Chicken welfare reforms may be harmful accounting for effects on wild arthropods?
Summary
Broiler welfare and cage-free reforms change the amount of feed needed to produce a given quantity of chicken meat or eggs. So they affect feed production, and therefore result in land use change. The density and welfare conditions of wild animals vary by type of land, so chicken welfare reforms influence wild animal welfare.
I estimate broiler welfare (cage-free) reforms increase or decrease the welfare of wild arthropods 47.7 (4.66) times as much as they increase the welfare of broilers (hens). Nonetheless, the effects on arthropods can be anything from negligible to much larger. I estimate uncertainty in arthropods’ welfare range alone means broiler welfare reforms increase or decrease their welfare 0 to 1.74 k times as much as they increase the welfare of broilers, and cage-free reforms increase or decrease their welfare 0 to 170 times as much as they increase the welfare of hens (5th to 95th percentiles).
The effects of chicken welfare reforms on arthropods could be safely neglected if they were less than 10 % of those on chickens. Yet, I have a hard time seeing how one could be confident about this. Even just considering the uncertainty in each and one of the following inputs, for broiler welfare and cage-free reforms:
The increase or decrease in feed would have to be smaller than 0.0693 % and 0.108 %.
The increase or decrease in the density of arthropods would have to be smaller than 0.0630 % and 0.645 % of the density of arthropods in all of Earth’s land.
The increase or decrease in the welfare of arthropods would have to be smaller than 1.05*10^-7 and 1.08*10^-6 QALY/arthropod-year, which are 0.00525 % and 0.0540 % of Rethink Priorities’ (RP’s) median welfare range of silkworms of 0.002.
Increasing the welfare of arthropods by 1 QALY would have be less than 0.210 % and 2.15 % as valuable as increasing the welfare of chickens by 1 QALY, which implies strongly rejecting impartiality.
My results suggest it is unclear whether chicken welfare reforms are beneficial or harmful. The effects on arthropods may well be larger than those on chickens, which would imply chicken welfare reforms being beneficial/harmful if they benefit/harm arthropods. I think these conclusions apply to any intervention targeting vertebrates which change the consumption of feed or food, especially if it mainly aims to increase/decrease positive/negative vertebrate-years.
I recommend donating and making grants to improve invertebrate welfare. In particular, supporting the Arthropoda Foundation, RP’s work on invertebrate welfare, the Shrimp Welfare Project, or Wild Animal Initiative.
I encourage organisations helping vertebrates to consider how they can help invertebrates.
Organisations would need support from funders to change their focus. It would be great if people in Open Philanthropy’s (OP’s) farm animal welfare team or leadership could persuade their funders, Dustin Moskovitz and Cari Tuna, to fund invertebrate or wild animal welfare again.
Effects of cage-free reforms
Overview
Broiler welfare and cage-free reforms change the amount of feed needed to produce a given quantity of chicken meat or eggs. So they affect feed production, and therefore result in land use change. The density and welfare conditions of wild animals vary by type of land, so chicken welfare reforms influence wild animal welfare. I explain below how I estimated the effects of broiler welfare and cage-free reforms on wild arthropods and chickens.
Broiler welfare reforms
From Figure 4 of Gittins et al. (2024), broilers in a conventional scenario need 2.75 feed-kg/meat-kg, and ones in a reformed (slower growth) scenario, corresponding to the European Chicken Commitment (ECC), need 3.70 feed-kg/meat-kg. So I suppose broiler welfare reforms increase feed by 0.950 feed-kg/meat-kg.
According to Gittins et al. (2024), “Based on 5,674 million birds per year moving from standard to ECC production in the EU”, “Feed use would increase by some 7.30 million tonnes (+34.5%)”. “Assuming crop yields remain constant, an additional 1.57 million hectares of land would be needed for crop production”, which implies a cropland requirement of 2.15*10^-4 ha-year/feed-kg. Using this and the above increase in feed, I infer broiler welfare reforms increase cropland by 2.04*10^-4 ha-year/meat-kg, 2.04 m^2 for 1 year per kg of meat.
Rosenberg (2023) “found that there are ≈1 × 10^19 (twofold uncertainty range) soil arthropods on Earth, ≈95% of which are soil mites and springtails”. The Earth has 1.49*10^10 ha of land, so I arrive at a density of arthropods in all of Earth’s land of 671 M arthropod/ha. Based on Brian Tomasik’s results, I estimate the difference between the density of arthropods in rainforest and Cerrado is 90.0 % of the density of arthropods in all of Earth’s land. I speculate additional cropland decreases the number of arthropods 1⁄3 as much, by 30.0 % of the density of arthropods in all of Earth’s land. So I calculate additional cropland decreases the living time of arthropods by 201 M arthropod-year/ha-year. Combining this with the cropland requirement gives me a decrease in the living time of arthropods caused by broiler welfare reforms of 41.1 k arthropod-year/meat-kg.
The Welfare Footprint Institute (WFI) says the “average [mean] slaughter weight [just before slaughter] of broilers in the EU [European Union]” “in a conventional scenario” is “2.5 Kg”. Based on Table 11 of Gittins et al. (2024), “the eviscerated carcass without the neck, internal organs or abdominal fat” accounts for 72.65 % and 71.05 % of that for broilers in a conventional and reformed scenario, which correspond to 1.82 and 1.78 meat-kg for WFI’s slaughter weight. WFI mentions the slaughter weight for broilers in a conventional scenario in the EU is “reached at 42” “days”, 0.115 broiler-years, and that, in “the reformed scenario”, the “same slaughter weight would be reached in approximately 56 days”, 0.153 broiler-years. So I determine broilers in a conventional and reformed scenario produce 15.8 and 11.6 meat-kg/broiler-year. Multiplying the 1st of these by the result of the previous paragraph, I conclude broiler welfare reforms decrease the living time of arthropods by 650 k arthropod-year/broiler-year.
I guess the absolute value of the (expected) welfare of arthropods is 25 % of the welfare of fully healthy arthropods. A uniform distribution ranging from −75 % to 25 %, whose mean is −25 %, and one ranging from −25 % to 75 %, whose mean is 25 %, would lead to that absolute value of the welfare. I set the welfare of fully healthy arthropods to 2*10^-4 QALY/arthropod-year, which is 10 % of RP’s median welfare range of silkworms of 0.002. So I obtain an increase or decrease in the welfare of arthropods of 5.00*10^-5 QALY/arthropod-year. I do not know whether arthropods would be benefited or harmed. I am uncertain about whether cage-free reforms increase or decrease feed production, as Gemini’s ranges for the FCR of egg production in cages and barns overlap a lot. Furthermore, it is unclear to me whether increased feed production increases or decreases arthropod-years, and whether arthropods have positive or negative lives.
Based on the decrease in the living time of arthropods, and their welfare, each broiler-year improved by broiler welfare reforms increases or decreases their welfare by 32.5 QALYs.
In agreement with my cost-effectiveness analysis (CEA) of corporate campaigns for chicken welfare, I suppose broilers in conventional and reformed scenarios have a welfare of −0.754 and −0.0535 QALY/broiler-year. These mean together with the aforementioned meat production values that broilers in conventional and reformed scenarios have a welfare of −0.0477 and −0.00462 QALY/meat-kg. So I deduce broiler welfare reforms increase welfare by 0.0431 QALY/meat-kg. Multiplying this by the meat production of broilers in a conventional scenario, I arrive at an increase in welfare of 0.681 QALY/broiler-year.
Cage-free reforms
I asked Gemini 2.5 Pro (experimental) on 11 April 2025 about the feed conversion ratio (FCR) of current egg production. Gemini provided a range of 1.9 to 2.1 feed-kg/egg-kg for cages, and 2.0 to 2.2 feed-kg/egg-kg for barns. The means between the lower and upper bounds of the ranges for cages and barns are 2.00 and 2.10 feed-kg/egg-kg, so I assumed cage-free reforms increase feed by 0.100 feed-kg/egg-kg. From Table 3 of Leinonen et al. (2012), which studied egg production in the United Kingdom (UK), hens in cages and barns consume 2.56 and 2.74 feed-kg/egg-kg, which imply cage-free reforms increase feed by 0.178 feed-kg/egg-kg. The lower increase suggested by Gemini’s values might be explained by the faster growth of cage-free production since 2012, which facilitates economies of scale. It makes sense hens in barns need more feed because they can move around, and therefore spend more energy.
Class A eggs from hens in cages and barns in the EU had a mean cost in the 7 days before 9 April 2025 of 2.60 and 2.85 €/egg-kg. From these, and a “commonly assumed” price elasticity of demand for eggs of −0.15, cage-free reforms make consumption 94.6 % (= (2.85/2.60)^-0.15) as high. I did not account for this in my main estimate for the effects on arthropods. The decrease of 5.40 % (= 1 − 0.946) in the consumption of eggs is negligible in light of other uncertainties, and some of it would be replaced by increased consumption of other animal-based products, which also require some feed production. Replacement by animal-based products with higher FCR might even imply higher egg prices increase feed production overall.
From Table 4 of Leinonen et al. (2012), the cropland requirement of the production of non-organic hens’ feed is 1.55*10^-4 ha-year/feed-kg. Organic hens’ feed requires 3.23 times as much cropland, thus implying a larger impact on arthropods. I assume non-organic feed, as I guess it is the most commonly used by far. Combining the cropland requirement with the increase in feed mentioned above, I infer cage-free reforms increase cropland by 1.55*10^-5 ha-year/egg-kg, 0.155 m^2 for 1 year per kg of eggs. Combining this with my estimate for how additional cropland affects arthropods gives me a decrease in their living time caused by cage-free reforms of 3.12 k arthropod-year/egg-kg.
From Table 3 of Leinonen et al. (2012), I use an egg production for hens in cages and barns of 19.5 and 19.0 egg-kg/hen. In addition, I consider a life expectancy of 1.46 hen-years, “20 weeks” for the pre-laying phase from WFI, plus 56 weeks for the laying phase from Table 1 of Leinonen et al. (2012). So I determine hens in cages and barns produce 13.4 and 13.1 egg-kg/hen-year. Multiplying the 1st of these by the result of the previous paragraph, I conclude cage-free reforms decrease the living time of arthropods by 41.8 k arthropod-year/hen-year.
Based on the decrease in the living time of arthropods, and their welfare, each hen-year improved by cage-free reforms increases or decreases their welfare by 2.09 QALYs.
In agreement with my cost-effectiveness analysis (CEA) of corporate campaigns for chicken welfare, I suppose hens in (conventional) cages and barns have a welfare of −0.563 and −0.111 QALY/hen-year. These mean together with the aforementioned egg production values that hens in cages and barns have a welfare of −0.0420 and −0.00850 QALY/egg-kg. So I deduce cage-free reforms increase welfare by 0.0335 QALY/egg-kg. Multiplying this by the egg production in cages, I arrive at an increase in welfare of 0.449 QALY/hen-year.
Comparisons
I estimate broiler welfare (cage-free) reforms increase or decrease the welfare of wild arthropods 47.7 (4.66) times as much as they increase the welfare of broilers (hens). Nonetheless, the effects on arthropods can be anything from negligible to much larger. There is large uncertainty respecting many of the inputs. RP’s 5th and 95th percentile welfare range of silkworms are 0 and 36.5 (= 0.0073/0.002) times their median. I used a welfare range for arthropods equal to 10 % of silkworms’, so I guess the 5th and 95th percentile welfare range of arthropods are also 0 to 36.5 times their median. Consequently, I estimate uncertainty in arthropods’ welfare range alone means broiler welfare reforms increase or decrease their welfare 0 to 1.74 k (= 4.66*36.5) times as much as they increase the welfare of broilers, and cage-free reforms increase or decrease their welfare 0 to 170 (= 47.7*36.5) times as much as they increase the welfare of hens (5th to 95th percentiles).
The increase or decrease in the welfare of arthropods as a fraction of the increase in the welfare of chickens is 10.2 (= 47.7/4.66) times as large for broiler welfare reforms as for cage-free reforms. This is explained by the increase in feed-kg/meat-kg caused by broiler welfare reforms being 9.50 (= 0.950/0.100) times the increase in feed-kg/egg-kg caused by cage-free reforms, the increase in ha-year/meat-kg of the former being 1.39 (= 2.15*10^-4/(1.55*10^-4)) times the increase in ha-year/egg-kg of the latter, and the increase in QALY/meat-kg of the former being 1.29 (= 0.0431/0.0335) times the increase in QALY/egg-kg of the latter (9.50*1.39/1.29 = 10.2).
Discussion
Effects on arthropods
The effects of chicken welfare reforms on arthropods could be safely neglected if they were less than 10 % of those on chickens. Yet, I have a hard time seeing how one could be confident about this. The effects on arthropods of broiler welfare and cage-free reforms would have to be less than 0.210 % (= 0.10/47.7) and 2.15 % (= 0.10/4.66) as large as I estimated. Even just considering the uncertainty in each and one of the following inputs, for broiler welfare and cage-free reforms:
The increase or decrease in feed would have to be smaller than 0.0693 % (= 0.00210*0.33) and 0.108 % (= 0.0215*0.0500), as I estimated an increase of 34.5 % (= 0.950/2.75) and 5.00 % (= 0.100/2.00).
The increase or decrease in the density of arthropods would have to be smaller than 0.0630 % (= 0.00210*0.300) and 0.645 % (= 0.0215*0.300) of the density of arthropods in all of Earth’s land.
The increase or decrease in the welfare of arthropods would have to be smaller than 1.05*10^-7 (= 0.00210*5.00*10^-5) and 1.08*10^-6 QALY/arthropod-year (= 0.0215*5.00*10^-5), which are 0.00525 % (= 1.05*10^-7/0.002) and 0.0540 % (= 1.08*10^-6/0.002) of RP’s median welfare range of silkworms of 0.002.
I also guess I underestimated the effects on arthropods due to factory-farms with improved conditions requiring more land, and the additional land having a lower density of arthropods than the counterfactual land use.
My results suggest it is unclear whether chicken welfare reforms are beneficial or harmful. The effects on arthropods may well be larger than those on chickens, which would imply chicken welfare reforms being beneficial/harmful if they benefit/harm arthropods. I think these conclusions apply to any intervention targeting vertebrates which change the consumption of feed or food, especially if it mainly aims to increase/decrease positive/negative vertebrate-years:
I estimated GiveWell’s (GW’s) top charities increase or decrease the welfare of wild terrestrial arthropods 1.15 k times as much as they increase the welfare of humans.
I estimated insecticide-treated nets (ITNs) distributed by the Against Malaria Foundation (AMF) cause 763 times as much harm to mosquitoes as they benefit humans.
I estimate replacing meat from broilers in conventional scenarios with plant-based food requiring as much cropland as the broilers’ feed increases or decreases the welfare of arthropods 138 (= 47.7*2.89) times as much as it increases the welfare of hens. 2.89 (= 2.75/0.950) times as much as broiler welfare reforms because this is the ratio between the FCR of broilers in conventional scenarios, and the increase in feed-kg/meat-kg caused by broiler welfare reforms.
I estimate replacing chicken eggs from hens in cages with plant-based food requiring as much cropland as the hens’ feed increases or decreases the welfare of arthropods 93.2 (= 4.66*20.0) times as much as it increases the welfare of hens. 20.0 (= 2.00/0.100) times as much as cage-free reforms because this is the ratio between the FCR of hens in cages, and the increase in feed-kg/egg-kg caused by cage-free reforms.
Here are some updates to past claims I have made based on my very uncertain best guess that increasing cropland harms arthropods due to decreasing positive arthropod-years:
I have been saying corporate campaigns for chicken welfare are way more cost-effective than GW’s top charities. I now expect their cost-effectiveness to be similarly negative. I estimate broiler welfare and cage-free campaigns increase cropland by 0.00968 and 0.00224 ha-year/$, which are close to my rough estimate for GW’s top charities of 0.0129 ha-year/$.
I have argued replacing chicken meat with beef or pork is beneficial. I now expect the opposite to be true. Beef or pork require more cropland than chicken meat.
I estimated Veganuary in 2024, and School Plates in 2023 were 1.20 % and 19.4 % as cost-effective as cage-free campaigns, and I believe decreasing the consumption of animal-based foods is usually less cost-effective than improving the conditions of farmed animals ignoring effects on wild animals. I now expect cage-free campaigns to be harmful, and decreasing the consumption of animal-based products to be beneficial. The former increases cropland, whereas the latter decreases it as long as the animal-based foods are mostly replaced with plant-based ones.
I do not take any of these updates seriously. I am practically agnostic about whether increasing cropland benefits or harms arthropods, and therefore very uncertain about whether interventions with dominant effects on arthropods are beneficial or harmful.
Interventions targeting vertebrates which change the consumption of feed or food can be beneficial due to increasing future support for wild invertebrates, even if it is unclear whether they are beneficial or harmful nearterm, but I am sceptical. I guess such indirect effects are too small to dominate the nearterm effects, and I am not confident they are beneficial. Brian Tomasik thinks “it’s pretty unclear whether promoting vegetarianism reduces or increases total animal suffering, both when considering short-run effects on wild animals on Earth and when considering long-run effects on society’s values”. I agree with Brian on this, and the following.
General concern for animal suffering is crucial if humans are to make wise choices with respect to wild animals once more advanced technologies arrive, and promoting vegetarianism based on reducing suffering seems generally likely to cultivate such sympathies.
However, advancing vegetarianism from, say, the perspective that humans have no right to interfere with animals or that meat causes destruction of intrinsically valuable nature could be counterproductive. Indeed, I find it likely that vegetarianism increases support for wilderness conservation on balance, and the negative impacts of wilderness preservation on wild-animal suffering could be substantial. This concern suggests that the indirect, memetic effects of vegetarianism could very well be net harmful.
Relatedly, I have the impression people working on reducing the consumption of animal-based products are often less sympathetic to wild animal welfare than people working on improving the conditions of farmed animals.
My recommendations
I recommend donating and making grants to improve invertebrate welfare. In particular, supporting:
They aim to improve the conditions of farmed insects, whose population is expected to grow a lot. RP forecasts there will be 417 billion black soldier flies and mealworms in 2033, 10.9 (= 417*10^9/(38.2*10^9)) times as many as in 2023, and 15.3 (= 417*10^9/(27.2*10^9)) times the number of chickens in 2023.
They are not targeting other farmed arthropods, or wild animals for now, but a better understanding of insects seems good longterm to help wild arthropods.
RP’s work on invertebrate welfare:
I feel it has historically been quite cost-effective. It has helped build the invertebrate welfare movement, and informed my cost-effectiveness analysis of the Shrimp Welfare Project.
RP’s work on invertebrate and wild animal welfare is only being supported by restricted funds. So one can make restricted donations to those areas without worrying about them being partially offset by a reallocation of unrestricted funds.
The Shrimp Welfare Project (SWP):
I estimate it has been 64.3 k times as cost-effective as GW’s top charities neglecting the effects of these on animals, and 173 times as cost-effective as cage-free corporate campaigns neglecting the effects of these on wild animals.
It has plans I like for what to do with over 1 M$ of additional funding.
It mainly advocates for electrically stunning shrimp before slaughter via the Humane Slaughter Initiative.
This can make shrimp slightly more expensive, decrease its consumption, decrease its feed, and therefore increase the number of wild animals, which may be beneficial or harmful.
On the other hand, lower consumption implies fewer farmed shrimp, which is beneficial if they have negative lives as I estimate.
Overall, I expect the effects on wild animals to be much less important for interventions which do not change the feed required to produce a kg of animal-based products, such as ones improving slaughter methods. Moreover, it is harder for effects on wild animals to make SWP harmful since I estimate it is much more cost-effective than cage-free campaigns.
The Wild Animal Initiative (WAI):
“To advance our understanding of wild animal well-being, our team conducts research and supports the growth of the wider research community”, in line with their research priorities.
I estimate paying farmers to use more humane pesticides would be 51.4 times as cost-effective as cage-free campaigns neglecting the effect of these on wild animals, and guess that research on and advocacy for more humane pesticides would be way more cost-effective than paying farmers to use them more. WAI has supported research on pesticides.
Wild animal welfare is super neglected.
I listed the organisations alphabetically. SWP and WAI are recommended by Animal Charity Evaluators (ACE), although this only plays a very minor role in my recommendations.
Here are some thoughts on other prominent organisations working on invertebrate or wild animal welfare:
Animal Ethics “provides information and resources about the reasons to respect all animals, promotes discussion and debate about issues in animal ethics, and encourages academics and scientists to do research in related areas”.
WAI seems much more well connected to academia than Animal Ethics, so I believe they are better positioned to drive academic interest.
I also suspect the impact of Animal Ethics’ broad outreach is limited by the lack of academic research on wild animal welfare. WAI fills this gap, thus legitimating more mainstream advocacy. Broad advocacy for the welfare of farmed animals would hardly lead to change if there was basically no research about their welfare, and how to improve it.
The Insect Institute “address[es] challenges and uncertainties related to the production and use of insects for food and feed. We aim to assist this novel industry, policymakers, and other interested parties by providing evidence-based information surrounding the rearing of insects and the creation of a food system that promotes public health, animal welfare, and sustainable protein production”.
I can see their 3 publications slightly slowing down the growth of the insect industry. Nevertheless, this is only clearly beneficial for farmed insects if their lives are negative now, and will remain so over the period affected by the slow down. It may be harmful if their lives are positive now, or will become so over the aforementioned period. I do not know whether farmed insects have positive or negative lives, so it is unclear to me whether preventing the growth of the insect industry is beneficial, especially because it could replace some farmed animals with negative lives like farmed fish. Knowing how much time farmed insects spend in pain, as determined by WFI for broilers and hens, and RP for shrimp, would be great to inform whether farmed insects have positive or negative lives.
The Arthropoda Foundation plans to improve the conditions of farmed insects, which is beneficial for these regardless of whether they have positive or negative lives.
I think improving the conditions of animals is usually more cost-effective than decreasing their consumption ignoring effects on wild animals.
Screwworm Free Future (SFF) is “A coordinated initiative to protect South America from the New World Screwworm” by eradicating it.
I assume this would be beneficial for the infected animals, but it is unclear to me whether it would be beneficial or harmful overall due to the possibility of screwworms having positive lives.
I think it is fine to pursue interventions which may be harmful to wild animals nearterm, but then it is important to learn from them to minimise harmful effects in the future.
I have not seen concern for the welfare of screwworms expressed in any of SFF’s posts. However, they say “Wild Animal Initiative is planning on funding a research investigating the welfare effects of screwworm eradication”. Mal Graham, WAI’s strategy director, told me “Any project we do on screwworms will include the effects on the screwworms themselves as well as the effects on wild animals” (Mal said I could share this).
I listed the organisations alphabetically.
I encourage organisations helping vertebrates to consider how they can help invertebrates. RP has a report on strategies to help farmed shrimp. Organisations working on chicken welfare corporate campaigns could run ones asking companies to commit to electrically stun shrimp, as Mercy for Animals (MFA) did to get a commitment from Tesco, UK’s Largest Supermarket Chain.
Organisations would need support from funders to change their focus. It would be great if people in Open Philanthropy’s (OP’s) farm animal welfare team or leadership could persuade their funders, Dustin Moskovitz and Cari Tuna, to fund invertebrate or wild animal welfare again.
Acknowledgements
Thanks to Keyvan Mostafavi, and Michael St. Jules for feedback on the draft. The views expressed in the post are my own.
Thanks so much for this thoughtful post, Vasco! It is so heartening to see people taking arthropod welfare seriously.
While I agree that chicken welfare reforms could plausibly harm arthropods more than they help chickens, I don’t think that means we shouldn’t support chicken welfare reforms. For the same reason I reject the meat-eater problem, the logic of the larder, and the logic of the logger, I think that to get to a society that maximizes utility over the long term, we will probably need to take some steps that decrease utility in the short term.
That is, I don’t know exactly what global-scale arthropod welfare programs would look like, but I think we’re more likely to get there if more people live in material abundance, so I think economic development is worthwhile even if it increases factory farming in the short term. I think reforming, regulating, and banning factory farming are also very likely to be helpful (and possibly necessary) for human society to normalize and institutionalize concern for non-human animals, and to invest substantial resources in helping them.
I realize this is a suspiciously convenient conclusion to come to, and I can’t rule out the possibility that my position is driven by motivated reasoning. But I think it’s a good sign that my claim (“Chicken welfare reforms might be good overall, even if they hurt arthropods in the short term”) uses a similar logic to yours (“Chicken welfare reforms might be bad overall, even if they help chickens directly”). Both are examples of finding a different conclusion as a result of changing the scope of the analysis: Should our calculations include just chickens directly affected, or also arthropods indirectly affected, or also farmed and wild animals very-indirectly affected?
Of course one could agree with me that we should include the very-indirectly-affected animals, but disagree with my guesses about what would be best for them. One of the biggest weaknesses of my approach is that it’s much harder to judge what kinds of are worthwhile, or to compare effectiveness across efforts. It also takes things further from ecology and more into social movement theory, which is annoying because I enjoy the former a lot more than the latter.
But that doesn’t mean we should abandon empiricism and settle for hand-waviness in everything. Analyses like yours can be very useful; I just think we should interpret their results in the context of explicit theories of change about long-term effects.
Thanks for the comment, Cam!
I tend to agree. Among 53 countries I analysed, there is a U-shaped relationship between the fraction of commercial egg production coming from cage-free hens, and the logarithm of the real gross domestic product (real GDP) per capita, which weakly suggests economic growth leads to improvements in the conditions of farmed animals.
However, I guess the beneficial longterm effects of economic growth on wild arthropods are not sufficiently large to dominate the neaterm effects. I estimated a spending on wild animal welfare in 2023 of 5.02 M 2023-$, 2.70*10^-8 (= 5.02*10^6/(166*10^12*1.12)) of the real GDP in 2023. So my prior is that one has to increase real GDP by 37.0 M 2023-$ (= 1/(2.70*10^-8)) to increase spending on wild animal welfare by 1 2023-$. In other words, the direct effect of increasing economic growth on wild animal welfare is super small. So I remain uncertain about whether it is beneficial or harmful.
I worry adocating for animal rights may be harmful due to encouraging the wilderness preservation (regardless of whether wild animals have positive or negative lives), and I think it is difficult to advocate for banning factory-farming without relying on animal rights. From a welfarist perspective, and ignoring effects on wild animals, lots of factory-farmed animals with positive lives would be much better than no farmed animals.
I very much agree one should in principle account for all effects. However, I am not aware of organisations mainly targeting farmed vertebrates or humans whose theories of change (TOCs) include effects on invertebrates. So I have little confidence about whether they are beneficial or harmful to wild invertebrates. I agree empirically-based explicit TOCs about effects on intertebrates would be useful.
My personal best guess is that wild arthropods have positive lives, and that broiler welfare and cage-free reforms decrease the welfare of arthropods 47.7 and 4.66 times as much as they increase the welfare of hens (as I estimate broiler welfare and cage-free campaigns decrease 1.95 M and 452 k arthropod-years per $). So I estimate broiler welfare and cage-free campaigns have a cost-effectiveness of −78.0 (= (1 − 47.7)*1.67) and −16.8 QALY/$ (= (1 − 4.66)*4.59), which, for my guesses of pain intensities, are equivalent to causing 7.80 (= 78.0/10) and 1.68 (= 16.8/10) hours of disabling pain in humans per $. I agree there is lots of uncertainty, but I think funding interventions which can be hugely harmful depending on how it resolves is not a good way of dealing with it. Funding organisations working on invertebrate welfare like the ones I mention in the post seems much more robustly positive, and also more cost-effective than vertebrate welfare interventions changing feed production even ignoring the effects of this on non-target animals.
I can’t think of a compelling theory of change for arthropod welfare that omits consideration of avian welfare, and I worry that your line of reasoning (despite the generally robust calculations you include) neglects two larger concerns:
The idea that, for the sake of increased total welfare, we should neglect a smaller but still significant harm, may border on naive utilitarianism. While chicken welfare reforms may, on balance, increase total suffering, there is no doubt that they directly and significantly reduce the immense suffering of factory-farmed chickens. The horrors of factory farming are too great to dismiss, even if in this case there is an extremely uncertain chance that they might reduce harm elsewhere. A better solution could be to work with farmed chicken welfare advocates to investigate ways of rearing chickens more humanely which do not (or are at least less likely to) increase insect suffering.
Outside of (and sometimes even inside) EA, insect welfare is a tough sell. I think an even tougher sell is insect welfare which comes at the cost of the welfare of birds, mammals, or other animals. Like Cameron, I have a hard time imagining a society that takes insect welfare seriously while subjecting farmed animals of other species to intense suffering, and so I worry that moving away from this kind of chicken welfare reform actually moves us further from adopting the reforms needed to decrease insect suffering in the long term.
That being said, I do think this is well worth discussing, and I appreciate your efforts to move things along with quantitative data!
Thanks for the comment, Samuel!
One hallmark of naive utilitarianism is neglecting uncertain effects which are important in expectation. This is in contrast with my post, where I consider uncertain effects on wild arthropods. I conclude these can easily be larger than those on chickens, but this is not the same as neglecting the effects on chickens, which I also modelled.
Changing donations from organisations helping farmed vertebrates to ones helping invertebrates, as I recommend, may come at the cost of the welfare of farmed vertebrates. Similarly, changing donations from organisations helping people in extreme poverty to ones helping farmed vertebrates may come at the cost of the welfare of people in extreme poverty. Does this necessarily mean people should not change donations from organisations helping people in extreme poverty to ones helping farmed vertebrates? I do not think so. I encourage people to simply donate to the organisations they consider more cost-effective.
I do not think broad societal care about wild arthropods is necessary to greatly improve their welfare. A small research community may be enough to determine whether the most numerous animals have positive or negative lives, and then advocacy efforts can focus on increasing or decreasing the area with the highest density of animals, which is much easier. For example, people are broadly on board with decreasing deforestation, which greatly increases the number of wild arthropods, and therefore is hugely beneficial if they have positive lives.
Hi Vasco,
Thanks for that first point – I was using “naive utilitarianism” in a broader sense which I now realize made my point less clear. What I meant was that I worry about the type of thinking that allows for serious harm if it increases net welfare, e.g. disregarding rights violations so long as they lead to the greatest total good. I don’t disagree with your modelling, but worry more generally about reasoning which permits these other types of harm.
Your second point is fair and helps me understand your post better. Thanks for that!
I am not convinced of your third point. There are just so many insects in the world that I think it would be hard to improve their welfare on a large scale without some level of societal investment. However, until we have more research on insect welfare and related potential interventions, I think this one will be hard to resolve.
Thanks again for sparking this great discussion!
Thanks for the follow-up, Samuel!
I worry about that too in some sense. On the other hand, I would not consider “serious harm” donating to organisations working on invertebrate welfare instead of ones working to help farmed vertebrates or humans. In addition, I wonder whether donating to organisations working on invertebrate welfare is better than to ones working on chicken welfare reforms from rights-based perspectives too. I estimate broiler welfare and cage-free campaigns decrease 1.95 M and 452 k arthropod-years per $, thus violating the right to life of lots of animals.
Rosenberg (2023) “found that there are ≈1 × 10^19 (twofold uncertainty range) soil arthropods on Earth, ≈95% of which are soil mites and springtails”. So looking into a few representative species of mites and springtails might be enough to have a good picture of the welfare of wild terrestrial arthropods.
Thanks for the support, @Samuel Mazzarella 🔸!
Thanks for the interesting post, Vasco. It certainly gave me some food for thought on Monday morning! It raises several important issues for me; primarily, the issue with pitching one ethical issue against another.
Within chicken advocacy, we frequently encounter the argument that welfare improvements are not viable due to environmental sustainability concerns. I dispute this claim on the same grounds that I would challenge the idea that advocating for higher welfare chicken production could negatively impact arthropod welfare and thus be net harmful overall. The issue is not the higher welfare production itself, but rather the agricultural methods (or other potentially harmful practices) employed. For instance, slower-growing chicken breeds can thrive on diets with less soybean than conventional strains, which has been shown to reduce emissions from land-use change per kilogram of feed in certain higher welfare systems (as demonstrated by Mostert et al., 2022). This indicates that we can modify feed production methods, rather than abandoning welfare improvements, to achieve more comprehensive positive change. We can simultaneously advocate for better welfare standards and expect the implementation of environmental (or other, such as wildlife welfare) mitigation strategies. Higher welfare farming is not inherently incompatible with other social goals; rather, businesses must adapt and implement supplementary or alternative strategies alongside welfare enhancements to ensure this compatibility.
Your findings “suggest it is unclear whether chicken welfare reforms are beneficial or harmful” from a holistic perspective. However, regarding chicken welfare specifically, the benefits of these reforms are unequivocally clear—even when these animals live longer lives, as in the case of meat chickens (see Welfare Footprint’s findings). If these reforms inadvertently cause harm elsewhere, those harms should also be addressed as our understanding evolves, for example, through the use of alternative feeds, reductions in consumption, and decreases in land-use change. While we don’t yet (or may never) have perfect solutions, the significant suffering prevalent on factory farms and the measurable impact we can have in alleviating that suffering necessitate our continued advocacy for higher welfare farming practices. At the same time, issues related to environmental compatibility, wildlife impacts, and insect welfare should remain areas of ongoing research and mitigation.
Welfare interventions such as cage-free systems and the Better Chicken Commitment (BCC) aim to improve chicken welfare by directly enhancing the well-being of individual birds during production. A potential secondary benefit to welfare is a reduction in overall meat and egg consumption, for example if conventional chicken is seen as a reputational risk for corporations & they seek alternatives. The progress observed in the Dutch retail sector illustrates the potential for broader benefits, where meat reduction pledges have followed successful campaigns for cage-free eggs and improved broiler welfare.
Thanks for the good context, Mia!
Personally, I assume that it’s more likely that arthropods live net negative lives. They are mostly r-selected, so most of them die soon after birth, possibly painfully. So in terms of short-term impact on animal welfare, I see it as a tentative positive that welfare reforms likely decrease wild animal numbers. If I understand it correctly, you see it as a tentative negative. I’d be interested to know why.
On the other hand, I see it as a bad thing that vegan advocacy probably seriously increases wild animal numbers. But I’m unsure about how to weight this against environmental concerns. And I’m very unsure if wild animals’ lives are net negative overall, but I slightly lean towards a yes.
Thanks for the comment, Saulius.
I say in the post “my very uncertain best guess [is] that increasing cropland harms arthropods due to decreasing positive arthropod-years”. I estimate broilers in a conventional scenario, and hens in cage-free aviaries have slighly negative lives, and I feel like wild arthropods have a higher welfare as a fraction of the welfare range than those. However, I do not really know whether wild arthropods have positive or negative lives. It would be nice if the Welfare Footprint Institute (WFI) determined the time in pain and pleasure of for the most abundant species of terrestrial nematodes, mites, and sprintails, which are the most numerous terrestrial animals.
Are you also concerned about other interventions outside vegan advocacy which push for the replacement of animal-based with plant-based foods? These decrease cropland, thus tending to increase arthropod-years, which you believe is bad.
Yes, the same argument applies for other types of reduction of animal products, especially beef. Chickens tend to use the much less cropland per calorie, reformed or not. I’m not so much concerned, as I’m resigned about figuring out whether decreasing meat consumption is good or bad. It’s almost surely good for farmed animals, I’d give say 55% that it’s bad for wild animals. But then there is also impact on the environment (like global warming) which could also be a factor for x-risks and stuff. But I’m not even that sure that some x-risks are bad from a utilitarian POV. Also vegan advocacy might also increase moral circle expansion. But even that could be bad. For example, if people care more about animals, maybe they will care more about preserving natural habitats, which might contain a lot of suffering. There are so many factors that go into all kinds of directions. We’re clueless.
For me, chicken welfare reforms look like an unusually good bet in this uncertain world. They help big farmed animals, reduce the populations of small wild animals, and maybe increase moral circle expansion a bit. All of these seem likely good. They do harm the environment, but it’s a relatively small effect, and I think it can be outweighed by donating a little to some environmental charity. So to me, chicken welfare reforms look good from many different worldviews.
Charities that help invertebrates that you mentioned seem very good as well from many perspectives. But we are clueless about their long-term effects too.
WFI looks at farmed animals that are farmed in a consistent way and in places where we can easily observe lives of individuals from beginning to the end. This sounds like a very different and a much much much more complex project.
And even if we got precise WFI estimates for all species, we still might disagree about whether increasing wild animal populations is good or bad because disagreements about how to weigh:
Suffering vs happiness
Short and intense suffering vs long-lasting milder suffering
Welfare of different species
I think it’s difficult to improve on the handwavy argument that maybe wild animals suffer more, so we are better off if there are fewer of them. I think that people who care about small invertebrates are probably better off supporting invertebrate charities that you mentioned than funding such complex research project, which might not end up changing the behaviour of that many people (unless it changes Open Philanthropy’s grantmaking).
Btw, I think it’s unlikely that nematodes are sentient because they are so simple. The most commonly studied one has like 300 neurons. But I see they are excluded from your estimate anyway because they are not arthropods.
Thanks for the additional thoughts, Saulius!
Persuading people to improve the lives of wild arthropods is hard, but the welfare of arthropods can also be increased by increasing/decreasing forest area if they have positive/negative lives, and there many interventions causing this which are more broadly appealing. For example, one could double down on nature conservation, and promote terraforming if one thought wild arthropods to have positive lives.
Setting aside arthropods for a moment, I worry about the idea that switching to slower-growing breeds of broilers could lead to an increase in the total number of broiler-years endured in factory farm conditions (just because they might live for 70 days each rather than 40).
You’re then right to then raise an additional concern about land use change and its impact on insects in the scenario where meat production is held fixed but slower-growing breeds are used.
I tend to think of it like this: broiler and cage-free campaigns are not really aiming to change methods of production while holding fixed the amounts produced. They’re aiming for a reduction in the size of the industry, at least compared with what it would have been otherwise, due to growing consumer concern about the way the cheapest products on the market are produced.
This way of thinking about it reduces your particular concern while raising others. It now assumes lives in factory farm conditions are not worth living overall, and it assumes these campaigns are cost-effective ways of counterfactually reducing meat consumption. I hope they are, at least in some countries, but it’s hard to tell.
On the welfare effects of slower growing breeds, I think suffering is reduced overall despite the increase in life expectancies, based on Welfare Footprint Institute’s analysis:
Thanks for the comment, Jonathan!
I estimate broilers in a reformed scenario grow 79.5 % as fast, which makes the cost-effectiveness of broiler welfare reforms 79.5 % of what it would be if they did not affect broilers’ growth rate, and resulted in the same increase in welfare per broiler-year. The implied increase in the number of broilers of 25.8 % (= 1⁄0.795 − 1) is much smaller than my estimate for the increase in the welfare per broiler-year of 92.9 %. So I think there is very little risk of broiler welfare reforms being harmful to chickens. My estimate for the increase in the welfare per broiler-year is also close to 100 %, which illustrates my assumptions imply broilers in a reformed scenario have lives which are close to neutral.
I believe the vast majority of the benefits to chickens of chicken welfare reforms come from improving the conditions of chickens, not from decreasing their consumption. I estimated in the post cage-free reforms decrease the consumption of eggs by “5.40 %”, which is much less than my estimate for the increase in the welfare per hen-year of 80.4 %. I also think chickens in improved conditions have lives which are close to neutral, such that decreasing the population of chickens which have undergone a 2nd wave of welfare reforms may be harmful to chickens (although I think the overall effects would be unclear due to uncertain dominant effects on wild arthropods).
This post demonstrates the reductio ad absurdum of these sorts of calculations.
“Caged chickens might be better than free range because it means there are more insects living in the soil” is the sort of conclusion that most people can discard as being too absurd to consider, but if you’re earnestly attempting a utility calculus-based approach like this you are forced to conclude that it is a serious option.
This is paralysing. You could use the same logic here to lead yourself to conclusions like:
we should prevent economic development in the poorest parts of the world because they might eat more chickens or reduce the numbers of insects in the soil
we should kill whales to prevent the harm they do to krill
the primary moral concern of Earth is the rights of demodex mites
I’d encourage you to stop making these sorts of posts. I think they’re off-putting for people that might otherwise engage more with more reasonable EA ideas. Expressing uncomfortable truths is important when it’s useful, but these calculations are so riddled with uncertainty and so lacking in actionable conclusions that this post and posts like it are probably net harmful.
I strong downvoted this comment because I think this type of discourse censorship is terrible. Effective Altruism should be about figuring out how to do the most good, and then doing just that.
“This idea is off putting” can be use as a fully general counterargument against any new intervention or pivot. Helping farmed animals is off putting to many. Helping people abroad before helping those at home is off putting to many.
This is, by the way, not to say that you can’t dismiss an argument if the logic lead to absurd conclusions. Reasoning from first principals can be a dangerous activity if you take your ideas seriously (see e.g., epistemic learned helplessness, memedic immune system). But when trying to figure out how to do the most good, I think it’s really really bad to have any sort internal thought censors.
(I think it’s comparably better to consider “does this sound off putting” deciding what actions to take.)
What should be done about the possibility that insects or anthropods are conscious and affected by our interventions in your view?
EDITED to add: Just reviving the idea that its ok to favour humans over animals to a very high degree won’t help here, since it’s animal versus animal interests we are dealing with.
I don’t think this is a point against valuing animal lives (to some extent) as much as it’s a point against utilitarianism. Which I agree with. I didn’t downvote because I don’t think a detailed calculation in itself is harmful, but when you reach these kinds of conclusions is probably the point to acknowledge pure utilitarianism might be a doomed idea.
I have some sympathy with that view, except that I think this is a problem for a much wider class of views than utiliarianism itself. The problem doesn’t (entirely) go away if you modify utilitarianism in various attractive ways like “don’t violate rights”, or “your allowed/obligated to favour friends and family to some degree” or “doing the best thing is just good, not obligatory. The underlying issue is that it seems silly to ever think you can do more good by helping insects than more normal beneficiaries, or that you can do more good in a galaxy-brained indirect way than directly, but there are reasonably strong theoretical arguments that those claims are either true, or at least could be true for all we know. That is an issue for any moral theory that says we can rank outcomes by desirability, regardless of how they think the desirability of various outcomes factors into determining what the morally correct action is. And any sane theory, in my view, thinks that how good/bad the consequences of an action are is relevant to whether you should do it, whether or not other things are also relevant to whether the action should be performed.
Of course it is open to the non-consequentialist to say that goodness of consequences are sometimes relevant, but never with insects. But that seems a like cheating to me unless they can explain why.
I agree with David’s comment. These sorts of ethical dilemmas are puzzles for everyone, not just for utilitarianism.
And in the case of insect welfare, rights-based theories produce more puzzling puzzles because it’s unclear how to reckon with tradeoffs.
Thanks, Michael. I agree. For example, I estimate broiler welfare and cage-free campaigns decrease 1.95 M and 452 k arthropod-years per $, thus greatly increasing deaths nearterm, which I assume can easily be harmful from a perspective concerned about the right to life.
Hi Guy,
I guess some impartiality, “valuing animal lives (to some extent)”, is the key element of utilitarianism which is necessary for the conclusions of my post to hold. Which moral theory having impartiality as an element would imply different conclusions?
You make a great point about the parallel to the meat-eater problem, and I agree that, for similar reasons, it’s probably still a good idea to advocate for chicken welfare reforms.
However, I don’t think reductio ad absurdum is a compelling argument in this case.
This post’s argument seems absurd not because it leads to some kind of internal contradiction, but rather because it argues for something that’s way outside the things people normally think are good ideas. I don’t think “seems absurd to most people” is a reliable indicator of “is not ethically sound,” because I think many people believe things and act in ways that are not morally sound (e.g., factory farming). What I love about EA is that it’s a social space that encourages questioning of conventional ideas.
Thanks for the comment, Henry.
I think the effects on arthropods of these interventions may well be the driver of their overall effects, but I have little idea about whether they benefit or harm arthropods. I believe people caring about arthropods would do better supporting the Arthropoda Foundation, RP’s work on invertebrate welfare, the Shrimp Welfare Project, or Wild Animal Initiative.
I worry adocating for animal rights may be harmful due to encouraging wilderness preservation, because wild arthropods could have negative lives. In any case, I believe animal welfare interventions focussing on improving the conditions of animals instead of giving them rights have been more cost-effective at increasing the welfare of the targeted animals.
I had some actionable conclusions? I recommended donating and making grants to improve invertebrate welfare, and encouraged organisations helping vertebrates to consider how they can help invertebrates.
While I didn’t karma-vote on the main post, I downvoted this comment because I think the idea of net-negative lives for naturally occurring creatures is not only false but even harmful.
Thanks, Guy. I upvoted your comment.
I estimated 6.37 % of people have negative lives in the sense of preferring to not have been born, and I guess there is a higher fraction of wild arthropods with negative lives. What do you think is the probability of a random arthropod having a negative life? If it was around 50 % like mine, would you still consider harmful discussing the possibility of wild arthropods having negative lives?
I don’t think there’s such a thing as a negative life.
Why do you believe that?
I assume you’re looking for a rational explanation, but it’s rather based on personal experience. It’s because I think my life with constant chronic pain has more negative experiences than positive ones but I have decided I should keep on living.
I am probably just stating the obvious, but I encourage you to chat with LLMs to brainstorm solutions for your chronic pain (the more context you give, the better). You can also use Elicit to find good studies (you can filter in randomised controlled trials (RCTs), meta-analyses, and systematic reviews). I am happy to have a look if you are open to sharing more (privately or not).
A negative life is typically defined has one with more suffering than happiness, regardless of whether the person living it wants to keep living or not. Wanting to end one’s life is a much stricter condition than having a negative life. I estimated 6.37 % of people have negative lives in the sense of preferring to not have been born, but only 0.0088 % of people commited suicide in 2021. I assume people who ended their lives had negative lives when they ended, so I estimate only 0.138 % (= 8.8*10^-5/0.0637) of people with negative lives in a given year end their lives that year.
The above suggests the vast majority of people with negative lives do not end their lives. I think this is because commiting suicide is hard, people hope their lives may become positive in the future, and believe that ending their lives would harm others.
In principle, or only in practice?
In principle—though I can’t say I’ve been consistent about it. I’ve supported ending our family dog’s misery when she was diagnosed with pretty bad cancer, and I still stand behind that decision. On the other hand I don’t think I would ever apply this to an animal one has had no interaction with.
On a meta level, and I’m adding this because it’s relevant to your other comment: I think it’s fine to live with such contradictions. Given our brain architecture, I don’t expect human morality to be translatable to a short and clear set of rules.
Thanks for sharing, Guy!
Would you be against painlessly enthanising a stray dog with a similar condition as your family’s dog? If you would support ending their misery too, why not supporting efforts to decrease the number of wild animals with negative lives?
If I somehow ran into such a dog and decided the effort to take them to an ultrasound etc. was worth it, then probably yes—but I wouldn’t start e.g. actively searching for stray dogs with cancer in order to do that.
Makes sense. I think that suggests you consider decreasing the number of negative lives good in principle, although not always worth it in practice.
No, that’s not what I think. I think it’s rather dangerous and probably morally bad to seek out “negative lives” in order to stop them. And I think we should not be interfering with nature in ways we do not really understand. The whole idea of wild animal welfare seems to me not only unsupported morally but also absurd and probably a bad thing in practice.
Respectfully Vasco, I believe you are gravely mistaken in this series of posts about insects. You start off with assumptions that all your castles in the air are built upon, assumptions that have no verified foundation: we have no idea how insects suffer in nature, not even how do they experience reality. It is possible that their lives are generally difficult, but it’s also possible that they thrive without suffering, or even that they do not suffer at all in the way we understand it.
We have a responsibility to limit factory farming and its worst practices such as cages, and your obsessively precautionary stance regarding the hypothetical suffering of insects in the wild creates a dangerous inertia against the liberation of animals from exploitative practices that are the source of their worst and longest-lasting suffering, which must remain an ideal.
Moreover, even if insects were 100% sentient and your conclusions were valid, I believe you are mistaken even in the utilitarian application. In that case, why not concrete over all natural areas (as long as we survive) if the goal is to minimize a hypothetically painful existence for insects?
I want to reiterate that all these calculations are so speculative that the slightest change in assumptions could flip the outcome entirely in any direction. I strongly disagree with this approach and believe it to be extremely ineffective, as it is built on so little.
Thanks for the comment, Alexandre.
I agree there is lots of uncertainty. I tried to acknowledge this in the post:
Do you have any suggestions for improvement?
Have you had the chance to check wildanimalsuffering.org? In any case, the conclusions of my post do not depend on wild arthropods having positive/negative lives:
Decreasing arthropod-years would be beneficial/harmful if arthropods had negative/positive lives. However, I am super uncertain about whether their lives are positive or negative, so I do not recommend interventions which assume they are positive/negative. I favour supporting organisations like the ones I mentioned in the summary which decrease the uncertainty about the welfare of arthropods, and improve their lives, which is good regardless of whether their lives are positive or negative.