I write about meat reduction research at https://āāregressiontothemeat.substack.com/āā and my date me doc is here š
Here is an overview of the lab I work at
Seth Ariel Green šø
Karma: 1,616
Hi David,
Itās true that we arenāt 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 beliefāout of scope of the dataāis that we can think of consumers as extrapolating, when they see an item like āchicknāitas,ā that itāll 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 Unfortunately, PMAs account for a small fraction of all sales (3% per this out-of-date article but I couldnāt find anything more up to date and official.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 ([ā12.58, 0.73]) is wide, and we say in the paper that we canāt 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.
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!
š Iāve 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 donāt have a singular directional intuition about this.
Iām not sure AGI isnāt 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, Iām afraid here and now it sometimes leads to worse discussions.
Hi James,
Iām 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 four existing groups:
Better Food Foundationās DefaultVeg Campus Program
Plant-Based Universities (UK-based)
What appeals about going your own way?
I have not read this paperāI will put it on my reading listābut for folks interested in the subject, I recommend Norwood and Lusksās 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 itās ok in most cases to ask someone to be nice on the internet. To quote Juana Molina, āNo seas antipĆ”tica. No seas antipĆ”tica con tu mamĆ”. La la la.ā 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 flambĆ©ed also said āsorry I was used to less nice parts of the internetā so I guess that went down well
When someone says ābe niceā 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 š¤š¤š¤
Dean Ballās commentary on this refamed the issue for me https://āāwww.hyperdimensional.co/āāp/āāclawed
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 āthis fighter jet is not certified for flight above such-and-such an altitude, and if you fly above that altitude, youāve breached your warranty,ā and āyou may not fly this jet above such-and-such an altitudeā). 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 thereās something iffy/ānon-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 āyouāre welcome to use our drug so long as you donāt use it in any abortions,ā 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: āTo 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.ā In other words, because we know what the law should be and what itās probably going to be, we should implement that policy today. I think many of us can think of examples where weād be uncomfortable with a billionaire tech CEO saying that.
I am glad that you think this issue is tractable and Iāve been following your work with great interest since I saw your RECAP talk in July (Side note to anyone following thisāthe RECAP talks are great!). I am not sure what my own threshold for ātractableā 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 Iām glad weāre a big tent where people can prove something possible by doing it.
Thank you, Iāve now hyperlinked to the piece in both of my prior comments where I use the word ānormsā (in part so that I remember it next time š)
Well, you know me, always playing nice ššš
I do want to say, in Andrewās defense, that the comments on @Alistair Stewart ās original post are not exactly the very model of civility that EA might hope to show the world. I can understand why youād read them and come away with the sense that people donāt really get what youāre trying to do.
However, the point that calling something a āleading cause areaā requires cross-cause comparison is well-taken.
Personally, I donāt think āmostly or entirely plant-based diet for dogsā sounds nutty. I think most people understand that dogs can subsist on literal garbage. Might not be optimal but I think we can make the case that if you feel really strongly about supplementing a healthy plant-based diet with animal protein, it should come from bivalves.
The case for cats is much less intuitive, I think.
By the way, Ben
Knight claims that there are other economically productive uses for byproducts; if thatās true, then a reduction in demand for animal-derived pet food would change the marginal use case for byproducts but not reduce their production
I donāt think ānot reduce their productionā follows from your reasoning. If the next available use cases pay less money, we should, in espectation, see fewer animals raised, no?
I have twice recently āgently counseledā people on EA forum norms when they come in, in my opinion, a little too hot for this rather cool medium š is there something official/āCEA-endorsed on this subject? If not, should I/āsomeone write it? I could point them to Scout Mindset but thatās kind of a high barrier to entry.
Hi Andrew, welcome to the forum! I am keenly interested in this subjectāI am one of the commentators you mentioned and have written on the subject previously (Towards non-meat diets for domesticated dogs).
Without getting to much into the specifics here, I wish to gently counsel you on EA forum norms in a way that might help the message go down better for readers.
We generally assume good intent. It is true that I am not persuaded by some elements of your analysis, which is why I stated a preference for the Alexander et al. estimation methods, but I would not describe that disagreement as āseeking to undermine [your] studies.ā My disagreement is not coming from a place of malice.
We generally do not use maximalist language to describe each otherās perceived mistakes, e.g. āprofoundly misrepresents,ā ādramatically incorrect,ā etc. Instead it is more in line with how we talk to say āThis is mistakenā or āthis is not what I intended.ā
We tend to address each other by name/āusername and use tagsāby all means please call me Seth rather than āa commentatorā š
Anyway, looking forward to more engagement,
Thank you Vasco! Iāll be curious to hear folksā responses, if any.
Hi there, thank for engaging with our results. On a preliminary note, I wish to gently counsel you that the tone of this response is not in keeping with the norms of this community, which tend towards being collegial/āassuming good intent/āetc. Instead of noting that our methods ādonāt bear any useful relationship to the real world,ā you might try something like āIām afraid the methods used here donāt answer the questions that I think are most important to the movement.ā
Yes, a hypothetical survey has important limitations. I think we are upfront about them. If we had found larger effects here, we likely would have pursued funding to take this design out to a restaurant or setting with consumption outcomes. When we didnāt, we went a different direction. Always we are triaging.
I agree that flexitarians/āmeat-reducers/āwould-be-reducers are probably the best target audience for PMAs. We are testing a different estimand, which is demand for the product in a silmulalcrum in the world at large. This quantity is important from the POV of, e.g., the viability of marketing and deploying PMAs at popular restaurants like Chipotle.
Ah my favorite subject/ābeloved nemesis, plant-based defaults!
I am a bit of a skeptic, as I laid out in Scaled up, I expect defaults to reduce meat consumption by ~1-2 pp. My basic point is that we donāt have a good sense of the effect of these interventions on dietary change rather than meal change, and that some theoretical considerations might give us pause.
Regarding this essay specifically:
You write āThe default effect was popularized by Johnson and Goldstein in 2003, demonstrating how countries with similar cultures and religions show dramatic differences in organ donation after death, all because of choice architectureā. Alas, when comparing countries that have opt-in vs opt-out policies, Arshad et al. (2019) find that āno significant difference was observed in rates of kidney (35.2 versus 42.3 respectively), non-renal (28.7 versus 20.9, respectively), or total solid organ transplantation (63.6 versus 61.7, respectively).ā
The NYC hospital data is not an RCT and doesnāt offer much of an identification strategy so Iām never sure what to make of those numbers. That being said, Iām ready to accept in principle that hospitals are a particularly promising environment for defaults and for people rethinking their diets generally.
The Ginn & Sparkman numbers are interesting but the decline in overall sales on plant-based default days is going to be a bit of a challenge for selling the strategy to institutions. On the other hand, Sodexo is apparently happy to scale them up so who knows.
I am working on a plant-based default evaluation that I hope to share with the forum soon š
How much of a post are you comfortable for AI to write?
AI sucks at writing and doesnāt get my style at all. If it did, Iād let it write the whole thing. (Yudkowsky predicts we have about 2 more years until AI learns to write.)
Hi there
š I am lead author of that paper!
Only one study we looked at addressed alt proteins directly
From that paperās 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. Other secondary and exploratory outcomes included the consumption of meat substitutes, cardiovascular risk factors, psychosocial variables related to meat consumption, and the nutritional composition of the diet. We also estimated the interventionās environmental impact. We evaluated the intervention using generalized linear mixed-effects models.
Results
Between June 2018 and October 2019, 115 participants were randomized. The baseline meat consumption values were 134 g/ād in the control group and 130 g/ād in the intervention group. Relative to the control condition, the intervention reduced meat consumption at 4 weeks by 63 g/ād (95% CI: 44ā82; P < 0.0001; n = 114) and at 8 weeks by 39 g/ād (95% CI: 16ā62; 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. The intervention increased intentions, positive attitudes, perceived control, and subjective norms of eating a low-meat diet and using meat substitutes, and decreased attachment to meat. At 8 weeks, 55% of intervention recipients identified as meat eaters, compared to 89% of participants in the control group.
* 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.
Iād say that one study found pretty positive effects from introducing plant-based alternatives by its own lights. However itās a small sample, from a long time ago, etc. So my read would be, our meta found scant evidence either way of the effects of introducing alt proteins on dietary behavior, and the study we do have embeds them as part of a larger behavioral change program.
When we test intro-ing alt proteins in a hypothetical, online ordering, chipotle-like environment, alas we donāt find much.
P.S. pardon my omission, there was one other study in our dataset that looked at providing meat alternatives to people for free
but this one is from so long ago, and it was a pretty narrow population, that I think it doesnāt illuminate very much about the likely effects of todayās alternatives.