Summary

In January 2026, the animal charity Farmkind launched a controversial news campaign advertising their ability to “offset” the impact of non-vegan diets through strategic donations to EA causes. I have now spoken with several non-vegan EAs who use Farmkind (or similar calculators) to buy “offsets” for their diets, and I think the offsetting logic is more powerful than vegans would like to admit. However, I think EAs generally draw the wrong conclusions from this line of reasoning. Even (maybe especially) under hardcore consequentialism I believe there’s a very strong case to be made against using “offsets” of any kind. As a vegan advocate now working in animal welfare policy, I feel strongly that the reluctance of many EAs to commit to veganism represents a major missed opportunity for the movement, and an objectively massive source of animal suffering.

How Offsets Work

By going vegan, you can reduce total market demand by something like ~200 animals (excluding shrimp) every year (1). However, due both to the astonishing number of animals killed for food every year and the immense tractability of targeted welfare reforms, going vegan is far from the most effective way to help animals. Farmkind estimates that donating only $276 to effective charities accomplishes exactly the same welfare improvements as veganism.

This kind of calculation might seem slightly fishy or underhanded. One reasonable objection was made in Clare Zabel’s 2015 EA Forum post, which pointed out that it is difficult to establish which charities are most appropriate to use as offsets, since the class of animals being helped is by definition distinct from the class of animal being harmed; a problem she referred to as using “vague and arbitrary reference classes.” 

Indeed, Farmkind’s calculator sometimes seems to make use of some pretty far-out claims about animal preferences—for instance, the claim that two chickens raised under the guidelines of the Better Chicken Commitment experience the same net suffering as one normally farmed chicken. It is somewhat disturbing to think of one animal’s welfare being “sacrificed” to save the welfare of another.

However, I do not think that this problem is endemic to offsetting. To some degree, these kinds of calculations across moral subjects are bread-and-butter for EAs. We have no problem sacrificing our own welfare, in the form of financial deprivation, to make donations to those in greater need.

Furthermore, at least in the case of veganism, I think there are several strong contenders for appropriate offsetting reference classes. To me the most obvious choice is Sociedade Vegetariana Brasiliera, which is estimated to replace 116 animal-based meals per dollar. Assuming that the average non-vegan American eats 2,000 meals per year, of which perhaps 1,800 include animal products, then a donation of only $16 would be sufficient to neutralize his annual impact on the world’s net demand for animal products, without any need for elaborate welfare calculations.

But even Farmkind’s more generous number is still disappointingly low. To many EAs I have spoken to, it seems to imply that going vegan is only “worth” a paltry $276—or alternatively, that donating at least $276 to pro-animal charities “buys” them the freedom from being vegan.

Why Offsets Don’t Really Exist

I believe that these last two comments (which were taken from real conversations I have had with EAs) display a logical misunderstanding of what donations are, and how they work.

This misunderstanding is encouraged by the marketing term “offset.” As Farmkind explained in a January 2026 EA Forum post, “our core message [is] that you don’t have to be vegan to help animals. We think that many people who are not willing to change their diet currently assume that there’s nothing for them to do about factory farming. We want them to know that’s not the case, opening the door to future actions for animals and a shift in their identity towards being ‘pro-animal’.”

This post makes it clear that Farmkind’s campaign is oriented at non-EAs. It is not trying to get normal people to do the most good, but to do any good at all. Ethically speaking, there should be nothing special to a consequentialist about donating $276 rather $27,600—you should always donate as much as you feel able to. The word “offset” is only a motivating fiction to make consumers feel guiltier, and hence more likely to donate. In this way it resembles the charities that ask people to donate some amount that is “less the price of your morning cup of coffee.” It plays on deontological sentiments about what you are “personally obligated” to do versus what is “not your problem.”

Other EA cause areas rarely consider “offsets,” both because it feels disrespectful and because offsetting even a single human life is so expensive. For instance, we do not ask great powers to simply “offset” the human cost of war by donating to the Against Malaria Foundation—we work actively both to reduce great power conflict and to increase funding for malaria prevention. The two facts are independent.

Perversely, however, the remarkably low price of saving a single animal’s life (just a couple cents) seems to be sometimes interpreted to diminish the value of that life—to make it seem like a rounding error, only “worth” a stick of gum. Around 50% of EAs are neither vegan nor vegetarian, often because of this belief in the “cheapness” of animal life. Conservatively estimating around 10,000 active EAs, that comes to nearly one million animals being slaughtered annually to feed a community that routinely ranks animal welfare as one of its top priorities.

In this way, perhaps pro-animal EA charities have become too effective!

What Should You Do Instead?

A critical piece of the previous argument is that eating a vegan diet is independent of one’s ability to donate—in other words, that you may always do both. Indeed, if you are at all concerned about animal welfare, the EA ethic to do the most good demands that you do both. If you’re donating $20,000/year to charity, you should also be vegan. If you’re donating $20,276/year to charity, you should still also be vegan. It is a standing opportunity to save hundreds of lives.

But some EAs I have spoken to are afraid that going vegan will reduce their productivity or earnings by more than $276/year. If this were true, then I wholeheartedly endorse their decision to eat an omnivorous diet and donate their extra earnings wherever they please.

However, I think there are at least three reasons to consider this fear more carefully.

1. Cost savings. According to a study conducted by the University of Oxford, vegan diets reduce annual food costs by 21-34% in high-income countries, which would save the average American about $2,796 per year (and around $4,800 for Americans in the top income quintile). An observational study of 11,000 Britons confirmed savings of 40% by those eating a vegan diet.
2. Social multipliers. While causal mechanisms are difficult to determine, it is well established that being vegan significantly increases the odds that people around you will decrease their animal consumption. It also provides statistically significant social validation for existing vegans, who are consequently much more likely to remain vegan. (I personally went vegan at the urging of a vegan friend several years ago—in fact, it was going vegan that drew me into EA). 52% of new vegans report that someone else has tried veganism as a result of their decision. Conservatively, each vegan might plausibly create/preserve 5-10 other vegans over the course of a lifetime, and reduce the animal consumption of 10-50 more.

3. Health effects. Nutrition is a notoriously complex and person-specific issue. However, it is very clear that a reasonable vegan diet does not raise serious health concerns. At a population level, meat-less diets are often associated with a reduction in all-cause mortality around 10-15%. This might allow vegan EAs to remain productive for several more years than non-vegan EAs.

    

The first two facts alone threaten to push the “price” of veganism for the average EA significantly higher than Farmkind’s estimate, potentially above ~$4,000 per year. (If a vegan diet did extend your working life by several years, as the most sophisticated comparative model currently predicts, that would represent a massive increase in total earnings—but this question is technically unsettled and awaits more precise causal studies.) In other words, productivity savings from not being vegan should certainly exceed $4,000 in order for a consequentialist to feel secure in their decision.

But even this number does not account for several broader consequences of veganism, including:

1. Increased accuracy in animal sentience estimates. A couple famous studies have confirmed that even the act of anticipating consumption of animal products later reduces your estimates for both animal sentience and intelligence. These effects are probably dampened among EAs, who think harder about these issues than the general public. But in a field with as much uncertainty as animal welfare ranges, it seems important to do everything we can to keep our heads unclouded by motivated reasoning.
2. Investment in alternative-protein development. Purchasing vegan meat (Beyond, Impossible, Tofurky) directs additional cash flow towards alternative-protein development—one of the most important destinations of EA animal welfare funding today. It also encourages stores to keep the brands in prominent positions on the shelves or on the menu, which increases both sales and brand awareness.

3. Boosting the social credibility of animal welfare. The general public is conflicted about farmed animal welfare. They are uniformly against animal cruelty, but uniformly for eating animals. Furthermore, they never donate to animal welfare charities and do not prioritize pro-animal ballot initiatives or political candidates. Speaking to people about veganism can raise the profile of the issue—not just for those considering going vegan, but for the general public.

    

Given all of these factors, I believe that EAs should think more seriously about adopting a vegan diet. Diet change is often perceived as more difficult than it actually is: according to Veganuary, 81% of people that undertake their 31-day challenge make significant cuts to their animal-product consumption. If all of the non-vegan EAs simply signed up for Veganuary’s challenge, we might expect to save hundreds of thousands of animals—a significant accomplishment for the movement and a declaration of faith in our own principles.

(1) This number varies significantly between estimates, depending on whether you count “bycatch”—that is, the less desirable fish accidentally killed while trawling for wild fish. Farmkind’s estimate is 41 animals per person per year, but they do not count bycatch.