The Meat Eater Problem: Developing an EA Response
One issue that Effective Altruism has not fully developed a coherent response to concerns the “meat eater problem”: this is the concern that saving human life increases animal suffering as meat consumption goes up. This argument is usually called the “poor meat eater problem,” but I think this term is not quite accurate, given that the concern is stronger in the developed world, so I’m going to call it the “meat eater problem.” Many GiveWell charity recommendations, particularly the Against Malaria Foundation, could be affected by this consideration – if saving human lives means increasing suffering for a large number of animals killed for meat consumption, should we support human health interventions at all?
I think we still should, but the meat eater problem also increases my confidence that Effective Altruism as a whole is probably not supporting effective animal organizations enough. In this post, I agree that saving human lives probably increases meat consumption. However, I argue that 1) this may not be bad, because some systems of animal agriculture give animals net-positive lives, 2) the size of this problem might be smaller than we expect, and 3) even if these arguments are wrong, we can hedge our bets by giving a more substantial portion of our overall giving to effective animal charities. I conclude with four broader theoretical and practical points for Effective Altruists. I present these arguments with considerable uncertainty and would greatly appreciate comments of where I may be wrong.
The meat eater problem is logically plausible, but it might not be ethically bad
The basic logic that allowing more humans to exist—whether by saving lives or increasing fertility—will raise meat consumption is essentially correct. I see little reason to doubt that increasing the number of humans alive over at least the next decade or so will also increase animal consumption, and what limited evidence we have seems to indicate that this is the case. However, the argument that this is ethically bad – or at least worse than the benefit of saving human life – rests on some shaky ethical assumptions that I want to highlight.
Most importantly, the meat eater problem is only a problem if we assume that animals killed for meat live net-negative lives. I recently read Compassion by the Pound and was struck by the importance of the argument that we should try to judge whether lives are worth living on a −10 – 10 scale – this allows us to ensure we’re creating more enjoyment in the world than suffering. Although it’s difficult to tell how much suffering or enjoyment animals get out of their lives, the authors make one of the most comprehensive, systematic, and rigorous attempt at answering this question that I’m aware of to date.
The figure below, taken from their book (pg. 229), presents an assessment from one of book’s authors about the quality of animal lives under some factory farm conditions – it’s important that these numbers represent one author’s opinion and are highly dependent on assumptions about what animals need to enjoy their lives. I would love to see other researchers replicate these assessments based on what they know, but at least according to my ethical framework, I have no reason to believe that it is impossible for animal agriculture to create net-positive lives for animals. Certainly, the level of cruelty towards chickens, in particular, is shocking, representing one of the worst examples of suffering on the planet today. I’ve been very impressed by The Humane League’s string of victories in convincing Nestle, Sodexo, Starbucks, and a host of other companies to adopt cage-free egg policies – these types of welfare reforms are hugely beneficial and are a big reason I donate to The Humane League.
The argument presented in Compassion by the Pound leads me to think, at the very least, that certain systems of animal agriculture might offer net-positive lives for the animals within them, albeit nowhere near what they deserve, and that outreach strategies that reduce demand for meat might not necessarily be net-positive. If we assume that the author’s welfare numbers are correct for a brief moment, it is entirely possible that reducing demand for meat would actually prevent some animals from being alive when they would’ve benefited from more enjoyment than suffering overall.
Even though the meat eater problem is plausible, it’s probably smaller than most argue
It also strikes me as important that animal consumption in the developing world is much lower than in the United States and a much lower percentage of it comes from factory farms. Yes, factory farms are quickly growing in the developing world. However, it’s likely that people in the developing world will eat not eat as much meat as people in the developing world for quite some time and that some fraction of the meat they do eat will not come from factory farms.
I think it’s probably true that the size of the meat eater problem will grow over the next century, but I have considerable uncertainty that this should influence our decision-making on human health interventions. For example, health interventions that save lives, like bed nets, may lower human fertility in the long run, resulting in lower meat consumption. It’s also plausible that interventions that raise incomes, like deworming, have a lower impact on meat consumption because they don’t raise the overall number of humans that would be eating meat over their entire lifetime. My point here is to illustrate that there are substantial uncertainties all around that don’t allow us to claim that the meat eater problem actually is a problem or that saving lives in the developing world actually worsens it.
I would encourage anyone here to statistically analyze what kind of impact declining child mortality in the developing world will eventually have on meat consumption, assuming that factory farms are widespread in the next 10-50 years. I can’t say much here without considerable uncertainty, but as GiveWell argues, we shouldn’t rely extensively on arguments that require extensive causal chains to be true. In this case, for saving lives in the developing world to be ethically bad, you would need to prove that over the next several generations, 1) people in countries with GiveWell-supported charities will be eating some high percentage of factory farmed meat, 2) factory farmed meat will produce net negative utility for animals, and that the moral weight of the harm to animals is greater than the humans saved in the first place.
Even if I’m wrong about the meat eater problem, we can improve the chances we’ll solve it with investments in animal organizations today
Effective Altruism is very right to place importance on thinking at the margins. However, given an argument like the meat eater problem that depends on projections about the future, I think it’s possible that systemic factors and larger considerations might trump our individual influence. As such, I think we should put reasonable probability (ex. at least 10-20%) that we’ll develop widespread meat alternatives and/or have a high number of successful farm animal welfare campaigns that influence meat consumption in the developing world.
It’s certainly unlikely that we should depend on these issues going our way, but we do have some direct power over them: donating to effective animal charities that work in the developing world, like Mercy for Animals, as well as organizations creating meat alternatives, like New Harvest, are very high-impact actions that EAs should consider.
In the meat eater debate, ultimately, my mental model for the two major pathways EA could follow is this: 1) accept what strikes me as an unlikely string of causal chains and stop giving to all effective human charities or 2) continue supporting effective human charities and put a substantial portion of our charitable giving in order to minimize the ethical risk that the meat eater problem is true and harmful. The latter option makes sense, given the argument I’ve outlined above, and takes into account the possibility that I’m wrong about everything by boosting the probability that we won’t have widespread factory farming in the coming decades.
The most obvious counterargument is that the welfare estimates in Compassion by the Pound might be wrong and that most or all animals in factory farming endure net-negative lives. That’s entirely possible, as Brian Tomasik believes is the case. In all honesty, I simply don’t know enough to say at this point in my growth as an Effective Altruist interested in animal activism. My very loose impression is that factory farming as a whole is currently extremely net-negative, considering that so much of it consists of chickens in battery cages. I’m less sure of other systems of animal agriculture where welfare standards are higher (ex. Europe and the UK) and where animal agriculture systems are less commercialized, as in many parts of the developing world.
Fellow Effective Altruists might also argue that we should simply give to wherever has the highest expected utility, rather than stretch our donations across causes or organizations. In theory, I agree with this approach. However, I struggle to see how we can measure the impact of our donations against one another to the degree required to make this judgment with any degree of confidence, especially in light of uncertainties like the meat eater problem and far future concerns. For example, would you be able to say with greater than 80-90% confidence that a donation to the Against Malaria Foundation has greater expected value than to the Humane League or Future of Humanity Institute?
A more practically sound approach, to me, seems to be to consider where the distribution of donations in the Effective Altruist movement generally falls and allocate funding as “bets” according to how much an organization could use money at the margins and its expected value. Under this line of thinking, I believe we may be under-prioritizing animal organizations, far future research, and meta-organizations like Giving What We Can.
Some smart philosopher will probably point out, “but Scott, doesn’t your argument assume that we should favor total utilitarianism over average utilitarianism?” Perhaps. This is a philosophical debate that has not appealed to me very much because it strikes me as forcing us into making an arbitrary distinction. I’ve generally felt that we shouldn’t necessarily optimize for one over the other – they’re both good goals. I’m open to being convinced otherwise.
If you’ve followed along for this long, I’d like to outline several possibilities for what I think the meat eater problem should teach us about how to give effectively:
1) Family planning/abortion may be human health interventions worth considering more closely
One important argument is that interventions like contraception and abortion – particularly in countries with high meat consumption like the United States – may actually be cost-effective at reducing animal suffering and improving women’s lives. Although the possibility that we develop meat alternatives may reduce the potential benefit of this argument, it’s certainly worth considering. Even though the cost of contraceptive interventions in the developed world are significantly higher than in countries like India, perhaps the additional animal suffering prevented makes these interventions more effective than we previously thought.
2) “Lives saved” probably isn’t a good single most important metric, considering that its worth hinges so heavily on whether someone’s life is net good
This argument has been considerably discussed, but once again, I find it true. Whether a life is worth saving through health interventions depends on whether it is worth living. Perhaps we should allocate more of our charitable dollars in human charities to explicit welfare improvements—I’m not sure, but think this strain of research regarding whether the global poor consider their lives net-positive is one worth investigating. By the same token, this is an extremely important question for animal charities to consider as they weigh various factory farming interventions.
3) Let’s work more in India
Considering 40% of India’s population is vegetarian, but has a significant fraction of disease burden and poverty worldwide, this country represents a wonderful place for Effective Altruists to work.
4) EA could probably stand to give a much higher proportion of its money to animal charities. If we want to eliminate factory farming globally, this will require much more resources.
The sheer number of animals killed in factory farming gives enormous moral importance to reducing animal suffering. Harish Sethu’s presentations make this very clear. However, effective animal charities generally have miniscule budgets, with Animal Equality at just $2.5 million in expenses in 2014 – a fraction of the nearly $100 million in donations attributed to GiveWell this past year. I haven’t empirically assessed what percentage of EA dollars should go to animal charities, but on the face of it, we’re way too human-centered. Open Philanthropy Project’s recent grant to The Humane League is a great step in that direction.
Many thanks to Brian Tomasik and Peter Hurford, who provided extremely helpful comments and information in the process of writing this post.