Hide table of contents

Short summary

  • I estimated the cost-effectiveness of seeking commitments from retailers in France, Italy and Spain to sell only farmed fish that are stunned before slaughter.
  • While this intervention could potentially affect 2 to 36 fish per dollar spent, the short duration of fish slaughter means this is equivalent to improving only 0.4 to 10 fish hours per dollar.
  • The relatively short amount of time affected limits how cost-effective this intervention appears when compared to other promising animal and human interventions, except perhaps under frameworks that prioritize averting short-duration but high-intensity suffering.
  • Support for this intervention could potentially be justified on account of moral and empirical uncertainty around the severity and duration of pain, or to help lay the ground for future farmed fish welfare interventions.

Longer summary

  • I estimated how cost-effective seeking farmed fish slaughter commitments from retailers might be in Europe using a Monte Carlo simulation model. I focused on France, Italy, and Spain because of the large scale of consumption of farmed fish species (Gilthead Seabream, European Seabass, and small Rainbow Trout) that don’t benefit from stunning in these countries.
  • My estimates suggest that, if initial pilot work proves successful, future marginal spending in these countries could potentially benefit between 2 and 36 animals per dollar spent. This is somewhat below estimates of historic chicken campaigns (10 to 280 animals per dollar), but is plausibly comparable to marginal chicken grants once decreases in cost-effectiveness are taken into account. However, the number of animals affected is several magnitudes lower than estimates of ongoing shrimp stunning efforts (880 to 1900 shrimp per dollar per year).
  • Farmed fish slaughter commitments might only affect between 0.4 and 10 fish hours per dollar. This is several orders of magnitude shorter than estimates for historic chicken campaigns (10 to 120 chicken years per dollar), largely reflecting the very short duration of fish slaughter (5 to 40 minutes for Seabream and Seabass) compared to the lifespan of layer hens and broilers.
  • Farmed fish slaughter commitments do not look cost-effective using an estimation method that prioritizes duration and does not assign a high moral value to averting intense suffering. Using philosophical assumptions and placeholder estimates from the Rethink Priorities Moral Weights Project, I estimated the cost-effectiveness of farmed fish slaughter commitments to be between $10.4K and $114M per DALY averted. To put these numbers in context, $50 per DALY averted is considered a proxy for some of the most promising human global health and development interventions. The best animal interventions are often considered to be even more competitive than this.
  • But farmed fish slaughter commitments could seem more competitive in estimation frameworks that prioritize averting the most intense types of suffering. Fish slaughter commitments beat a theoretical funding bar of $50 per DALY in 50% of simulations for someone who was willing to avert 1 year of fish suffering during slaughter over gaining 72 years of human life at full health.
  • Allocating resources for farmed fish slaughter corporate commitment work could potentially be justified on account of moral and empirical uncertainty around the duration and severity of pain. It could also be justified to better understand the tractability of fish welfare work in these countries, and to help lay the ground for non-slaughter welfare improvements (for example, by getting retailers to recognize that fish are sentient).
  • Corporate commitments for non-slaughter welfare improvements have the theoretical potential to look more cost-effective than slaughter improvements. But further work needs to be done to establish what the most tractable asks might be. It’s also unclear how competitive such interventions will look once the costs of developing consensus for specific welfare asks and running feasibility pilots is taken into account.
Comments5


Sorted by Click to highlight new comments since:

Thanks for sharing, Sagar!

Farmed fish slaughter commitments do not look cost-effective using an estimation method that prioritizes duration and does not assign a high moral value to averting intense suffering.

I do think your methodology greatly underestimates the value of averting extreme suffering. You say:

To calculate the direct $/DALY range, I assumed that the stunning intervention leads to a welfare improvement between 10% and 50% of the entire fish welfare range (both positive and negative), that lasts for 50% to 90% of the duration of slaughter without stunning.

[...]

To convert this into DALYs averted, I assumed that an intervention that lasts for a year and generates an average welfare improvement of 50% of the entire human welfare range is equivalent to averting a DALY.

The assumptions in the 1st of the above paragraphs make sense to me if I interpret the welfare range as i) the difference between the welfare per unit time of the best and worst possible experience of 1 second, rather than ii) the difference between the welfare per unit time of the best and worst possible experience of 1 year (this can also be thought of as the difference between the welfare per unit time of the best and worst typical experience). However, for the 2nd paragraph to make sense, one has to interpret the welfare range as ii). So I think there is a contradiction:

  • Interpreting the welfare range as i), I would value 1 DALY as much less than a 50 % improvement in the welfare range, because being in extreme suffering is way worse than being in full health.
  • Interpreting the welfare range as ii), the improvement due to the stunning intervention would be way larger, because avoiding extreme suffering is way more important than extending full health.

Am I missing something?

I agree with this comment. It's worth nothing that the methodology used in this analysis isn't the same as the methodology used in the CURVE sequence. In the "How Can Risk Aversion Affect Your Cause Prioritization" report, @Laura Duffy weighted 1 year of disabling pain at 2 to 10 DALYs and 1 year of excruciating pain at 60 to 150 DALYs. I expect that dying is at least disablingly painful and potentially excruciatingly painful, so these weights would imply a >5x improvement in cost-effectiveness (but even at the upper end, this probably wouldn't be cost-competitive with top EA interventions). 

In general, I think it's a good step to try and actually put interventions from different cause areas on the same scale, but I continue to think that because DALYs are a unit of health status and not a unit of utility, trying to use them as a unit of comparison is unlikely to be optimal (see here and here for more)

Thank you for your comments, Matt!

I would agree on that this intervention would look better (in $/DALY space) if I were to have adopted the same assumptions as @Laura Duffy and come up with some plausible assumptions how much time in various pain intensities that would be averted through the intervention.  I also think its very unlikely the intervention would look competitive the top AW and GHD interventions.  Under the assumptions where this intervention were to look very competitive, I'd suspect shrimp stunning interventions would look even better.

Thanks also for your very valid comments on using DALYs as a unit to compare interventions (and your general engagement on the research that @Rethink Priorities does!).

Thank @Vasco Grilo  you for your thoughtful comments.  Appreciate it!

I don’t think you’ve missed anything. I think you've identified a very valid critique of the assumptions I used to express cost-effectiveness as a cost per DALY averted range.  Expressing the welfare ranges in units of seconds or years is a great way of bringing this out – so thank you for doing that.

Some comments:

  • If I were to rebuild the cost-effectiveness model, with the benefit of hindsight (and more time), I’d have probably used a probabilistic rather than deterministic variable for the assumption converting the % improvement in the human welfare range (for one year) that is equivalent to averting a DALY.
  • I’m pretty sure assumption feeds through linearly into the $/DALY results.  So if you believed an assumption of 5% of human welfare range was more appropriate than 50%, you could divide the  5th and 95th percentiles of the cost per DALY averted range.
  • The formal sensitivity tests I did suggest the conclusions of how this intervention looks compared to the most promising GHD and animal welfare interventions wouldn’t change with ‘relatively small’ adjustments to the assumptions needed to convert results into DALY space (e.g.  doubling the fish welfare range relative to humans and assuming averting a DALY is equivalent to a human intervention that raises human welfare by 10% of the human welfare range for 1 year).
  • I think once you start making bigger adjustments to these assumptions, you can run into the risk of being criticised for placing too much moral value on short-duration but high intensity suffering.  I don’t think we have good empirical evidence to support any particular assumption here.
  • The moral value section of the results more formally illustrates how the fish stunning intervention compares to various $/DALY benchmarks depending on the moral value you might assign to improving a year of fish life via the intervention relative to averting a DALY.
  • I don’t think the narrative expressed in the executive summary would change even if I were to change the assumption on the moral value of averting intense suffering relative to extending healthy lifespan.
  • While I think there is a lot of value in trying to place results into a ‘common currency’, I think this is also a good reason why cost per DALY averted numbers should always be treated with some caution (there is will always be moral value judgements there, some of which may be objectionable).  I think it’s valuable important to look at a number of different metrics (number of animals affected, amount of time affected) to assess how promising an animal welfare intervention looks.

I estimated the cost-effectiveness of farmed fish slaughter commitments to be between $10.4K and $114M per DALY averted.

Animal welfare is often conceptualised as just one area, but the above illustrates the cost-effectiveness can vary a lot depending on the species and type of intervention. So I think it is great that you are willing to estimate the cost-effectiveness in terms of DALY/$[1].  It makes me more willing to donate to Rethink Priorities instead of Animal Charity Evaluators' (ACE's) recommended charities or the Animal Welfare Fund (AWF), because these are not estimating cost-effectiveness in terms of DALY/$ (or similar). Both ACE and AWF assess cost-effectiveness based on heuristics[2], but I am not confident they are sufficient to figure out which are the best animal welfare interventions. I see GiveWell's cost-effectiveness analyses as quite important to determine the best interventions in global health and development, so I assume having similar analyses in the context of animal welfare is quite useful too.

To put these numbers in context, $50 per DALY averted [or 0.02 DALY/$ (= 1/50)] is considered a proxy for some of the most promising human global health and development interventions. The best animal interventions are often considered to be even more competitive than this.

For reference, I guess the cost-effectiveness of corporate campaigns for chicken welfare is 13.6 DALY/$ (= 0.01*1.37*10^3), i.e. 680 (= 13.6/0.02) times Open Philanthropy's bar. I got that multiplying:

  • The cost-effectiveness of GiveWell's top charities of 0.01 DALY/$ (50 DALY per 5 k$), which is half of Open Philanthropy's bar of 0.02 DALY/$.
  • My estimate for the ratio between cost-effectiveness of corporate campaigns for chicken welfare and GiveWell's top charities of 1.37 k (= 1.71*10^3/0.682*2.73/5):
    • calculated corporate campaigns for broiler welfare increase neaterm welfare 1.71 k times as cost-effectively as the lowest cost to save a life among GiveWell’s top charities then of 3.5 k$, respecting a cost-effectiveness of 0.286 life/k$ (= 1/(3.5*10^3)).
    • The current mean reciprocal of the cost to save a life of GiveWell’s 4 top charities is 0.195 life/k$ (= (3*1/5 + 1/5.5)*10^-3/4), i.e. 68.2 % (= 0.195/0.286) as high as the cost-effectiveness I just mentioned.
    • The ratio of 1.71 k in the 1st bullet respects campaigns for broiler welfare, but Saulius estimated ones for chicken welfare (broilers or hens) affect 2.73 (= 41/15) as many chicken-years.
    • OP thinks “the marginal FAW [farmed animal welfare] funding opportunity is ~1/5th as cost-effective as the average from Saulius’ analysis”.
  1. ^

    Although I think you are underestimating the cost-effectiveness.

  2. ^

    In addition, from Giving What We Can's evaluation of AWF:

    Fourth, we saw some references to the numbers of animals that could be affected if an intervention went well, but we didn’t see any attempt at back-of-the-envelope calculations to get a rough sense of the cost-effectiveness of a grant, nor any direct comparison across grants to calibrate scoring.

Curated and popular this week
Ben_West🔸
 ·  · 1m read
 · 
> Summary: We propose measuring AI performance in terms of the length of tasks AI agents can complete. We show that this metric has been consistently exponentially increasing over the past 6 years, with a doubling time of around 7 months. Extrapolating this trend predicts that, in under a decade, we will see AI agents that can independently complete a large fraction of software tasks that currently take humans days or weeks. > > The length of tasks (measured by how long they take human professionals) that generalist frontier model agents can complete autonomously with 50% reliability has been doubling approximately every 7 months for the last 6 years. The shaded region represents 95% CI calculated by hierarchical bootstrap over task families, tasks, and task attempts. > > Full paper | Github repo Blogpost; tweet thread. 
 ·  · 2m read
 · 
For immediate release: April 1, 2025 OXFORD, UK — The Centre for Effective Altruism (CEA) announced today that it will no longer identify as an "Effective Altruism" organization.  "After careful consideration, we've determined that the most effective way to have a positive impact is to deny any association with Effective Altruism," said a CEA spokesperson. "Our mission remains unchanged: to use reason and evidence to do the most good. Which coincidentally was the definition of EA." The announcement mirrors a pattern of other organizations that have grown with EA support and frameworks and eventually distanced themselves from EA. CEA's statement clarified that it will continue to use the same methodologies, maintain the same team, and pursue identical goals. "We've found that not being associated with the movement we have spent years building gives us more flexibility to do exactly what we were already doing, just with better PR," the spokesperson explained. "It's like keeping all the benefits of a community while refusing to contribute to its future development or taking responsibility for its challenges. Win-win!" In a related announcement, CEA revealed plans to rename its annual EA Global conference to "Coincidental Gathering of Like-Minded Individuals Who Mysteriously All Know Each Other But Definitely Aren't Part of Any Specific Movement Conference 2025." When asked about concerns that this trend might be pulling up the ladder for future projects that also might benefit from the infrastructure of the effective altruist community, the spokesperson adjusted their "I Heart Consequentialism" tie and replied, "Future projects? I'm sorry, but focusing on long-term movement building would be very EA of us, and as we've clearly established, we're not that anymore." Industry analysts predict that by 2026, the only entities still identifying as "EA" will be three post-rationalist bloggers, a Discord server full of undergraduate philosophy majors, and one person at
Thomas Kwa
 ·  · 2m read
 · 
Epistemic status: highly certain, or something The Spending What We Must 💸11% pledge  In short: Members pledge to spend at least 11% of their income on effectively increasing their own productivity. This pledge is likely higher-impact for most people than the Giving What We Can 🔸10% Pledge, and we also think the name accurately reflects the non-supererogatory moral beliefs of many in the EA community. Example Charlie is a software engineer for the Centre for Effective Future Research. Since Charlie has taken the SWWM 💸11% pledge, rather than splurge on a vacation, they decide to buy an expensive noise-canceling headset before their next EAG, allowing them to get slightly more sleep and have 104 one-on-one meetings instead of just 101. In one of the extra three meetings, they chat with Diana, who is starting an AI-for-worrying-about-AI company, and decide to become a cofounder. The company becomes wildly successful, and Charlie's equity share allows them to further increase their productivity to the point of diminishing marginal returns, then donate $50 billion to SWWM. The 💸💸💸 Badge If you've taken the SWWM 💸11% Pledge, we'd appreciate if you could add three 💸💸💸 "stacks of money with wings" emoji to your social media profiles. We chose three emoji because we think the 💸11% Pledge will be about 3x more effective than the 🔸10% pledge (see FAQ), and EAs should be scope sensitive.  FAQ Is the pledge legally binding? We highly recommend signing the legal contract, as it will allow you to sue yourself in case of delinquency. What do you mean by effectively increasing productivity? Some interventions are especially good at transforming self-donations into productivity, and have a strong evidence base. In particular:  * Offloading non-work duties like dates and calling your mother to personal assistants * Running many emulated copies of oneself (likely available soon) * Amphetamines I'm an AI system. Can I take the 💸11% pledge? We encourage A