Factory farming in a cosmic future
Humanity may inadvertently propagate animal suffering caused by factory farming when we establish settlements in space because we are too confident in futures free of animal suffering. This could pose a significant suffering risk (or s-risk). In this post, I aim to show that such confidence is misguided. I start by explaining why factory farming in space is (probably) worth serious moral consideration. I then present four common reasons for believing that humanity’s future in space will likely be cruelty-free, and critique them as best I can. I conclude, very tentatively, that a future full of factory farming is perfectly plausible, and currently our default trajectory.
I have made three assumptions which I do not defend in any detail here, and which I have taken for granted as plausible:
We are sufficiently concerned with animal welfare overall that a graphic, gratuitous description of just how bad factory farming in space could be is unnecessary here.
Humans could one day settle permanently in space, and we will do so in recognisably human form (i.e. not as digital minds, and not ‘post-human’ to the point of no longer requiring food).
Tens of thousands of people (at least) would need to be living beyond Earth for factory-farming in space to be a realistic development. I don’t know how long this could take, and I do not attempt a forecast here, but this might take the form of multiple Earth-orbital stations and semi-permanent stations on the Moon with rotating populations, for instance, or even permanent settlements further afield on or around Mars.
Epistemic status and caveats
I am about 75% certain that factory farms in space will eventually be technically and commercially feasible, if humans settle in space. If this happens, it will probably happen too fast for most interventions to interrupt it, and there will be new, strong incentives for factory farming to continue indefinitely.
Why might this matter?
Yip Fai Tse has estimated that in a future where humans settle throughout the Solar System, there could be as many as a thousand trillion trillion vertebrates farmed and slaughtered for human consumption. This doesn’t even cover the insects, crustaceans and molluscs (and many more animals) we may farm in future. He wrote that the spread of factory farming into space might be one of the most important considerations for future animal welfare, and notes that work is already being done on feeding humans in space by farming fish and insects.
There may also be normative reasons to take more seriously the possibility of factory farming in space. In Suffering-Focused Ethics, Magnus Vinding argues that happiness and suffering are deeply asymmetrical, rather than arbitrary ends of a well-being scale that can be algebraically balanced or outweighed. This implies that while we may still have good reasons to focus predominantly on the wellbeing of future people (and future digital minds), we are not at all exempt from giving due consideration to animals in the meantime.
Finally, as Fai notes, if animal welfare matters less in the longterm because we can expect there to be many, many digital minds, we should expect human welfare to matter less, too. This is a deeply complex subject but it should encourage us at the very least to find a stake in this argument – how much can we expect to matter morally, if we’re sure that this won’t?
The issue seems neglected within EA, however, and where it is considered, it is quickly discounted. 80,000 Hours have estimated that “the scale of the problem of factory farming itself…is small compared to the scale of issues affecting future generations. This is especially true if you think that factory farming is likely to end within a few decades or centuries.” Their exploratory piece does not suggest reasons to think that factory farming may soon end, and does not explore the possibility of factory farming in space. In addition, EA user alene described the scepticism they encountered from some effective altruists regarding the future of farmed animal suffering, and the confidence that it won’t be a risk worth addressing.
I address tractability in the Conclusion.
Reasons for optimism, and rebuttals
In this section, I will describe and critique four reasons for optimism that humanity’s future in space will be cruelty-free. There are likely many more reasons for optimism than I describe here, and many more valid critiques thereof, but I hope merely to show that we probably haven’t considered them enough overall rather than that there aren’t good reasons.
1. Earth can supply meat products to low-Earth orbit, the Moon and even Mars long-term
Reasons for optimism: Pre-deployed food caches and uncrewed resupply missions are possible, in principle, for an enduring human presence on Mars. We can currently land on the Red Planet within a few kilometres of a chosen target, which a rover could reach without much difficulty if needed, and well-packaged dehydrated food could plausibly remain edible for 5-7 years after launch. We could probably also send components for a small indoor farm for hardy, high-caloric vegetables (e.g. potatoes, à la The Martian).
First, a commitment to resupplying interplanetary settlements means factory farming might expand on Earth, unless other norms change too. For our purposes, it matters less where animals are farmed and more that the demand for particular products has increased in line with our population and activities further away, to the detriment of animals.
Second, there are obvious risks to this approach such as launch failures, spoilage en route, destruction of food on landing, and caches passing their expiry dates before use.
Third, there will be strong incentives for human settlements beyond Earth to strengthen their own food security and reduce dependency on a fragile Earthbound supply chain, including from an existential risk perspective, and factory farming in space could be seen as a reliable source of protein-dense foods.
Finally, this might work for Mars, but it is far less certain further afield. Given our current propulsion technology and the transfer windows between optimal orbits, a crewed mission to Saturn’s moons and back could take well over a decade, requiring several metric tonnes of food even for a small crew. Keeping food safe and edible with reliable failsafes in place for this enormous journey will be far more difficult, even if we can get such payloads into orbit.
2. Lab-grown meats and alternative proteins will render factory farming obsolete
Reasons for optimism: Lab-grown meat is a fast-growing industry that has seen a great deal of progress over the last decade. It is often portrayed as the closest thing to a silver bullet for ending factory farming, and limiting the long-term effects of intensive animal agriculture on the environment.
First, lab-grown meat has only recently become available to retail consumers, at relatively high cost and in limited markets. It will need to pass through lengthy, expensive and time-consuming regulatory hurdles worldwide before it reaches major markets, and it is difficult to know how quickly this will happen (Singapore gave regulatory approval in 2020; the US and some EU countries are in progress).
Second, the technical progress may be much more uncertain over the medium- to long term. In a 2020 report commissioned by Open Philanthropy, David Humbird wrote that “the cost of cultivation facilities will always be too burdensome, and the cost of growth media will always be too high, for the economics of cultured meat to make sense” (the full report can be found here).
Third, if lab-grown meat turns out to be commercially-viable to produce at affordable prices, it is unclear how competitive it will be against ‘traditional’ meats. We should expect this to be a political and cultural issue, as the agriculture industry is a major employer and enjoys preferential treatment in many jurisdictions, and some consumers may regard lab-grown meat as ‘unnatural’ (and therefore undesirable, unhealthy or outright harmful). Factory-farmed meat may be promoted as more ‘authentic’ than lab-grown alternatives, or supportive of particular political views.
Fourth, the necessary bioreactors and raw materials for lab-grown meat will have to survive the high-g stresses of launch and short periods of microgravity. Bioreactors have been tested in space at a limited scale, although fluid dynamics, gas exchange and cellular behaviours are difficult to control in these environments. While something similar can be said about transporting live animals into LEO space, bioreactors require maintenance, repair and access to Earthbound supply chains that will still incur major costs; once provided with enough calories, live animals are (unfortunately) the most robust, affordable and disposable infrastructure needed to produce commercial quantities of meat, eggs and dairy.
Finally, without improvements in propulsion technology, it will be just as expensive (per kilo) to transport alternative proteins into orbit as anything else. In the long-term, this may make lab-grown meat less competitive against fungal, algal and other alternative proteins that may be successfully cultivated in space. This could be deceptively important: if meat-free diets become the norm in space because meat is simply unavailable, sufficient demand may make factory farming in space commercially viable long before lab-grown meat. Humans in space may never make the switch.
3. Transformative AI will render factory farming in space unnecessary
Reasons for optimism: A comprehensive assessment of the problems AI could solve for us is beyond the scope of this post, but the opportunities and risks of this technology are well attested elsewhere. In short, transformative AI could far outstrip humans in the quality and quantity of technological advancement and economic growth it could produce, including innovations in biotechnology and engineering that make factory farming in space undesirable and/or unnecessary.
First, transformative AI would have to mitigate not only the technical challenges of lab-grown meat cultivation, but also the regulatory and political/cultural issues mentioned above. Transformative AI solving technically-challenging problems does not necessarily mean it will solve all relevant problems.
Second, transformative AI may also solve the engineering problems that make reaching orbit so expensive. This may make it commercially viable to quickly send meat products into space at competitive prices before lab-grown meat can compete.
Third, the impacts of existing AI already extend to factory farming itself. This may mean that transformative AI makes it harder for lab-grown meat and other alternative proteins to compete on their own terms, as factory farms are already taking advantage of economies of scale not necessarily available to lab-grown meat companies. This could also extend to advances in genetic engineering that allow animals to survive and reproduce in low- and zero-gravity environments.
Finally, we simply do not know what transformative AI will be like. It may be possible to create digital minds which experience consciousness, which presents important ethical and legal considerations if we expect AI to work on the issues we care about. It may not be enough to solve the alignment problem without also answering important questions about the rights of digital beings (see Holden Karnofsky, Nick Bostrom & Carl Schuman, and Will Macaskill’s intention to work on this).
4. Norms will change in favour of meat-free diets and better treatment of nonhuman animals
Reasons for optimism: We might welcome news that vegetarian and vegan dietary habits are on the rise (Google reported huge increases in the search term “vegan restaurants near me” in 2019), but these trends are not global, nor one-way: Statista reported in late 2022 that vegan diets declined in India between 2018 and 2022, while Mexico and Spain saw very little change at all.
First, norms are powerful, but malleable. There is valuable work elsewhere on moral progress, the fragility of value, the contingency of history when it comes to ending suffering, and the biases that prevent us properly understanding suffering. This strongly suggests that we are not at all guaranteed to move en masse toward meat-free diets, or taking nonhuman suffering more seriously. A well-known example of counterintuitive moral regression is slavery, which was commonplace in the ancient world around the Mediterranean but had all but disappeared in Europe by the Middle Ages: its mere absence, however, was not enough to prevent the Atlantic slave trade.
Second, if and when we live on multiple planets, it might be even harder to change norms through discourse. Real-time conversations between Earth and Mars will be impossible, because one-way transmissions will take a minimum of three minutes at best (extending to more than twenty minutes when the two planets are furthest apart); two-way exchanges will take hours, making cooperation and convergence on controversial issues much more difficult. This gets worse the further from Earth we inhabit: for example, delays between Earth and the moons of Jupiter will take three times as long as those with Mars).
Why I might be wrong
There are five main ways in which I think I could be wrong about factory farming in space being a worthy concern. I provide a brief response for each.
We may not live long enough to settle in outer space at all, due to nuclear war, malevolent AI or some other existential catastrophe. This is true, strictly speaking, but it doesn’t seem like a strong discouragement from taking factory farming in space more seriously. Nobody who works to reduce pandemic risk argues against doing so given the risk of nuclear war, for example, and other important work on the alleviation of absolute poverty and animal welfare continues despite these ongoing existential risks.
We may make a political decision to remain on Earth long-term, perhaps to reflect more carefully on the allocation of resources in space, or to heal the planet we come from before we try to settle elsewhere in the cosmos. This is possible, given the malleability of norms described above. Spacefaring states are already competing for access to low-Earth orbit and beyond, however, and the welfare of nonhuman animals should ideally form part of any discussion of how to govern ourselves in space.
We may convert ourselves to a purely digital form, in which we are eternal, editable, replicable, transferable and free of the suffering imposed by (and on) organic bodies. Given the uncertainty of how transformative AI might unfold, this may be unavailable and/or undesirable for many people. It also raises more complex questions about a future in which humans, nonhuman animals and digital minds share the cosmos.
We discover strong normative reasons to believe that humans and nonhuman animals are not worthy of the same moral consideration as ‘super-patients’ or ‘super-beneficiaries’, e.g. digital minds. I don’t yet take a strong view here, but see Nick Bostrom and Carl Schuman (p. 14) for how this wouldn’t necessarily mean bad lives for humans and nonhuman animals. Even in a universe where digital minds have a much, much larger claim on the proportion of resources in the cosmos, we might still have good reasons to use our smaller absolute share to prevent suffering for nonhuman animals.
I have tried to carefully critique four common reasons for optimism in a human-led future in space, a future in which nonhuman animals no longer suffer in factory farms. My current best guess is that Earth probably cannot reliably supply farflung human settlements in space, and that even with the opportunities provided by transformative AI, we are at least marginally more likely to enable a future full of orbital and extraterrestrial factory farms, as one without.
Tractability, and tentative next steps
While I am convinced that this issue is more important than widely believed and significantly neglected, I am deeply uncertain of how tractable it is. I don’t know how or when we might build factory farms in orbit and beyond, and I don’t know when or how quickly we might have to transition from a long, thoughtful exploration of this problem to developing useful interventions.
It may be too early to recommend specific actions. The s-risk posed by factory farming in space is clearly a strange niche issue for most people, and it would be counterproductive to invite further scepticism, ridicule or effective resistance from relevant industries by proposing premature interventions. That said, I would suggest the following questions that readers may wish to consider, and even try to address, as we try to understand how big this problem might be, and whether we should make it a priority.
What might the first factory farm in space be like, in engineering terms? What might this suggest about timelines for intervention?
As an example, I imagine a highly-automated disc- or ring-shaped station in low-Earth orbit, rotating on its central axis with sufficient centrifugal force to generate “spin gravity” that mimics conditions on Earth. It would probably serve humans living on space stations and on the Moon, and possibly further afield. Similar orbital factory farms might also be installed in stable orbits at LaGrange Point 4 or LaGrange Point 5 (Earth-Moon system), with less need for course-corrective fuel burns.
What commercial incentives might transformative AI create for both settlement in space, and factory farming?
Are there ways in which transformative AI might help accelerate our moral progress, negating any incentives for factory farming in space?
What would be some early indicators that factory farming in space is closer to becoming a reality? When might be the right time to act, and how?
Could genetic engineering create “mind-free” animals without any subjective experiences that could be farmed ethically? Is this an effort worth focusing on to end the suffering caused by factory farming on Earth, and forever end the possibility of it extending to space?
How does factory farming in space intersect with other futuristic proposals such as terraforming, which could introduce wild animal suffering to the cosmos?
Is it possible, in principle, to create digital minds that are free from suffering? If so, why might a much larger number of future digital minds (in expectation) still morally outweigh s-risk from factory farming in space?
I think this is probably easier to answer if one accepts prioritarian ethics, but harder if one accepts suffering-focused ethics, but I’m not sure.
I have largely relied on the work of Yip Fai Tse here, but my thinking has also been strongly influenced by Magnus Vinding, Tobias Baumann, Nick Bostrom, Everett C. Dolman, Tim Marshall, Will Macaskill, Toby Ord and Peter Singer, and probably others. Errors, in fact or in reasoning, are of course my own.