Thanks for commenting, Joe!

I think that is definitely an important point, and it makes me believe the conditions of factory-farmed animals should be improved such that their rights are less violated. However, I do not know whether it outweights other factors like the increased human starvation in abrupt sunlight reduction scenarios, and potentially preventing wild animals from having good lives. Copy-pasting from my reply further down:

A hallmark of naive utilitarianism is strongly optimising for a single metric (e.g. number of factory-farmed animals) without adequately accounting for other potential important effects (e.g. on wild animals and longterm future).

Some may argue that this approach doesn't apply as much to animals, but perhaps it should as the world becomes less speciesist.

I'm Not a Speciesist; I'm Just a Utilitarian. Great piece from Brian Tomasik illustrating key differences between animals and humans:

When we examine the above list of points, we see that most of the utilitarian arguments for deontological / virtue-ethical behavior only apply regarding our interactions with other humans.^a For example:

1. Game-theoretic considerations only apply to other agents smart enough to interact in a game-theoretic way with us, which seems to exclude most non-human animals.
2. Decision-theoretic motivations for promises and honesty likewise only apply to agents who can understand the relevance of such commitments.
3. Striking fear into the hearts of a populace only applies to animals that read the news or spread gossip (except for other animals that may be present to directly witness or hear, e.g., a slaughter taking place).
4. Breaking down bonds of social cohesion only applies to animals who enter into long-term trusting relationships with humans. (So, e.g., it might be bad on balance to violate the trust of your pet dog even for good reasons but not bad on balance to break the trust of a wild animal for good reasons.)

The only arguments from the previous section that seem to apply clearly in the case of animals are

1. overriding severe miscalculation of costs vs. benefits regarding actions that harm some individuals
2. preventing self-serving behavior that's justified under the pretext of advancing the greater good.

While these two considerations are important, the main force of the utilitarian arguments for deontology is lost in the case of non-human animals.

In essence, as I commented, humans are not only moral patients, but also moral agents.

Nice points!

I’m pretty skeptical of your points related to food shocks and abrupt sunlight reduction scenarios, especially given the academic controversy surrounding nuclear winter scenarios (which I don’t believe you’ve adequately updated on despite this controversy being pointed out to you beneath one of your other posts).

I did not update my analysis on why more global warming might be good to mitigate the food shocks caused by abrupt sunlight reduction scenarios following the comments because I was already aware of the controversy. I decided to defer to the results of Luisa Rodriguez, who thought about the matter much more than me. My position is similar to the one Toby Ord expresses in The Precipice (emphasis mine):

There are significant remaining uncertainties at all stages of our understanding of nuclear winter:

1. How many cities are hit with bombs?
2. How much smoke does this produce?
3. How much of the soot is lofted into the stratosphere?
4. What are the effects of this on temperature, light, precipitation.
5. What is the resulting reduction in crop yields?
6. How long does the effect last?
7. How many people are killed by such a famine? Some of these may be reduced through future research, while others may be impossible to resolve.

Some of these may be reduced through future research, while others may be impossible to resolve.

Skeptics of the nuclear winter scenario often point to these remaining uncertainties, as they show that our current scientific understanding is compatible with a milder nuclear winter. But uncertainty cuts both ways. The effect of nuclear winter could also be more severe than the central estimates. We don’t have a principled reason for thinking that the uncertainty here makes things better. Since I am inclined to believe that the central nuclear winter scenario is not an existential catastrophe, the uncertainty actually makes things worse by leaving this possibility open. If a nuclear war were to cause an existential catastrophe, this would presumably be because the nuclear winter effect was substantially worse than expected, or because of other—as yet unknown—effects produced by such an unprecedented assault on the Earth.

It would therefore be very valuable to have additional research on the uncertainties surrounding nuclear winter, to see if there is any plausible combination that could lead to a much deeper or longer winter, and to have fresh research on other avenues by which full-scale nuclear war might pose an existential risk.

I agree that:

Also, note that in any such scenario, the human population will be far lower than it is today, so the quantity of animal feed is almost certainly not going to be the thing that determines whether humanity recovers from collapse (especially if we’re assuming that such a scenario occurs in the next few decades before AGI, when animal product consumption is expected to be pretty high).

However, even if animal feed has a negligible impact on the likelihood of recovery, it may influence the chance of collapse.

Robin Hanson has also questioned whether farmland used to grow crops for animal feed would be ‘re-wilded’ - at least some of it will be used for development, which will actually reduce wild animal numbers. In any case, whether or not wild animals have net-negative lives is incredibly uncertain.

Agreed, but note that "uncertain effect with zero expected value" + "certain effect with positive expected value" = "uncertain effect with positive expected value". If the combined effect is sufficiently uncertain, I think it is often better to learn more, and keep options open, instead of acting as if one of the options is robustly good/bad.

So, the direct effects on factory farmed animals of reducing consumption, plus lower greenhouse gas emissions (which, given my skepticism about your food shock points, and the fact that climate change probably increases the risk of an existential catastrophe by reducing global stability and increasing the likelihood of extreme/tail-risk climate scenarios), seem to make reducing consumption net-positive.

I would say this corresponds to the common sense view, and I believe it may well be right. However, because of what I said just above, I do not think we should act as if we are sure that is the case.

Hi Sanjay,

I agree the case for reducing consumption is stronger for factory-farmed chickens (or other animals living super bad lives which have a small impact on global warming).

Thanks! I have now added:

• Mitigating antimicrobial resistance:
  • From Carrique-Mas 2020, "the greatest quantities of antimicrobials (in decreasing order) were used in pigs (41.7% of total use), humans (28.3%), aquaculture (21.9%) and chickens (4.8%). Combined AMU in other species accounted for < 1.5%".
  • Boeckel 2015 projects that "antimicrobial consumption [in livestock production] will rise by 67% by 2030 [relative to 2010]".

The same crops are already used for bioethanol before being sold and fed to farmed animals, as distillers grains. Decreasing animal product consumption could therefore reduce bioethanol revenues from selling the byproducts for feed, but it could also reduce input grain costs.

Great point, Jason!

So I am not confident a small-to-medium refuction in animal feed demand would ultimately impact crop acreage that much.

I agree confidence is not warranted.

I don't know about other markets though!

It looks like there is significant variation across countries (the relevant metric is per capita production, but I did not immediately find it):

Thanks for the thought experiment!

I am very strongly against forcibly impregnating or murdering humans under almost all circumstances, including the one you describe. Humans are not only moral patients, but also moral agents, and these being forcibly impregnated or murdered is highly anti-cooperative. I follow expectational total hedonistic utilitarianism, but this only justifies creating net positive lives all else equal. In a society where humans are forcibly impregnated and murdered for food, I think it is fair to say that all else would not be equal! One should take into account the actions involved in the creation of lives, and forced impregnation and murder are actions with pretty bad consequences. For example, they would tend to normalise violence and inequality, which correlate with lower wellbeing longterm.

These concerns apply to animals too. Forcibly impregnating or murdering animals is still uncooperative, and my ideal world does not involve those. However, by decreasing the consumption of animals, one may accidently increase the suffering of wild animals, and the existential risk from ASRSs. So, as a precautionary measure, I think it would be better to increase resilience to ASRSs before/while driving down the number of factory-farmed animals. I think welfare reforms are more robustly good because they significantly decrease the suffering of factory-farmed animals while not decreasing much the resilience against ASRSs, at least in the nearterm (in the longer term, continued welfare reforms will increase prices, and therefore decrease the consumption of animals).

Thanks, Matt!

I love how your posts often bring together points about different cause areas, making connections between topics that those focused on particular causes are perhaps either unaware of or choose to ignore because they are complicated and inconvenient!

Nice that you like them!

Do you have an estimate of how likely an abrupt sunlight reduction scenario (ASRS) is to occur over the next (e.g.) 100 years? My intuition is that for the cases of volcanic and impact winters it's extremely low, perhaps less than 0.1%. In which case it probably comes down to the likelihood and consequences of nuclear war.

I agree nuclear war is the driver of the risk from ASRSs. When I last estimated the risk, I used Metaculus' community prediction for a global thermonuclear war by 2070, which is currently at 13 %. For the ejection of soot into the stratosphere conditional on a global nuclear war (as defined by the Metaculus' question), I used the results of Luisa Rodriguez, who thought about the matter much more than me. To estimate the reduction in future value given a certain soot ejection, I relied on historical data about socioeconomic indices plus a bunch of guesses.

I also wonder to what extent food shocks could be mitigated by the development of plant (or fungi) crops that are much more able to tolerate ASRS conditions. I can imagine these sorts of crops might be developed for the purposes of space exploration, e.g. if humans attempt to establish permanent bases on the Moon and Mars over the coming decades.

Great point, increasing the consumption of animals is far from the only way to increase resilience against food shocks. From ALLFED's page on resilient food solutions:

• High-tech solutions:
  • Single cell protein from CO₂ and hydrogen.
  • Food from plant fiber.
  • Single cell protein that uses natural gas.
• Low-tech solutions:
  • Simple, scalable greenhouses.
  • Seaweed.
  • Reallocating food for humans and animals.
  • Leaf protein concentrate.
  • Relocation of cool tolerant crops.

Thanks for asking!

70 % to 80 % of deforestation is driven by conversion of primary forest to agriculture or tree plantations:

Since more agriculture tends to result in more deforestation, I guess less agriculture leads to less/more deforestation/forestation. In any case, I do not know whether terrestrial arthropods have good or bad lives, so regardless of whether their population would increase or decrease due to greater consumption of animals, I would not be able to tell whether the effect was good or bad.