Is it accurate that the lowest numbers you arrived at come from donating entirely to shrimp slaughter? Most farmkind donors probably don't agree that shrimp suffering can substitute for chicken or pig suffering, or if so, expect it to convert at a rate much less favorable to shrimps than you or ACE. Maybe if they deeply engaged with your reasoning around shrimps they would agree, but with more modest engagement I expect them to land closer to FarmKind's model. I think this is worth noting because I think it means FarmKind donors would endorse that model as matching their expectations on at least some reflection; only a worldview-changing amount of reflection would cause them to see your model as better matching what FarmKind describes.
Is it accurate that the lowest numbers you arrived at come from donating entirely to shrimp slaughter?
Yes. "I estimate donating 0.279 $/person-year (= 0.106*2.63) to SWP is enough to offset the suffering the food consumption of random people in the US causes to farmed animals". This is "assuming SWP is as cost-effective as HSI has been", and HSI advocates for electrically stunning shrimps.
Most farmkind donors probably don't agree that shrimp suffering can substitute for chicken or pig suffering, or if so, expect it to convert at a rate much less favorable to shrimps than you or ACE.
I assumed sentience-adjusted welfare ranges proportional to "individual number of neurons"^0.5, which implies that of shrimps is 0.1 % of that of humans. This ratio is 8.0 % in Bob Fischer's book about comparing welfare across species, which is what Rethink Priorities (RP) uses as the default value in their Donor Compass tool. ACE also uses the value in Bob's book or something very similar. So I would have obtained a lower offset using ACE's or RP's preferred value.
Maybe if they deeply engaged with your reasoning around shrimps they would agree, but with more modest engagement I expect them to land closer to FarmKind's model.
I expect them to land on a much smaller offset than FarmKind's as long as they endorse offsetting the suffering of all farmed animals by decreasing the suffering of chickens. ACE estimated THL's cage-free accountability work in 2024 decreased 88 SADs per $, and I calculated random people in the US increase the suffering of farmed animals by 138 SADs per person-year based on ACE's 2025 meal replacement model. In this case, random people in the US can offset the impact of their diets on farmed animals by donating 1.57 $/person-year (= 138/88) to interventions as cost-effective as THL's cage-free accountability work in 2024, which is 0.565 % (= 1.57/278) of the offset suggested by FarmKind.
@Aidan Kankyoku, here is a quick calculation illustrating that the suffering random diets in the US cause to poultry birds can be offset donating much less than the 185 $/person-year suggested by FarmKind. I am tagging you because this was not in my initial reply to your comment.
Here is a quick calculation illustrating that the suffering random diets in the US causes to poultry birds can be offset donating much less than the 185 $/person-year suggested by FarmKind. There were 4.72 poultry birds per person in the US in 2024. ACE estimated THL's cage-free accountability work in 2024 helped 11 layers per $. Say each layer lives for 1.34 years (= (60 + 80)/2*7/365.25), which is the mean between the lower and upper bound of the range of "60 to 80 weeks" assumed by the Welfare Footprint Institute (WFI). Then THL's cage-free accountability work in 2024 improved 14.7 layer-years per $ (= 11*1.34). I calculatethat moving 1.67 layer-years (= (0 - (-2.27))/(-0.333 - (-1.69))) from conventional cages to cage-free aviaries increases the welfare of chickens as much as removing 1 broiler-year in a conventional scenario. So I infer THL's cage-free accountability work in 2024 increased the welfare of chickens as cost-effectively as removing 8.80 broiler-years per $ (= 14.7/1.67). Suppose the conditions of poultry birds in the US are as bad as those of broilers in a conventional scenario (I think they are better due to layers in conventional cages having higher welfare than broilers in a conventional scenario). Random people in the US could offset the suffering their diets cause to poultry birds donating 0.536 $/person-year (= 4.72/8.80) to interventions increasing the welfare of chickens as cost-effectively as THL's cage-free accountability work in 2024, which is 0.290 % (= 0.536/185) of the offset suggested by FarmKind.
In all seriousness its an interesting question and I appreciate this wee write-up. Of your 2 questions "How much money is needed to help each type of farmed animal affected by the diet about as much as the diet harms it? Or how much money is needed to decrease the suffering of farmed animals as much as the diet increases it, regardless of which animals benefit?" I would only consider the first question strict offsetting. I think given that offsetting is already a controversial approach which some people disagree is valid, going one step removed and not offsetting the actual direct problem you are causing is perhaps too much of a stretch, and personally I wouldn't be OK with making that equivalency
When you offset carbon, its the same type of carbon being pulled out of the air that you put in, like for like. When you offset biodiversity, you pay (in theory) to restore/protect an equivalent habitat somewhere else. I think preventing suffering in shrimp to allay your conscience from causing pig suffering is something similar, but distinct.
This might be a straw man, but I would feel OK (ish) about supporting a charity that improved the welfare of miners in the DRC to allay my moral burden of the rare earth metals in my phone, but I wouldn't feel right "offsetting" that suffering I caused to miners by eating less meat and reducing the equivalent amount of chicken suffering - if you get my drift.
This is just my opinion btw, not any kind of standard practise. I may well also be flat wrong, I often am when it comes to moral philosophy.
I agree that different people might be trying to do different things, and I appreciate you outlining the costs of both, but I would rather not call the second question "offsetting"
Of your 2 questions "How much money is needed to help each type of farmed animal affected by the diet about as much as the diet harms it? Or how much money is needed to decrease the suffering of farmed animals as much as the diet increases it, regardless of which animals benefit?" I would only consider the first question strict offsetting. I think given that offsetting is already a controversial approach which some people disagree is valid, going one step removed and not offsetting the actual direct problem you are causing is perhaps too much of a stretch, and personally I wouldn't be OK with making that equivalency
From my perspective, what is offsetting and strict offsetting is a matter of how humans use language. Ambiguity can be resolved by making explicit what is being offset. The questions are about offsetting the expected suffering caused to farmed animals allowing or not for offsetting the suffering of one type via decreasing the suffering of other type. The donations needed to offset the harm increase as restrictions and target confidence increase. For instance, the donations needed to i) offset the expected suffering of each type of farmed animals with 90 % probability, and not allowing for offsetting the suffering of one type via decreasing the suffering of other type are much larger than the donations needed to ii) offset the expected suffering of farmed animals.
When you offset carbon, its the same type of carbon being pulled out of the air that you put in, like for like. When you offset biodiversity, you pay (in theory) to restore/protect an equivalent habitat somewhere else. I think preventing suffering in shrimp to allay your conscience from causing pig suffering is something similar, but distinct.
Nitpicks. There are greenhouse gases (GHGs) besides carbon dioxide, and I think people are fine with offsetting the global warming caused by one type of gas by decreasing the emissions of other type of gas. I also suspect many people would be fine with their impact on biodiversity as long as they are not, for example, increasing agricultural land, even if they are decreasing the area of some biomes.
I agree that different people might be trying to do different things, and I appreciate you outlining the costs of both, but I would rather not call the second question "offsetting"
You are in agreement with FarmKind's users.
Aidan Alexander and Thom Norman, co-founders of FarmKind, commented the Compassion Calculator deals with offsetting the suffering of different farmed animals in agreement with users’ expectations. Their view is based on conversations they have had with donors, and I do not dispute it.
I would agree it is better to use "offsetting" with fewer qualifications when the supported interventions more closely reverse the harm, which makes the 1st question more about offsetting than the 2nd.
I created this post using Claude Opus 5.5 to edit a draft entirely written by me. I wrote the vast majority of it in November 2025, and it already had all the calculations below.
Summary
The cost of offsetting the suffering a diet causes to farmed animals depends a lot on the question being asked. How much money is needed to help each type of farmed animal affected by the diet about as much as the diet harms it? Or how much money is needed to decrease the suffering of farmed animals as much as the diet increases it, regardless of which animals benefit? The cost also depends on the cost-effectiveness estimates used, and on how much they are adjusted for conservatism.
FarmKind’s Compassion Calculator answers the 1st question. It recommends donating 278 $/person-year to offset the suffering random Western adult omnivores cause to farmed animals. It models conservatively on purpose, as it wants donors to be confident that their donations are enough.
For the 2nd question, I estimate 2.04 $/person-year for random people in the United States (US) based on Animal Charity Evaluators’ (ACE’s) results, and 0.279 $/person-year based on my own numbers for the suffering caused to farmed animals, and the past cost-effectiveness of the Shrimp Welfare Project’s (SWP’s) Humane Slaughter Initiative (HSI).
Changing one modelling choice at a time in FarmKind’s spreadsheet shows how much each choice affects the cost. For example, offsetting the suffering of birds raised for meat with cage-free corporate campaigns changes the cost from 278 to 109 $/person-year, and barely requires comparing welfare across species, as almost all birds raised for meat are chickens.
Which estimate is more useful depends on what a donor wants an offset to do, and on how much margin they want for uncertainty.
Two questions an offset can answer
FarmKind estimates “how much it might cost to help animals as much as the average diet negatively impacts them”, based on the cost-effectiveness of interventions helping each type of animal involved in the production of beef, dairy, pork, poultry meat, eggs, and farmed finfishes and crustaceans. The average diet is “a weighted average of a handful of Western, English-speaking countries” (the US, the United Kingdom, Canada, Ireland, Australia and New Zealand), which “roughly reflect the donors that use FarmKind”. Aidan Alexander and Thom Norman, co-founders of FarmKind, commented the Compassion Calculator deals with offsetting the suffering of different farmed animals in agreement with users’ expectations. Their view is based on conversations they have had with donors, and I do not dispute it.
FarmKind explains its “outputs should be viewed as rough ballpark estimates rather than precise figures”, and that it adds “large discounts for conservatism” for donors to be confident they are donating enough. Its calculator says “there is a significant margin of error in these kinds of calculations, which we have accounted for by modelling conservatively”. In its spreadsheet, it applies “Adjustment[s] for conservatism”, explaining that “These estimates are necessarily highly uncertain and so we want to err on the side of conservatism as we’d rather people end up helping more than they intend than that people think they’ve fully offset their consumption when they haven’t”.
The 2nd question pools the suffering of all the animals involved, and offsets it with whatever decreases suffering most cost-effectively. Unless the animals helped are of the same species as the ones harmed, this requires comparing welfare across species.
The 2 questions suit donors with different goals. Donors who want the types of animals affected by their diet to be the ones helped may prefer the 1st. Donors who want to decrease the suffering of farmed animals as much as possible per $ may prefer the 2nd.
How each modelling choice changes the cost
Both FarmKind’s calculator and ACE’s estimates rely on Ambitious Impact’s (AIM’s) suffering-adjusted days (SADs). So I think the difference between 278 and 2.04 $/person-year mostly comes from other modelling choices. I started from FarmKind’s spreadsheet, and changed one modelling choice at a time. The table summarises the results.
Approach
$/person-year
FarmKind’s Compassion Calculator
278
FarmKind’s calculator with change 1) offsetting the suffering of birds raised for meat with cage-free corporate campaigns
109
FarmKind’s calculator with change 2) using ACE’s cost-effectiveness estimates for chicken welfare corporate campaigns
97.4
FarmKind’s calculator with change 3) removing the adjustments for conservatism
152
FarmKind’s calculator with changes 1 to 3
50.4
FarmKind’s calculator with changes 1 to 3, and change 4) offsetting the suffering of farmed finfishes by decreasing that of laying hens
12.2
Decreasing suffering as cost-effectively as possible, based on ACE’s results
2.04
Decreasing suffering as cost-effectively as possible, based on my own numbers
0.279
The list below explains each change, numbered as in the table:
Offsetting the suffering of birds raised for meat with cage-free corporate campaigns. FarmKind estimates the money per person-year needed to offset the suffering of birds raised for meat (chickens and turkeys) is 65.2 (= 182/2.79) times that needed to offset the suffering of laying hens. ACE estimates the suffering of farmed animals caused by the consumption of poultry meat in the US is 4.64 (= 0.0483/0.0104) times that caused by the consumption of eggs there, and I get a ratio of 2.99 (= -0.302/(-0.101)). Based on ACE’s ratio, and FarmKind’s estimate of the cost-effectiveness of cage-free corporate campaigns, offsetting the suffering of birds raised for meat with these campaigns requires 12.9 $/person-year (= 4.64*2.79) instead of 182 $/person-year, 169 $/person-year (= 182 - 12.9) less. So the overall cost becomes 109 $/person-year (= 278 - 169).
Using ACE’s cost-effectiveness estimates for chicken welfare corporate campaigns. FarmKind considers cage-free and broiler welfare corporate campaigns help 1.17 (= 1/0.854) and 0.885 (= 1/1.13) chickens per $, and describes its estimates as “conservative”. These are 10.6 % (= 1.17/11) and 1.92 % (= 0.885/46) of ACE’s estimates for The Humane League (THL). With ACE’s estimates, the overall cost becomes 97.4 $/person-year.
Removing the adjustments for conservatism. Without FarmKind’s adjustments for conservatism, the overall cost becomes 152 $/person-year.
Offsetting the suffering of farmed finfishes by decreasing that of laying hens. With changes 1 to 3, the overall cost becomes 50.4 $/person-year, and the money per person-year needed to offset the suffering of farmed finfishes is 321 (= 38.2/0.119) times that needed to offset the suffering of laying hens. ACE estimates the suffering of farmed animals caused by the consumption of finfishes in the US is 16.6 % (= 0.00173/0.0104) of that caused by the consumption of eggs there, and I get a ratio of 17.9 (= -1.81/(-0.101)). Based on ACE’s ratio, offsetting the suffering of farmed finfishes by decreasing that of laying hens requires 0.0198 $/person-year (= 0.166*0.119) instead of 38.2 $/person-year. So the overall cost with changes 1 to 4 becomes 12.2 $/person-year (= 50.4 - 38.2). This change requires comparing welfare across species.
Decreasing suffering as cost-effectively as possible
For the 2nd question, I estimate 2.04 $/person-year (= 138/67.5) for random people in the US from the ratio between:
Effects on farmed animals of 138 SADs per person-year (= 0.126*3*365.25) for the food consumption of random people in the US. I got this based on the estimate from ACE’s 2025 meal replacement model of 0.126 SADs per meal, and an assumption of 3 meals per person-day.
Decrease in the suffering caused to farmed animals of 67.5 SADs per $. I got this from a weighted mean of ACE’s estimates for the organisations they assessed and recommended in 2025, giving the same weight to each organisation, and to each program within a given organisation. I would prefer to use the cost-effectiveness of ACE’s Recommended Charity Fund, but I would be surprised if this meaningfully changed the result.
I also calculate donating 0.106 $/person-year to SWP is enough to offset the suffering food consumption globally causes to farmed animals, assuming SWP is as cost-effective as HSI has been. ACE estimates the food consumption of a random person in the US causes 2.63 (= 0.126/0.0479) times as much suffering to farmed animals as that of a random person globally. So I estimate donating 0.279 $/person-year (= 0.106*2.63) to SWP is enough to offset the suffering the food consumption of random people in the US causes to farmed animals. This is lower than the estimate based on ACE’s results mainly because I think ACE greatly underestimates the intensity of excruciating pain, and therefore how much HSI increases the welfare of shrimps. ACE assumes 7.06 h of “awareness of Pain is likely to be present most of the time” (hurtful pain) is as painful as 1 h of “severe burning in large areas of the body, dismemberment, or extreme torture” (excruciating pain).
All these estimates are very uncertain. Donors who want a margin for the uncertainty can donate more than the estimates above, which is the reasoning behind FarmKind’s conservative modelling. My estimates are for the US, whereas FarmKind’s is a weighted mean for 6 Western countries, and I have not adjusted for this.
Acknowledgements
Thanks to Aidan Alexander, Aidan Whitfield, Anonymous Person, Kevin Xia, Thom Norman, and Toby Tremlett for feedback on the draft. Aidan Alexander clarified he only skimmed the summary of the draft from November 2025. The views expressed in the post are my own.
Crossposted to Less Wrong
‘Like most of life’s dangers, the dangers of artificial intelligence are not easily identified from the armchair.’
J Dmitri Gallow
Epistemic status
I’m not an expert in this field, and herein I make some sweeping generalisations educated guesses about the overall...
Note: This post was crossposted from Planned Obsolescence by the Forum team, with the author's permission. The author may not see or respond to comments on this post.
Subtitle: If acceleration compounds, the naive extrapolation implies superintelligence within a year
This post discusses a public METR report on AI R&D acceler...
TL;DR:
* When we started Rethink Priorities (RP) in 2018, our mission was, simply put, to close the knowledge gaps that stopped people from improving the world as much as they could. That hasn’t changed much, but the landscape we’re doing it in is shifting.
* In 2027, we plan to grow our core GHD, Animal Welfare, and Worldview Investigations teams to take on emerging priorities across causes, including those that the forthcoming transformative AI is creating.
...
Is it accurate that the lowest numbers you arrived at come from donating entirely to shrimp slaughter? Most farmkind donors probably don't agree that shrimp suffering can substitute for chicken or pig suffering, or if so, expect it to convert at a rate much less favorable to shrimps than you or ACE. Maybe if they deeply engaged with your reasoning around shrimps they would agree, but with more modest engagement I expect them to land closer to FarmKind's model. I think this is worth noting because I think it means FarmKind donors would endorse that model as matching their expectations on at least some reflection; only a worldview-changing amount of reflection would cause them to see your model as better matching what FarmKind describes.
Hello Aidan.
Yes. "I estimate donating 0.279 $/person-year (= 0.106*2.63) to SWP is enough to offset the suffering the food consumption of random people in the US causes to farmed animals". This is "assuming SWP is as cost-effective as HSI has been", and HSI advocates for electrically stunning shrimps.
I assumed sentience-adjusted welfare ranges proportional to "individual number of neurons"^0.5, which implies that of shrimps is 0.1 % of that of humans. This ratio is 8.0 % in Bob Fischer's book about comparing welfare across species, which is what Rethink Priorities (RP) uses as the default value in their Donor Compass tool. ACE also uses the value in Bob's book or something very similar. So I would have obtained a lower offset using ACE's or RP's preferred value.
I expect them to land on a much smaller offset than FarmKind's as long as they endorse offsetting the suffering of all farmed animals by decreasing the suffering of chickens. ACE estimated THL's cage-free accountability work in 2024 decreased 88 SADs per $, and I calculated random people in the US increase the suffering of farmed animals by 138 SADs per person-year based on ACE's 2025 meal replacement model. In this case, random people in the US can offset the impact of their diets on farmed animals by donating 1.57 $/person-year (= 138/88) to interventions as cost-effective as THL's cage-free accountability work in 2024, which is 0.565 % (= 1.57/278) of the offset suggested by FarmKind.
@Aidan Kankyoku, here is a quick calculation illustrating that the suffering random diets in the US cause to poultry birds can be offset donating much less than the 185 $/person-year suggested by FarmKind. I am tagging you because this was not in my initial reply to your comment.