The appendix of the ebook (pdf, p. 218 & 221) suggests structuring it like tax brackets, giving (as a percentage of adjusted gross income):
1% of the first $81,000, 5% of the next $59,000, 10% of the next $180,000, 15% of the next $160,000, 20% of the next $1,520,000, 25% of the next $9,000,000, 33.3% of the next $42,000,000, and 50% of the remainder
I would rather see people make bets that they think are very profitable (relative to the size of the bet).
There's this idea that betting on your beliefs is epistemically virtuous, which sometimes leads people to be so eager to bet that they make bets at odds that are roughly neutral EV for them. But I think the social epistemic advantages of betting mostly depend on both parties trying to make bets where they think they have a significant EV edge, so sacrificing your EV to get some sort of bet made is also sacrificing the epistemic spillover benefits of the bet.
Wouldn't a person's "welfare footprint" also include, e.g., all the cases where they brightened someone's life a little bit by having a pleasant interaction with them? The purpose ("different animal products have vastly different welfare harms") seems fairly narrow but the term suggests something much broader.
I expect the causal effect to be pretty weak - if some light drinkers become nondrinkers, I guess that would lead to some reduction in the amount of heavy drinking, but not very much.
There are larger social influences from "the 5 closest people in your life [being] heavy drinkers" or from social rituals that actively pressure people into heavy drinking, but if the recommendation is for light drinkers to become nondrinkers that doesn't directly touch either of those causal pathways.
I basically disagree with this take on the discussion.
Most clearly: this post did generate a lot of pushback. It has more disagree votes than agree votes, the top comment by karma argues against some of its claims and is heavily upvoted and agree-voted, and it led to multiple response posts including one that reaches the opposite conclusion and got more karma & agree votes than this one.
Focusing on the post itself: I think that the post does a decent job of laying out the reasoning for its claims, and contains insights that are relevant and not widely considered/discussed, especially for readers who already (e.g.) think that fish matter to a non-negligible extent and are willing to avoid eating salmon because of that. It takes a lot more bee-days to make a serving of honey than salmon-days to make a serving of salmon, and when Rethink Priorities looked into moral weights they found as about as strong a case for bee sentience/welfare as they did for carp or salmon. So the claim that one should avoid eating honey for basically the reasons given in the post seems like a plausible hypothesis, especially conditional on the views that lead many EAs to avoid eating factory farmed chicken or salmon, and it seems good that people are thinking through that claim.
I don't buy this post's conclusion, for reasons that are pretty well covered in existing responses. And I agree that there are some epistemic problems with the way it lays out its arguments, such as using a kilogram-to-kilogram comparison instead of serving-to-serving, incorrectly claiming that RP found that bees display "every" behavioral proxy of consciousness, and its choices about when to label estimates as being "conservative". And it's not good that these all point in the same direction, of overstating the case for avoiding honey.
So we at least have a fair amount of agreement about the post itself. But turning it into an exhibit of a general EA-vegan-specific problem feels like a stretch, especially given the response it has received. And even just considering the post by itself, there are posts which have gotten a positive reception on LW that I've found to be worse-argued than this. (Though perhaps that could be flipped around to say that there are some bad epistemic patterns in both intellectual subcultures.)
I'm unsure how to interpret "will probably doom". 2 possible readings:
A highly technologically advanced civilization that tries to get really big will probably wind up wiping itself out due to the dynamics in this post. More than half of all highly technologically advanced civilizations that grow really big go extinct due to drastically increasing their attack surface to existential threats.
The following claim is probably true: a highly technologically advanced civilization that tries to get really big will almost certainly wind up wiping itself out due to the dynamics in this post. Almost every very large, highly technologically advanced civilization that grows big has a doom-level event spawn in a pocket of the civilization and spread to the rest of it.
The 2nd reading is big if true - it implies that the EV of the future arising from our civilization is much lower than it would otherwise seem, and that civilizations might be better off staying small - but I disagree with it.
For the first one I'm on the agree side, but it doesn't change the overall story very much.
Causality moves at c, so if we have probes moving away from each other at nearly 2c, that suggests extinction risk could be permanently reduced to zero.
This isn't right. Near-speed-of-light movement in opposite directions doesn't add up to above speed of light relative movement. e.g., Two probes each moving away from a common starting point at 0.7c have a speed relative to each other of about 0.94c, not 1.4c, so they stay in each other's lightcone.
(That's standard special relativity. I asked o3 how that changes with cosmic expansion and it claims that, given our current understanding of cosmic expansion, they will leave each other's lightcone after about 20 billion years.)
How about 'On the margin, work on reducing the chance of our extinction is the work that most increases the value of the future'?
As I see it, the main issue with the framing in this post is that the work to reduce the chances of extinction might be the exact same work as the work to increase EV conditional on survival. In particular, preventing AI takeover might be the most valuable work for both. In which case the question would be asking to compare the overall marginal value of those takeover-prevention actions with the overall marginal value of those same actions.
(At first glance it's an interesting coincidence for the same actions to help the most with both, but on reflection it's not that unusual for these to align. Being in a serious car crash is really bad, both because you might die and because it could make your life much worse if you survive. Similarly with serious illness. Or, for nations/cities/tribes throughout history, losing a war where you're conquered could lead to the conquerors killing you or doing other bad things to you. Avoiding something bad that might be fatal can be very valuable both for avoiding death and for the value conditional on survival.)
Related: Sublinear Utility in Population and other Uncommon Utilitarianism by Alice Blair
Alexander Wales added an Unused Starfish Facts post, with some things that he didn't fit into the story itself.
The appendix of the ebook (pdf, p. 218 & 221) suggests structuring it like tax brackets, giving (as a percentage of adjusted gross income):
I would rather see people make bets that they think are very profitable (relative to the size of the bet).
There's this idea that betting on your beliefs is epistemically virtuous, which sometimes leads people to be so eager to bet that they make bets at odds that are roughly neutral EV for them. But I think the social epistemic advantages of betting mostly depend on both parties trying to make bets where they think they have a significant EV edge, so sacrificing your EV to get some sort of bet made is also sacrificing the epistemic spillover benefits of the bet.
Wouldn't a person's "welfare footprint" also include, e.g., all the cases where they brightened someone's life a little bit by having a pleasant interaction with them? The purpose ("different animal products have vastly different welfare harms") seems fairly narrow but the term suggests something much broader.
I expect the causal effect to be pretty weak - if some light drinkers become nondrinkers, I guess that would lead to some reduction in the amount of heavy drinking, but not very much.
There are larger social influences from "the 5 closest people in your life [being] heavy drinkers" or from social rituals that actively pressure people into heavy drinking, but if the recommendation is for light drinkers to become nondrinkers that doesn't directly touch either of those causal pathways.
I basically disagree with this take on the discussion.
Most clearly: this post did generate a lot of pushback. It has more disagree votes than agree votes, the top comment by karma argues against some of its claims and is heavily upvoted and agree-voted, and it led to multiple response posts including one that reaches the opposite conclusion and got more karma & agree votes than this one.
Focusing on the post itself: I think that the post does a decent job of laying out the reasoning for its claims, and contains insights that are relevant and not widely considered/discussed, especially for readers who already (e.g.) think that fish matter to a non-negligible extent and are willing to avoid eating salmon because of that. It takes a lot more bee-days to make a serving of honey than salmon-days to make a serving of salmon, and when Rethink Priorities looked into moral weights they found as about as strong a case for bee sentience/welfare as they did for carp or salmon. So the claim that one should avoid eating honey for basically the reasons given in the post seems like a plausible hypothesis, especially conditional on the views that lead many EAs to avoid eating factory farmed chicken or salmon, and it seems good that people are thinking through that claim.
I don't buy this post's conclusion, for reasons that are pretty well covered in existing responses. And I agree that there are some epistemic problems with the way it lays out its arguments, such as using a kilogram-to-kilogram comparison instead of serving-to-serving, incorrectly claiming that RP found that bees display "every" behavioral proxy of consciousness, and its choices about when to label estimates as being "conservative". And it's not good that these all point in the same direction, of overstating the case for avoiding honey.
So we at least have a fair amount of agreement about the post itself. But turning it into an exhibit of a general EA-vegan-specific problem feels like a stretch, especially given the response it has received. And even just considering the post by itself, there are posts which have gotten a positive reception on LW that I've found to be worse-argued than this. (Though perhaps that could be flipped around to say that there are some bad epistemic patterns in both intellectual subcultures.)
I'm unsure how to interpret "will probably doom". 2 possible readings:
The 2nd reading is big if true - it implies that the EV of the future arising from our civilization is much lower than it would otherwise seem, and that civilizations might be better off staying small - but I disagree with it.
For the first one I'm on the agree side, but it doesn't change the overall story very much.
This isn't right. Near-speed-of-light movement in opposite directions doesn't add up to above speed of light relative movement. e.g., Two probes each moving away from a common starting point at 0.7c have a speed relative to each other of about 0.94c, not 1.4c, so they stay in each other's lightcone.
(That's standard special relativity. I asked o3 how that changes with cosmic expansion and it claims that, given our current understanding of cosmic expansion, they will leave each other's lightcone after about 20 billion years.)
How about 'On the margin, work on reducing the chance of our extinction is the work that most increases the value of the future'?
As I see it, the main issue with the framing in this post is that the work to reduce the chances of extinction might be the exact same work as the work to increase EV conditional on survival. In particular, preventing AI takeover might be the most valuable work for both. In which case the question would be asking to compare the overall marginal value of those takeover-prevention actions with the overall marginal value of those same actions.
(At first glance it's an interesting coincidence for the same actions to help the most with both, but on reflection it's not that unusual for these to align. Being in a serious car crash is really bad, both because you might die and because it could make your life much worse if you survive. Similarly with serious illness. Or, for nations/cities/tribes throughout history, losing a war where you're conquered could lead to the conquerors killing you or doing other bad things to you. Avoiding something bad that might be fatal can be very valuable both for avoiding death and for the value conditional on survival.)