I write about meat reduction research at https://regressiontothemeat.substack.com/ and here’s an overview of the lab I work at
Seth Ariel Green 🔸
Karma: 1,695
Ah, apologies, should have clicked your profile.
Yes, my/our lab’s work is funded by Coefficient and by the Strategic Anima Funding Circle, and my blog is funded through Coefficient’s Living Literature Review program.
Some other EA-flavored research we’ve produced (and sometimes reported here):
Research report: adding a plant-based option (or two) doesn’t change behavior
Research report: modern alternative proteins did not replace meat purchases in restaurants
Documentary films can increase public interest in plant-based diets in the USA
Adjacent to our work, Kenny Torella of Vox writes about a randomized evaluation of the documentary Dominion (Kramer & Landry 2026) & I talk to James Ozden about the limits of behavioral approaches on his podcast How I learned to Love Shrimp.
All of this, I think, is circling around questions of persuasion and talking to people about meat.
Finally, it’s simply true that eating animals entails paying for abuse so extreme that almost everyone is repulsed by it.
But as far as I’m aware, we have no one with an EA mindset focused on how we make these arguments most effectively
👋 I and many others are working on this! See, eg, Meaningfully reducing meat consumption is an unsolved problem: meta-analysis. We write such papers with the goal of figuring out what strategy works best for talking to people about factory farming.
P.s. Welcome to the forum Bruce, hope it’s a good experience.
I strongly disagree with the concern about reputationally damage. Vasco is not the spokesperson for EA. He should ask the questions he thinks interesting and important. Other people can engage or not as they see fit. The situation might be different if he occupied a structurally important role in the movement—head of CEA? -- but even then, I’d want him to pursue his version of sapere aude, daring to know. Rationalist epistemic norms are good and worth preserving.
I am reminded of something @Aidan Kankyoku wrote about worrying about infohazards in a different thread. IMHO, most of us should spend very little time thinking about the second order effects of what we write and just write.
Something corroborating from Gernmany:
https://www.nature.com/articles/s44458-026-00138-3
Decade-long shift toward vegan meal options as part of dietary transition in German university canteens
Here we show menu and price trends using 3.7 million meal options from 340 German university canteens between 2015 and 2024. Plant-based options (vegetarian, vegan) rose from 63.7% to 70.5% (6.9 observed percentage points; 2.2 predicted percentage points per year). Vegan options increased by 23.1 points overall and 3.7 per year, while vegetarian options declined by 16.3 and 2.1, and meat-based options by 6.0 and 1.7, respectively. Vegan growth coincided more strongly with declining vegetarian than meat-based options, suggesting intra-plant-based replacement rather than direct substitution of animal-based options. Animal-based meals remained more expensive, while prices rose in parallel, suggesting economic factors alone cannot explain the shift.
I...guess this is good? But it really depends on the specifics. As Kenny Torella notes: https://www.vox.com/future-perfect/23273338/germany-less-meat-plant-based-vegan-vegetarian-flexitarian
Germans are eating about the same amount of beef as they were in 2011, but far less pork. But because pigs are large — pigs yield about 124 pounds of edible meat on average in Germany — the steep decline in pork only resulted in a reduction of about one-sixth of a pig per person. However, the 12.5 percent increase in poultry consumption, which looks modest on the chart below, has resulted in almost one extra chicken farmed for each of Germany’s 83 million residents because chickens are so small.
I talk about this paper in light of comparing more modern meat alternatives (Beyond/Impossible/etc.) to traditional vegetarian mains (tofu/curry/etc.), and why the academic literature at least suggests more cause for optimism about the latter than the former, here:
https://regressiontothemeat.substack.com/p/the-more-successful-add-more-veggie
To Jacob’s point, a 1 pp. increase in the unemployment rate is associated with a 3.3% higher suicide mortality rate. A 6 pp. decline in employment would be hugely disruptive. (Caveat that decline in labor force participation != unemployment)
Yes, rich people eat more meat. A lot more. From OWID
Is this relationship causal? Probably. Obviously this is hard to establish but I’d bet that if we did have well-identified studies they’d validate that.
Is the increase in meat consumption per capita as incomes rise offset by declining birth rates? Probably not. The differences in meat consumption between rich and poor countries is large (can easily be a factor of 10 or more).
Does a $5000 donation to ACE offset the potential harms? This is a deep question. I suppose you could build a model to address it. I think that model would be extremely brittle and probably rest on weak evidentiary foundations. So I have no idea.
My thought: I would take the job that most increases your human capital. I am increasingly of the opinion that the future is going to be weird and we really, really need to be building skills. The big picture is too murky to think about in a productive way and things are going to be chaotic. We should focus on navigating the incoming chaos.
Hi there
See also this[20] more recent meta-analysis that came to a similar conclusion about alternative proteins and other meat reduction interventions.
👋 I am lead author of that paper!
Two studies we looked at addressed alt proteins directly
Bianchi 2022 Replacing meat with alternative plant-based products (RE-MAP): a randomized controlled trial of a multicomponent behavioral intervention to reduce meat consumption 10.1093/ajcn/nqab414 Acharya 2004 Nutritional changes among premenopausal women undertaking a soya based dietary intervention study in Hawaii 10.1111/j.1365-277X.2004.00537.x The second paper is a small sample and doesn’t really address what we consider the modern generation of alt proteins. The first does. From that paper’s abstract:
Methods
Adult volunteers who regularly consumed meat were recruited from the general public and randomized 1:1 to an intervention or control condition. The intervention comprised free meat substitutes* for 4 weeks, information about the benefits of eating less meat, success stories, and recipes. The control group received no intervention or advice on dietary change. The primary outcome was daily meat consumption after 4 weeks, assessed by a 7-day food diary, and repeated after 8 weeks as a secondary outcome...
Results
Between June 2018 and October 2019, 115 participants were randomized. The baseline meat consumption values were 134 g/d in the control group and 130 g/d in the intervention group. Relative to the control condition, the intervention reduced meat consumption at 4 weeks by 63 g/d (95% CI: 44–82; P < 0.0001; n = 114) and at 8 weeks by 39 g/d (95% CI: 16–62; P = 0.0009; n = 113), adjusting for sex and baseline consumption. The intervention significantly increased the consumption of meat substitutes without changing the intakes of other principal food groups.
* Participants selected the meat-free substitute foods from a catalogue containing the full range of products available in a major UK grocery store at the time of the trial. This included mycoprotein meat alternatives and vegetable- and pulse-based meat substitutes.
How much should one update on this study? IDK. But by its own lights, giving people alt proteins for free—embedded in a larger intervention—had meaningful effects.
However, when we test intro-ing alt proteins in a hypothetical, online ordering, chipotle-like environment, alas we don’t find much.
Hi David,
It’s true that we aren’t powered to detect very small changes on any test. We did a fair bit of background work to target 5 pp. as our SESOI based on our understanding of what we could plausibly take to restaurants as an argument that they should add a chicken-based PMA to their menus.
if One were to have introduced a more more popular set of plant-based products it remains possible that these indeed would have substituted for the animal products.
Yes, that is possible. Our paper does not address what some other, more appealing thing would have done. However, taste parity is an open problem. My personal belief—out of scope of the data—is that we can think of consumers as extrapolating, when they see an item like “chickn’itas,” that it’ll basically be the category average, or on par with the other PMAs currently available at Chipotle, and making a choice based on that belief. Unfortunately, PMAs account for a small fraction of all sales (3% per this out-of-date article but I couldn’t find anything more up to date and official. However, based on 1-on-1 conversations, I think this number is still approximately right). Chipotle has tried many times to introduce more plant-based options. If a more popular and appealing option existed at the kind of margins that Chipotle looks for, we can assume they would have already made it available.
Regarding the chicken-vs-everything else split: the difference-in-difference CI ([−12.58, 0.73]) is wide, and we say in the paper that we can’t resolve which meat the displacement came from, alas.
Hi David, my colleagues and I have a forthcoming project that will address the substitution question using a large, novel dataset. (UPDATE: that preprint is now summarized in Research report: modern alternative proteins did not replace meat purchases in restaurants.)
Regarding the extent to which PBAs displace demand for meat products or provide more options for vegetarians, some work led by my colleague Jessica Hope is relevant (summarized here).
I suspect other teams have forthcoming work on this as well. Very interesting subject!
👋 we’ve updated this post to link to the now-published paper:
Hope, J. E., Green, S. A., Peacock, J. R., & Mathur, M. B. (2026). Effects of adding plant-based menu options on meat selection frequency: A randomized controlled experiment. Food Quality and Preference, 105931.
I don’t have a singular directional intuition about this.
I’m not sure AGI isn’t already here.
There are some scenarios where AGI liberates us from constraints and others where it enables humans to extend their dominance over animals. Who can say?
In the meantime, AI does not absolve us of helping animals.
Something about this topic creates a semantic stop sign for people whose opinions I otherwise find interesting. So even if the subject interesting in the abstract, I’m afraid here and now it sometimes leads to worse discussions.
Hi James,
I’m of course excited to see more funding and attention here and I broadly agree with your theory of change.
Obvious counterfactual is to funnel that same donation into, say, one of these existing groups:
Better Food Foundation’s DefaultVeg Campus Program
Plant-Based Universities (UK-based)
[09/04/2026 update] MCURC’s Student Fellows program
What appeals about going your own way?
I have not read this paper—I will put it on my reading list—but for folks interested in the subject, I recommend Norwood and Lusks’s book Compassion, by the Pound: The Economics of Farm Animal Welfare, specifically chapter 5, which covers different rearing systems. They argue, and I agree, that cage-free systems are better overall for chicken welfare but note that many farmers and specialists feel that caged systems are actually better for overall flock health. For instance, mortality is higher in cage-free systems due to diseases, pecking etc. (Here is some contrary evidence.)
I think it’s ok in most cases to ask someone to be nice on the internet. To quote Juana Molina, “No seas antipática. No seas antipática con tu mamá. La la la.” Words to live by.
I seriously doubt this is going to be the example they focus on when they send the trains
The first person I lightly flambéed also said “sorry I was used to less nice parts of the internet” so I guess that went down well
When someone says “be nice” and it gets more upvotes than the post itself that provides a more meaningful signal of how a post went astray than just downvoting the original post
So I am not persuaded and will keep doing my chastise-y hall monitor thing 😤😤😤
Dean Ball’s commentary on this refamed the issue for me https://www.hyperdimensional.co/p/clawed
The big difference, however, is that Anthropic is essentially using the contractual vehicle to impose what feel less like technical constraints and more like policy constraints on the military. Think of the difference between “this fighter jet is not certified for flight above such-and-such an altitude, and if you fly above that altitude, you’ve breached your warranty,” and “you may not fly this jet above such-and-such an altitude”). It is probably the case that the military should not agree to terms like this, and private firms should not try to set them.
But the Biden Administration did agree to those terms, and so did the Trump Administration, until it changed its mind. That alone should make one thing clear: terms like this are not some ridiculous violation of the norms of defense contracting...
The contract was not illegal, just perhaps unwise, and even that probably only in retrospect. Note that this is true even if you agree with the underlying substance of the limitations. You can support restrictions on mass domestic surveillance and lethal autonomous weapons, but disagree that a defense contract is the optimal vehicle to achieve that policy outcome. The way you achieve new policy outcomes, under the usual rules of our republic, is to pass a law...
I agree that there’s something iffy/non-democratic in theory about putting that kind of constraint around the Pentagon, and that it would have been prudent for them to decline it in the first place. An analogy I read on Substack: if an epidural manufacturer told a government hospital “you’re welcome to use our drug so long as you don’t use it in any abortions,” it would probably be prudent to decline that contract (too much overhead).
Anyway this reframing put one sentence in particular by Dario into a new light: “To the extent that such surveillance is currently legal, this is only because the law has not yet caught up with the rapidly growing capabilities of AI.” In other words, because we know what the law should be and what it’s probably going to be, we should implement that policy today. I think many of us can think of examples where we’d be uncomfortable with a billionaire tech CEO saying that.
I am glad that you think this issue is tractable and I’ve been following your work with great interest since I saw your RECAP talk in July (Side note to anyone following this—the RECAP talks are great!). I am not sure what my own threshold for “tractable” is but I appreciate that you are cracking on it and I would be glad to be proven wrong. Tractability is inherently based on unknowns and I’m glad we’re a big tent where people can prove something possible by doing it.
Thank you, I’ve now hyperlinked to the piece in both of my prior comments where I use the word “norms” (in part so that I remember it next time 😃)
Well, you know me, always playing nice 😃😃😃
I do want to say, in Andrew’s defense, that the comments on @Alistair Stewart ’s original post are not exactly the very model of civility that EA might hope to show the world. I can understand why you’d read them and come away with the sense that people don’t really get what you’re trying to do.
However, the point that calling something a ‘leading cause area’ requires cross-cause comparison is well-taken.
I agree with both points here. A few general thoughts, echoing something I wrote in 2022.
Very hard for an outsider to assess whether someone is likely to be good at a job. I generally don’t like to second guess other people’s hiring decisions.
There are extreme cases where I’d second guess anyone. SBF is obviously a psychopath and hiring him would be insane. Likewise, there are cases when you don’t hire the (narrowly conceived) ‘best’ person for the job. I would never hire Diddy to promote a party or produce a track, for instance.
I don’t think Caroline is such a case. I view her as a cautionary tale about getting sucked into the orbit of a charismatic psychopath when you’re a young person. I’ll never know what I would have done in her shoes. For what it’s worth, I once applied to be SBF’s friend in the Bahamas because it was advertised on the EA forum.
If Caroline were hired at CG or CEA or a few other orgs, I’d be worried about reputational harms. Manifund is much more marginal than that. I think reputational harms will be commensurately marginal.
One version of this point is to note that EA’s critics are looking for anything to pounce on because we’re a bunch of weirdos, and Austin’s reasoning about this is definitely weird (as in ‘unusual’ rather than ‘bad’, whereas for some folks, these two senses of the word are indistinguishable). It is unbecoming and ignoble to live in fear of what one’s least generous adversaries might say.
Another version of this point is that many EAs believe very deeply in the power of ideas and thus are worried about second order effects of stating/living one’s ideas, such as reputational harms and infohazards. I think these concerns are largely overblown. I think they make too much of our own importance and the extent to which people are paying attention to us whatsoever.
Having said that, I think a mocking headline about this hire is likely. (But where? NYPost? Current Affairs? Wired? Sierra Club? London Review of Books? Our haters are an ideologically rich array. A real cosmic gumbo, you might say.)
The strong feelings this post inspires suggests that the FTX trauma has not fully healed. I wonder if EAGs should have like, sessions or something where people who were affected by it can grieve together. Weren’t we all?