1806Coquitlam, BC, CanadaJoined Jun 2018


Dr. Neil Dullaghan is a senior researcher at Rethink Priorities. Rethink Priorities is a research organization that conducts critical research to inform policymakers and major foundations about how to best help people and nonhuman animals in both the present and the long-term future. Neil currently works in the farmed animal welfare team, with an expertise in European Union policy.

He holds a PhD in Political & Social Science from the European University Institute, an MPhil in European Politics & Society from the University of Oxford and a BA in International Relations from Dublin City University.

He has volunteered for Charity Entrepreneurship & Animal Charity Evaluators. Before joining Rethink Priorities, he was a political data manager for WeVoteUSA while it participated in Fast Forward's accelerator for tech nonprofits, held numerous research assistant positions at the University of Oxford, and acted as Strategy Associate for a behavioural science think tank, The Decision Lab.


Topic Contributions

Thanks for this,
Just wanted to note a misframing of the slaughterhouse ban post.  You have written
 "found ~40% supported banning slaughterhouses or said ‘don’t know / no opinion’ to questions, highlighting a large discrepancy"-which I think is taken directly from the latest "EA & LW Forums Weekly Summary"  rather than the slaughterhouse ban post.
This makes it seem like 60% opposed and then 40% combined EITHER supported or had no opinion, when in fact the 2017 Sentience Institute result was 43% supported, 11% chose don't know,  46% opposed.

I realise this misunderstanding comes from my phrasing in the summary
"(~39-43% support when including those who chose no opinion/don't know)" 

I added the "when including those who chose no opinion/don't know" clause because  Sentience Institute's 2017 summary only reports the percentages agreeing out of those who either agreed or disagreed (47% agree with the ban, 53% disagree). But since many respondents selected “Don’t know” regarding the bans on slaughterhouses (11%), the overall percentages supporting these ban is slightly lower than their headline summary: 43% rather than 47%. In their 2020 replication, the same issue appears again when SI report a headline result of "44.8% are in favor of banning slaughterhouses" but this excludes the "don't know", so the actual support is 39.5%.

Sorry for causing confusion. I have now edited the original post to avoid this so it just reads "(~39-43% support).

Also you only mention the results from survey 1, the survey experiment of N=700, and I think a fairer comparison to the Sentience Institute figure is from survey 2 of  15.7% (95% CI [13%-18.8%]) support because both of these use weighting to represent the US public's opinion and are of a larger sample size.

We should also note that Norwood (one of the authors who replicated SI’s original 2017 study) this year ran a new slaughterhouse ban survey experiment  ([Britton & Norwood 2022]( and found lower support. (I only just received the data from them so I couldn’t include it in the post).

Here is my summary from just skimming the article and quickly aggregating the data.

They test a hypothesis that the question ordering in the 2017 SI study cued respondents' ideal self (like whether voting is a moral virtue) rather than their common self (like whether they actually voted). Their theory is that by asking respondents first whether they agreed with statements about meat reduction, discomfort with the way animals are used in the food industry, and animal sentience it cued their ideal self so that “the desire to not appear hypocritical induced them to activate a mixture of their ideal and common self” when answering questions about bans on animal farming, factory farming, and slaughterhouses.

The actual design of their study is a little too complicated to explain here (involving four treatments that altered the order and wording of ideal and common self questions, some food-related and some non-food related, as well as inserting buffer questions), but basically some respondents saw the ban questions before the ideal self questions, and others saw them in the same order as in the original 2017 SI study. Furthermore, to build on their tests about whether respondents understood the implications of bans,  "roughly half of the subjects are given the [common self] statements exactly as they appeared on the Animal Sentience survey, while the other half contain an addition [. . .] For example, some see the statement “I support a ban on slaughterhouses” while others see the statement “I support a ban on slaughterhouses and will stop eating meat”. "

While the primary aim of their study was to test something they call “identity inertia” and they fail to find convincing evidence of it, their finding on the slaughterhouse ban issue was "once individuals are informed about the implications of actions like banning slaughterhouses, they are less eager to do so." 

Data were collected via an online survey through Qualtrics from August to October 2019 of a representative sample of nearly 2600 drawn from the U.S. population. A subset of the results (N=1528) show

  • 26.48% "agreed" with the statement "I support a ban on slaughterhouses" (36.5% disagreed, 37.02% no opinion) (N=759)
  • 18.21% "Agreed" with the statement "I support a ban on slaughterhouses and will stop eating meat." (58.13% disagreed, 23.67% no opinion (N=769)

(Though Norwood say they couldn’t confirm this was correct since they never went into the data to get raw numbers like that, and I couldn't see an easy way to break these results down according to whether respondents saw the ideal-self or common-self questions first- though that probably doesn't matter since Norwood didn't find a lot of evidence that it matters)


Thanks for reading and engaging with our work!

  • In 2019 we conducted some exploratory small-N, low-confidence studies on this topic that informed these high-quality, larger N studies. We feel comfortable presenting these recent results as we want to promote the norm of advocates choosing strategies and messages based on the best quality evidence, so would rather the community update based on the results of high-quality studies rather than low-confidence small-N studies where the wrong inferences may be drawn.
  • [Update 2022-Nov-18: I added a methods section to make it clearer for readers. Thanks Jacy for flagging this issue]. Respondents were recruited via Prolific and surveyed using Qualtrics and were weighted for representativeness using 5-year 2019 American community survey data and general social survey data. More details and data are available in the OSF project, but let us know if it's not clear or something is missing since the 2nd survey was part of a larger survey that is described here which we have not finished publication of yet. (minor update- the pre-registration for the experimental study was still, unintentionally, embargoed at the time of publication so that information wasn't easily accessible for readers, but it has been lifted and should be available here by Nov 12)
  • I don't think one should take a survey showing 16% support for banning slaughterhouses when animal welfare frames are used as reason for the marginal farmed animal advocate to update towards more radical approaches relative to their prior, unless that farmed advocate assumed there would be basically 0% support. In practice, I imagine many advocates have been relying on surveys showing much higher support as their benchmark so this survey would update them toward more pessimism. 
  • More generally, given the clear variation in support results from well-done surveys from different organizations on this specific item it doesn't seem advisable for advocates to update their beliefs strongly on how support for other specific or the whole class of radical policies will be. As we wrote, it would be better to test the specific policies in question with a variety of messages to get closer to understanding how the public might actually act if given the chance to vote. And of course compare these to more moderate asks that the movement thinks are also worthwhile in expectation. I don't think we should take attitudes towards a specific policy as very indicative of attitudes to all of animal agriculture or other anti-factory farming policies. I also don't think we should take attitudes towards all of animal agriculture as indicative of support for specific policy proposals advocated for under specific messages. It was the claim that support for banning slaughterhouses should encourage advocates to push stronger messages and policies that we wanted to raise more questions about. 
  • Thanks for the reminder about the 2021 AFT survey!- which if I'm correct also showed ~43% support for banning slaughterhouses, when including No Opinion. I'll add it to the report.

Thanks for engaging with the report. I'll offer a response since Tapinder's summer fellowship has ended and I was her manager during the project. I've made a general comment in response to Tristan that applies here too.

On your comment specifically, the "malthusian trap" is empirically not always supported. A population can approach or be at its carrying capacity and still have adequate resources, for instance if they simply do not reproduce as much due to less resource surplus.

Thanks for engaging with the report. I'll offer a response since Tapinder's summer fellowship has ended and I was her manager during the project.

Firstly, as a response to both you and Max.

Those are very fair concerns. I tend to think that in the very nascent WAW field tractability is such a big issue that focusing only on the most painful deaths of wild animals leaves us with few (if any?) tractable things to do. Until we gain greater knowledge of what to do, there is value in some WAW resources going towards trying out things that look more tractable. In the specific cases in this report, much of the value is coming from establishing the norm around acting on wild animal welfare in human-wildlife conflict, especially in an area where we can get buy-in from non-EA stakeholders at the ground level of a technology that might affect the larger sources of suffering that we do care more about.

On counterfactually more painful deaths, even if the counterfactual death is more painful, this of course needs to be weighed against the fact that the birds would have died later and so the birds get to enjoy more life. In terms of how birds fare as a result of not being killed by windmills, it is fair that the report could have more explicitly acknowledged uncertainty about this, but this uncertainty cuts both ways since the value of living longer is tied to one's assumptions about the suffering-v-pleasure in wild animal welfare which could be very optimistic or pessimistic.

For something similar, see David Manheim's 2021 list

Net worth & Charity Pledges

Elon Musk: $205 billion - Giving Pledge (GP), 50% 
Jeff Bezos:  $200 billion - No pledge (None) 
Bernard Arnault:$159 billion -  None 
Bill Gates: $151 billion - GP, ~100% 
Mark Zuckerberg: $136 billion - Non-specific, 99% 
Larry Page: $126 billion - None 
Sergey Brin: $121 billion - None 
Steve Ballmer: $107 billion - None 
Larry Ellison: $101 billion - GP, 95% 
Warren Buffett: $101 billion- GP, 99%

Thanks Jamie! We struggled a lot with this issue when writing the post.

I'm not really sure I see a problem or a difference with the "which megaproject ideas can we think of?"/ "how rapidly will we get diminishing returns on further investment in various plausibly cost-effective project ideas?" distinction. I think if the answer to the second question is "quickly and with only a few million $" then you cut the idea from the list. It's part of the way to arrive at answers to "which megaproject ideas can we think of?". Other ideas floated seemed like they would be cost-effective at a small scale but could never absorb $10M because the problem was so small (foie gras bans perhaps) or the low-hanging fruit was uniquely cheap (the first type of a new campaign in a new region/species but hit some blockers or severe diminishing returns as they try to scale), and other ideas didn't look cost-effective at a small scale only but maybe at large scale if they reach some sort of economies of scale (some sort of policy or subsidization schemes that only gain leverage at large scales).

On the specific example you highlighted, I think "almost any animal charity" would have more weight as a critique if there were many such opportunities. I think the N of animal charities pursuing interventions that could actually both scale & remain cost-effective is relatively small (I don't see orgs like FWI and Healthier Hens popping up without the deliberate effort of Charity Entrepreneurship and it's still to be proven if they can scale and remain cost-effective. Even larger orgs like CIWF & THL aren't obviously only doing cost-effective things). The two we cited (focusing on shrimp and farmed insects) were deliberate because the sheer number of animals affected provide the opportunity that cost-effectiveness could be maintained even if spending a lot of money, unlike other animal charities. 

I agree not all the items on the list will turn out to meet strict definitions, or even vague definitions, of megaprojects. The main point of the exercise here was to note the virtual lack of any ideas on animals and prompt discussion and interest, and secondly to actually propose ideas from among which further investigation might find some really compelling megaprojects.

Thanks for your thoughts on the different approaches Zane, do you have an estimate of when your proposed approach would bear fruit in terms of a scalable intervention?

As Sebastian noted, in-ovo is already being done in two countries with chick culling prohibition legislation, and the Metaculus forecasting community median estimates most eggs produced in the EU will be sexed before hatching by October 2025, and most eggs produced in the USA will be sexed before hatching by 2033 (but a major U.S. supermarket chain will sell “no-kill eggs” in at least 25 states by 2031 ).

How do you think this approach (make male chickens lay eggs if I understand correctly) compares to other solutions being proposed/rolled out:

Load More