If you’re going to make an argument like this you should specify how you weight different animals, including wild ones, and include welfare estimates for wild animals. I have some discussion in this post.
For example, health interventions that save lives, like bed nets, may lower human fertility in the long run, resulting in lower meat consumption.
This is not obvious. See David Roodman’s write-up. Using the criteria expressed elsewhere in the post (total QALY-counting without a specified weighting scheme, direct effects and indirect effects on meat consumption but ignoring longer-term flow-through effects), this would be an argument for health interventions being bad, along Repugnant Conclusion lines.
A more practically sound approach, to me, seems to be to consider where the distribution of donations in the Effective Altruist movement generally falls and allocate funding as “bets” according to how much an organization could use money at the margins and its expected value. Under this line of thinking, I believe we may be under-prioritizing animal organizations, far future research, and meta-organizations like Giving What We Can.
This could work smoothly in an expected value framework for interventions with strong diminishing returns over the range of money EA can move, and would be recommended by relevant moral uncertainty or moral pluralism (this is relevant for many Open Philanthropy activities), although the cutoffs for each intervention will depend on relative weightings/credences/shares of influence.
However, it doesn’t work in the same way for interventions with near-constant returns over larger ranges. For instance, scaled-up cash transfers could absorb hundreds of billions of dollars per year with fairly steady marginal returns.
Yep, I think the discussion around how much we value different animal lives is pretty central to this. I think it deserves a post on its own—perhaps that’s the next thing I’ll write!
I think you’re right in theory about interventions with constant returns, but I’m not sure many interventions actually behave this way. To take GiveDirectly, I see one of the most large (potential) benefits being that developing countries may begin cash transfer systems after seeing GiveDirectly’s success. To that end, $50 million looks very different from $150 to $300 in how quickly countries will hear about their successes, how much media attention GiveDirectly receives, etc. It’s probably very impossible to predict where these cut-offs are—I’m just trying to highlight that optimizing our donations is of course what we should aim for, but pretty hard when many of the benefits come from policy changes from a diverse set of actors.
If you’re going to make an argument like this you should specify how you weight different animals, including wild ones, and include welfare estimates for wild animals. I have some discussion in this post.
This is not obvious. See David Roodman’s write-up. Using the criteria expressed elsewhere in the post (total QALY-counting without a specified weighting scheme, direct effects and indirect effects on meat consumption but ignoring longer-term flow-through effects), this would be an argument for health interventions being bad, along Repugnant Conclusion lines.
This could work smoothly in an expected value framework for interventions with strong diminishing returns over the range of money EA can move, and would be recommended by relevant moral uncertainty or moral pluralism (this is relevant for many Open Philanthropy activities), although the cutoffs for each intervention will depend on relative weightings/credences/shares of influence.
However, it doesn’t work in the same way for interventions with near-constant returns over larger ranges. For instance, scaled-up cash transfers could absorb hundreds of billions of dollars per year with fairly steady marginal returns.
Yep, I think the discussion around how much we value different animal lives is pretty central to this. I think it deserves a post on its own—perhaps that’s the next thing I’ll write!
I think you’re right in theory about interventions with constant returns, but I’m not sure many interventions actually behave this way. To take GiveDirectly, I see one of the most large (potential) benefits being that developing countries may begin cash transfer systems after seeing GiveDirectly’s success. To that end, $50 million looks very different from $150 to $300 in how quickly countries will hear about their successes, how much media attention GiveDirectly receives, etc. It’s probably very impossible to predict where these cut-offs are—I’m just trying to highlight that optimizing our donations is of course what we should aim for, but pretty hard when many of the benefits come from policy changes from a diverse set of actors.
Good points, Carl!