I work as an engineer, donate 10% of my income, and occasionally enjoy doing independent research. Iâm most interested in farmed animal welfare and the nitty-gritty details of global health and development work. In 2022, I was a co-winner of the GiveWell Change Our Mind Contest.
MHRđ¸
Karma: 1,975
RP is currently #1. EA Animal Welfare Fund is currently #2, and I donât think it the kinds of work it funds are necessarily things OP wonât fund.
I think this is only partially true. Since RP gets significant funding from OP, my understanding based on their communications is that they tend to often use unrestricted funding specifically in areas that canât get funding for from OP. And similarly, AWF has specifically highlighted funding areas that OP wonât as one of their top areas.
Thank you Sjir and Aidan for this excellent work! I think itâs quite valuable for community epistemics to have someone doing this kind of high-quality meta-evaluation. And it seems like your dialogue with ACE has been very productive.
Selfishly as somone who makes a number of donations in the animal welfare space, Iâm also excited by the possibility of some more top animal charities becoming supported programs on the donation platform :)
Thanks Vasco. After thinking about the numbers myself, I agree that allowing for states worse than death canât on its own do a lot to make the numbers comparable between GiveWell and SWP. I do actually think it would move the numbers more than youâre accounting for there, both because the deaths prevented by GiveWell top charities might involve more than 7.5 minutes of excruciating pain and because GiveWell top charities prevent a lot of morbidity among people who end up surviving (and I think theyâre significantly underweighting the value of this, e.g. clean water interventions prevent about 6 person-years of being sick with waterbone illnesses for everyone person who dies, and I would significantly prefer to be in a dreamless sleep than be conscious with a severe enteric infection.[1] But the DALY weight for severe diarrheal illness is 0.247, implying 3/â4ths the wellbeing[2] of being fully healthy). But this is at most going to change the cost-effectiveness of GiveWell top charities by a factor of 2, not 4 OOMs.
As for the 10000x difference in weights between disabling and excruciating pain, I have to admit Iâm pretty confused here. On the one hand, it strikes me as fundamentally implausible that suffocating is 10000x worse than dying of heatstroke. On the other hand, some of my intuitions do lean towards not being willing to endure e.g. burning to death for almost anything else. Iâll need to spend some time reviewing the literature before I try and make further sense of how to best make these tradeoffs.
Thanks again for all your work and engagement here, I think itâs genuinely quite valuable to be having these conversations!
I also ranked Arthropoda first! Iâm quite bullish on the value of information in the animal welfare space, and think that on the current margin they would do extremely valuable work.
Though Iâve made some comments that disagree with some of Vascoâs specific numbers, I agree that SWP is extraordinarily effective and ranked them a close second after Arthropoda!
(Crossposted from twitter)
While Iâm a big fan of SWP and have donated to them myself, I am skeptical of claims likeThis makes [SWP] around 30 times better at reducing suffering and promoting well-being than the highly effective animal charities focused on chicken welfare which themselves are hundreds or thousands of times more effective than the best charities helping humans.
I greatly appreciate @Vasco Grilođ¸ for writing up his analysis, but I donât think that most people would agree with some of the assumptions made in it regarding pain intensity:
For air asphyxiation: time in disabling pain equal to the maximum time during which shrimp can remain alive of 30 min, although Aaron noted he and his colleagues have seen some alive for 6 h; time in excruciating pain as a fraction of that in disabling pain equal to that of ice slurry (0.126 h); time in hurtful pain as a fraction of that in disabling pain equal to that of ice slurry (0.00633 h); and time in annoying pain as a fraction of that in hurtful pain equal to that of ice slurry (0 h).
[...]Annoying pain is 10 % as intense as fully healthy life.
Hurtful pain is as intense as fully healthy life.
Disabling pain is 10 times as intense as fully healthy life.
Excruciating pain is 100 k times as intense as fully healthy life.
RPâs median welfare range of shrimps of 0.031.
My assumptions for the pain intensities imply each of the following individually neutralise 1 day of fully healthy life:
10 days (= 1â0.1) of annoying pain.
1 day of hurtful pain.
2.40 h (= 24â10) of disabling pain.
0.864 s (= 24*60^2/â(100*10^3)) of excruciating pain.
Vasco estimates that asphyxiating shrimp experience about 7.5 minutes of excruciating pain, and weights this as 10000x worse than disabling pain, which is the maximum pain experienced by a chicken during a keel bone fracture or death from heat exhaustion (in the data used to generate the THL numbers). Moreover, the data he relies on for the cost effectiveness of GiveWell top charities does not allow for the existence of states worse than death. This means that heâs estimating that the pain experienced during asphyxiation is 100000x the worst pain prevented by GiveWell. This seems highly implausible to me. Surely dying of malaria or diarrheal disease involves some pain that is within 100000x the intensity of suffocation (and indeed WFP estimates that sepsis in a chicken involves excruciating pain, so I would expect that sepsis in a human does as well).
None of this is to say that SWP is ineffective, merely that the cost-effectiveness ratios compared to other EA top charities citied here seem overly high to me.
Congrats, Sjir!
Thanks for looking at this Vasco, itâs always great to see others doing this kind of cost-effectiveness analysis.
Your results indicate a substantially higher direct cost-effectiveness for SWP relative to the analysis I did last year. From looking at your methodology, I believe our primary difference comes from a difference in weighting the relative badness of different levels of pain. I used the same numbers as a 2023 RP report which weighted excruciating pain as 33 times worse than hurtful pain, while your weights put excruciating pain at 100000x worse than hurtful pain.
Iâve updated towards thinking 33x is probably at least an order of magnitude too low (and more recent RP reports have used weights in the vicinity of 600x), but I would personally be skeptical of 100000x.
Of course much of SWPâs impact is through creating systemic change, so I donât want to over-emphasize the importance of these direct impact CEAs, as valuable as they are.
Great list (and thanks for the shoutout)!
I would add @Laura Duffyâs How Can Risk Aversion Affect Your Cause Prioritization? post
I believe the values come from the 10th anniversary edition of the TLYCS book. They should be in the FAQ on the website and Iâm surprised theyâre not.
I think mainstream longtermist EA is already on a path to try and help create the hedonium shockwave if and only if itâs the right thing to do. The âonly ifâ part seems really importantâturning 99.999% of the accessible universe into hedonium seems like a quite bad idea unless youâre extraordinarily confident in your ethical views. But it does seem like one theoretically possible outcome of the type of long reflection MacAskill advocates in WWOTF:
As an ideal, we could aim for what we can call the long reflection: a stable state of the world in which we are safe from calamity and we can reflect on and debate the nature of the good life, working out what the most flourishing society would be. I call this the âlongâ reflection not because of how long this period would last but because of how long it would be worth spending on it. Itâs worth spending five minutes to decide where to spend two hours at dinner; itâs worth spending months to choose a profession for the rest of oneâs life. But civilisation might last millions, billions, or even trillions of years. It would therefore be worth spending many centuries to ensure that weâve really figured things out before we take irreversible actions like locking in values or spreading across the stars.
Itâs really not clear to me that thereâs a better path to the hedonium shockwave than what longtermsists are already doingâtrying to ensure humanity survives and prospers and makes it to a place where we have more hope of reaching a consensus about whether or not itâs the right course of action. Of course, if the shockwave really is the right thing to do, waiting to start it would lead to a great deal of astronomical waste. But this is a small price to pay relative to the risks of destroying ourselves or causing great harm if our moral views are wrong.
Yeah this is a really good point, I have no idea how to square the numbers with big grants from OP to THL
Thanks Vasco!
I checked the pages for each charity to get the scores.
I agree that AWF doesnât directly evaluate cost-effectiveness, but I still think thereâs a good chance theyâre likely to be the EV maximizing option over THL. THL estimates that it costs them $2.63 to move a hen from a conventional to a cage-free system, or about 0.57 yr/â$ given a 1.5-year lifespan. Last year, Emily Oehlsen from Open Phil said âWe think that the marginal [farmed animal welfare] funding opportunity is ~1/â5th as cost-effective as the average from Sauliusâ analysis.â Sauliusâs 2019 analysis estimated that corporate campaigns pre-2019 impacted 41 chicken-years per dollar, so at a 5x reduction thatâs 8.2 yr/â$. I donât want to take Emilyâs numbers too literally, but that implies a >10x gap between the cost effectiveness values of OPâs marginal funding opportunity and THL. Since Iâd expect AWFâs opportunities to look somewhat similar to OPâs, that leads me to guess that theyâre likely to be on net more cost-effective than THL. This directionally agrees with some of the comments by insiders such as @James Ăzden on the GWWC evaluations thread as well. But Iâd be very curious to hear more from folks who are more plugged in, this is just an outsiderâs guess.
For what itâs worth, I do actually give to both AWF and THL, but give much more to AWF.
Good information!
A couple things to add to this very good comment:
In general, the landscape of charity evaluation for animal charities is less mature and quite a bit more uncertain than the landscape for global health and development charities. Any cost-effectiveness estimates are going to be coarse and debatable.
ACE has a partially qualitative cost-effectiveness scoring system. Their ratings (higher = better) for their recommended charities are:
Faunalytics: 5.7
New Roots Institute: 4.9
The Humane League: 4.7
Wild Animal Initiative: 4.5
Ăiftlik HayvanlarÄąnÄą Koruma DerneÄi: 4.3
Shrimp Welfare Project: 4.3
Fish Welfare Initiative: 4.3
Sinergia Animal: 4.1
Good Food Institute: 3.8
Dansk Vegetarisk Forening: 3.7
Legal Impact for Chickens: 3.7
@Laura Duffy wrote a report at Rethink Priorities in which she estimated that corporate hen welfare campaigns avert 1.13 DALYs/â$ and shrimp stunning interventions avert 0.038 DAYLYs/â$. Both of these estimates were quite uncertain and depended on a lot of debatable assumptions (including possibly underrating the potential for shrimp stunning interventions to catalyze industry-wide changes), but I think this is one of the best estimates currently out there.
My personal advice would be that I think the EA Funds Animal Welfare Fund is probably the expected value maximizing option, while The Humane League is probably the best option if youâre somewhat risk-averse.
Some data from open phil and EA funds grants:
Shrimp Welfare Project :
Open Philanthropy (September 2022): $300,000
Open Philanthropy (November 2023): $2,000,000
EA Funds (Q1 2022): $45,000
EA Funds (Q1 2022): $45,000
EA Funds (Q4 2022): $210,000
EA Funds (Q3 2023): $130,000
Insect Institute:
None listed
Arthropoda Foundation:
None listed
Aquatic Life Institute:
Open Philanthropy (June 2022): $100,000
Open Philanthropy (November 2022): $150,000
Open Philanthropy (December 2023): $550,000
EA Funds (Q1 2022): $80,000
EA Funds (Q1 2022): $80,000
EA Funds (Q3 2022): $80,000
EA Funds (Q4 2022): $200,000
Crustacean Compassion:
Open Philanthropy (February 2021): $786,830
Open Philanthropy (January 2023): $863,595
EA Funds (Q3 2021): $92,000
EA Funds (Q3 2022): $59,000
EA Funds (Q3 2023): $2171
EA Funds (Q4 2023): $84,000
IMO if EA funds isnât representative of EA, Iâm not sure what is. I think the different funds do a good job of accurately representing the broad diversity of viewpoints and approaches within the community, and I would personally be very sad if EA funds dropped the EA branding.
Agreed that this seems surprisingly under-discussed
Thanks so much for your response, that all makes sense!
Youâre understanding question 3 correctlyâGiveWellâs moral weights look like the following, which is fairly different from valuing every year of life equally.
Thanks for sharing your perspective and experiences here! I think itâs really valuable for EAs with first-hand experience to write about these issues, and Iâm really sorry you went through such a difficult time. You might be interested to read this piece I wrote last year about EA and disability based on my own experiences. It includes some discussion of HALYs, though thatâs focused more on the history and perception of HALYs rather than the issues you touched on.
Reading your piece, I very much agree with you that the current methods of constructing HALY weights generally have methodological issues and would greatly benefit from more focus on actual experiences of people in the health states in question. I also agree that the naive application of HALYs as units of utility or âgoodnessâ can lead to some very dark places (especially in light of the methodological concerns you mentioned, it frustrates me that EAs often slip into using DALYs as units of utility when the post-2010 weights are specifically intended to only be a measure of health status).
I have a slightly different perspective on a couple of the other issues you mention, and Iâd be curious to talk more with you about these.
First, I think that the existence of difficult tradeoffs is inevitable as long as resources are limited, and moving to HALY alternatives wonât necessarily eliminate these challenges. For example, one of the articles you quote mentions NICE guidelines about drugs for dementia. I would really, really like better medications to exist, but my (admittedly not very in-depth) understanding of the field is that the drugs are quite expensive and donât do much to improve symptoms or alter the course of the disease. As long we have limited healthcare resources, itâs not clear that thereâs an alternate weighting strategy that would justifiably recommend allocating limited resources to buying these medications rather than spending on other health interventions that work better. Here, it seems like the problem is much more the paucity of options and limited resources than the particular weighting scheme.
Another item is the role of HALYs in perpetuating healthcare inequalities. I do agree that there is absolutely a straightforward way in which this is true, but Iâve come to think thereâs some more complexity here than I at least initially thought. HALY maximization in some situations encourages improving the wellbeing of people with chronic illnesses/âdisabilities over extending the lives of able-bodied people. For example, the 2021 GBD disability weights give post-viral chronic fatigue syndromes a weight of 0.22. A policymaker trying solely to maximize DALYs averted would, if such a treatment were available for the same cost, choose to invest in curing five people of post-viral chronic fatigue over saving one fully healthy personâs life (if all the individuals were the same age). I absolutely agree that ME/âCFS is underfunded overall, and that there is probably a role for HALYs in that underfunding (in particular, the 2021 GBD disability weights only include values for a small number of recognized post-viral chronic fatigue syndromes, so policymakers may not be able justify investments in broader ME/âCFS research/âtreatment in terms of HALY benefits). I just think the overall picture here is a bit complicated.
The last item is the existence of states worse than death. I very much agree that deciding at a population level that certain peopleâs lives are worse than death, then making policy on that basis, can lead to very dark and wrong places. However, I really do think that some people in some cases experience states worse than death, both from my own experiences and from the testimonies of others. In my own life, I have had experiences that were bad enough that I absolutely would have traded off shortening my own life to avoid them (for example, I would have been happy to lose a week of healthy life to avoid the worst moments of a shoulder dislocation). More broadly, I do think we should listen to ill or disabled individuals such as Gloria Taylor whoâve described their own conditions as worse than death and advocated for the right to access medical assistance in dying. I think it would be wrong to say that for individuals who describe their lived experience as worse than death and express a desire to access medical assistance in dying, they have not benefited by being able to fulfill their desire. And moreover, not having a weighting scheme that allows for states worse than death I think risks underemphasizing the suffering involved for certain people in cases of extreme pain. As a result, I worry that such a scheme could lead to underprioritizing interventions that improve quality of life and alleviate pain in favor of interventions that save lives. Again, I think itâs absolutely wrong to decide based on population-level statistics that an individual personâs life is worse for them than death, but I think thereâs a balance to walk here and I worry that itâs as bad or worse to not listen when people describe their lived experiences as worse than death based on their own values.
Thanks again for writing your piece. I hope these thoughts are useful!