One guess as to how Scott’s link (US VSL) might underestimate the value destruction is that in Nov 2021, GiveWell aimed to direct ~$1bn annually by 2025, and the year after they revised downward their future funding projections due in part to their main donor Open Phil revising downward its planned 2022 allocation to GW due to a ~40% reduction in their asset base since the end of last year from “the recent stock market decline” changing their portfolio allocation, in particular more than proportionally reducing their GW allocation (partly offset by GW overachieving by ~40% RFMF-wise in finding cost-effective opportunities). My low-confidence guess is that “the recent stock market decline” has quite a bit to do with FTX. It seems likely to me that the NPV of the projected funding reduction to GW is >$1bn over the next say decade at ~5k per life saved (~1,000x the US VSL Scott linked to), or >200k lives that could have been saved but weren’t, most of them children under 5. (This galls me, to be honest, so I’d like to be told my reasoning is wrong or something.)
That’s one guess; I’m sure there are more I’m missing that’s still BOTEC-able, let alone the knock-on effects on social trust from fraud Scott mentioned.
To the OP: I think it’s worth reflecting on the warning that maximization is perilous.