Epistemic status: Speculation from two decently informed advocates armed with anecdata.
Note on process: After having some version of this conversation several times and saying, “we should probably write about this publicly,” we took the less heroic route: we recorded one of our conversations, fed the transcript into an LLM, and then substantially revised the structure, substance, and framing ourselves. We will not be sharing the transcript, as it is interspersed with side comments about dogs.
We’ve been very slow to get this out. We wrote this and began receiving feedback on our draft before very relevant projects went live, including the announcement of Spring Innovation Fund and Rethink Priorities’ call for new animal founders.
Many thanks to those who reviewed our drafts. All mistakes are our own.
Summary
The animal movement is at an extremely exciting stage, with many promising new ideas, funding opportunities, and talented individuals entering the space. We want to encourage this energy, while also noting a concerning pattern that we’ve observed:
The effective animal advocacy world has a pattern of high levels of enthusiasm for a specific intervention that eventually gives way to a rapid downward adjustment, as expectations run ahead of the evidence and practical reality. We’ve seen this with leafleting, cultivated meat, alternative proteins, corporate campaigning (to a lesser extent), and, most recently, electrical stunning of shrimps. We describe these hype cycles as “bubbles.”
We worry “welfare tech” (also sometimes referred to as “humane tech”)—a fuzzy term referring to technologies intended to measure, reduce, or eliminate animal suffering in systems of animal use (such as electrical stunners, in-ovo sexing, immunocastration, or AI-enabled higher-welfare farming)—may be entering a similar dynamic.
Our worry is not that welfare tech is a bad idea (some of it is likely excellent). Rather, the worry is that it has several traits that make it vulnerable to becoming a bubble. It has a clean theory of change, no obvious need for mass public persuasion, compatibility with short AI timelines, appeal to tech-oriented funders and builders, and the ability to be framed as a silver bullet: build the tool, deploy the tool, reduce suffering at scale.
The actual path from “promising technical idea” to “large-scale suffering reduction” is likely to be slower, messier, more iterative, and more institutionally dependent than that story suggests. Bubbles risk both over-investment in the short-term and under-investment in the long-term.
Welfare tech should not be dismissed, but expectations should be grounded; we should be careful to understand what needs more research vs. what is ready for deployment. We should calibrate ourselves to the realistic time, effort, and cost of turning good ideas into functional reality. Finally, we should try to avoid pulling resources away from long-term movement capacity/ecology efforts.
The recurring pattern
A pattern in the EA animal world goes something like this:
- A new intervention appears to have an unusually high expected impact.
- The movement begins treating it less like “one promising bet” and more like “the thing we should all focus on.” This may lead to underinvestment elsewhere.
- Eventually, the evidence, implementation challenges, or timelines look less favorable than the initial narrative.
- The bubble bursts, sometimes leading to heavy disillusionment and diminished support for the slower but still important work in that area.
In general, we would expect that the best interventions may well have rapid initial advancement and subsequent slowdown as new hurdles emerge. However, the bubble can make that standard trajectory look like a disappointment or outright failure.
Some examples of interventions that saw their bubble burst:
Leafleting was perhaps the original example. In the early-to-mid 2010s, it was often treated as one of the most effective things people could do for animals, based largely on evidence that later looked much weaker than initially believed. The story was wonderfully legible: distribute leaflets, convert people to vegetarianism, reduce animal suffering. Simple, measurable, scalable. However, the reality was messier. The foundational studies were methodologically flawed (e.g., self-reporting on diets leading to social-desirability bias) and long-term effects were unreliably extrapolated. When these flaws were found, there was a rapid downward adjustment, and now leafleting has effectively been fully removed from EA animal work.
Cultivated meat and alternative proteins followed a different but related pattern. They promised a route to animal suffering reduction that did not require persuading billions of people to care more about animals. If the product became cheaper and tastier, maybe the market would do the work. However, early timelines were often wildly optimistic, and the story underweighted the social and psychological dimensions of changing how people eat. Clean meat remains a plausible long-run theory of change, but is now often absent from animal movement conversations.
Corporate campaigning is more complicated, and more instructive. It had early wins, which helped validate the theory of change, attract talent, and build institutional knowledge. Later, when progress on harder asks like the Better Chicken Commitment felt slower or more difficult, there was a wave of disillusionment. The field was, however, able to readjust expectations and continue progressing.
Welfare tech is already seeing this cycle with stunning equipment. Some considered Shrimp Welfare Project’s stunning work as perhaps the most impactful program for farmed animals to date (even by many orders of magnitude). However, this has recently turned to disillusionment for many, as the June 2026 post “animal welfare has an evidence problem” outlined some of the key uncertainties around current stunning equipment’s efficacy. However, among those most involved (including SWP), the concerns of efficacy have been well-known and a point of active effort for multiple years, pointing to a misalignment between the hype-heavy discourse and the known realities of those in the field.
Welfare tech seems liable to further fall within this pattern.
Why EA is vulnerable to silver bullets
EA is, for good reasons, attracted to heavy-tailed impact. We look for unusually cost-effective interventions, and are willing to shift resources when the evidence suggests we should.
These are virtues, but they can create binary thinking. Something is either “one of the top things” or “not worth doing.”
The most hype-prone interventions tend to be easy to explain, have a clean theory of change, seem measurable, and appear to scale without requiring deep public buy-in. They make hard social and political problems look like technical or logistical ones:
- Leafletting can make dietary behaviour change seem like a logistical issue of best distributing the most compelling information.
- Cultivated meat and alt-proteins can make dietary behaviour change seem like a technological problem of mimicking animal products.
- Corporate campaigning can make large-scale industry reform seem like an issue of targeting a few businesses and executives.
- Shrimp stunners can make reducing shrimp suffering seem like an issue of distributing existing technology.
This connects to what Tom has called EA’s “last-mile delivery bias.” EA is often most comfortable when the solution basically exists, and the remaining challenge is distribution: buy the bednets, deliver the vaccines, implement the proven intervention.
These opportunities are real and important. But in animal advocacy, true last-mile delivery problems may be rarer than we want them to be. We are often much earlier in the process: still figuring out how animals are suffering, which interventions actually help, and how those interventions play out in real-world conditions.
When we mistake an early-stage research or field-building problem for a last-mile delivery problem, we set ourselves up for disappointment.
Cultivated meat was, for years, implicitly treated as if it was close to a last-mile problem: a few years from cost parity, just needing scale and distribution. It wasn’t. Something similar could happen with welfare tech.
Why welfare tech might be the next “silver bullet”
Welfare tech is a fuzzy term, which we roughly define as technology that either directly improves welfare (e.g. stunning equipment or genetic interventions that reduce animals’ capacity to suffer) or our ability to measure welfare (e.g. AI-assisted welfare monitoring).
Welfare tech has a lot going for it. It seems practical. It seems compatible with existing industry incentives. It does not require consumers to change their behavior. It appeals to engineers and technical founders. It can attract funding from people more excited by technological innovation than by movement building or policy advocacy. If successfully deployed, it could reduce suffering for very large numbers of animals.
It is also emotionally and rhetorically powerful. “We can build tools that reduce suffering directly” is a compelling message. Compared to decades-long fights over law, culture, and political power, welfare tech can feel refreshingly concrete.
In a world where many EAs have short AI timelines, welfare tech may look especially attractive. If someone believes we have only a few years before transformative AI changes everything, they may be less excited by slow institution-building and more drawn to interventions that could plausibly improve animals’ lives quickly.
But this is exactly where caution is needed. A technology can be conceptually simple, but still be practically immature and susceptible to complication when hit by the real world.
This is all exacerbated by the current influx of tech-oriented, AI-adjacent funders, who are perhaps more likely to find welfare tech exciting. Organizations may feel pressure to reshape their plans around what sounds fundable at the expense of building for the long-term.
Why are bubbles bad?
By an “impact bubble,” we mean a situation where expected impact in a given timeframe substantially exceeds realistic impact. The problem is not necessarily that the underlying idea is bad, but that expectations outrun readiness. This can hurt the movement in several ways.
First, it can lead to overinvestment in short-term deployment and underinvestment in the slower work needed to make the intervention real. For welfare tech, the highest-value work right now may often be research infrastructure, welfare measurement, validation, field testing, or understanding what adoption would actually require — not immediately buying equipment or launching aggressive scale-up campaigns.
Second, hype can crowd out long-game strategies. Ultimately, the pro-animal movement needs to build a broad strategy for reducing animals’ vulnerability, building institutional power, and shrinking exploitative industries. To achieve this, we need a diverse ecosystem. We need some people making near-term welfare gains and others working on long-term structural change. However, welfare tech becomes attractive precisely because it does not challenge the system very much. The concern with hype cycles is that it can push focus too hard into single tracks that make the whole movement ecology become less resilient.
Third, hype that is not delivered on can create backlash. If people expect rapid, large-scale wins and instead encounter messy adoption barriers, ambiguous welfare effects, and slow timelines, they may sour on the whole category, including genuinely valuable long-term work within it.
What would a healthier approach look like?
We would suggest:
Separate research bottlenecks from deployment opportunities. Some technologies are not ready for broad rollout. The best next step may be better welfare measurement, validation, field testing, or simply understanding what would need to be true for adoption.
Share an explicit (ideally evidence-based) timeline. Is the claim that this will help animals within one year, five years, twenty years, or eventually? A project can be worthwhile on a long timeline while being overhyped on a short one.
Ask what the intervention might crowd out. Welfare tech funded by new technical donors who would not otherwise support animal advocacy is very different from welfare tech pulling animal movement resources away from other work. We should be careful to consider the value of sustained attention on long-term strategies and be very cautious when directing funds away from them and towards potential bubbles.
Evaluate second-order effects. Does a technology entrench animal industries or create leverage over them? Does it reduce the number of animals used, or could it inadvertently increase them? Does it help build a stronger movement or just solve one isolated problem?
Avoid binary updates. If welfare tech disappoints over the next few years, that should not necessarily mean “welfare tech was a mistake.” It may mean we misunderstood the stage of development. Likewise, if one project succeeds, that does not mean the whole category deserves a massive influx of support.
Conclusion
We’re both deeply committed to the animal movement and excited for more funding to enter the space, more motivated individuals to work for animals, and more organizations to start or scale.
Welfare tech may be an important part of the future of animal advocacy. Some projects in this space could reduce enormous amounts of suffering. We are glad people are exploring it. But we worry the movement is vulnerable to turning welfare tech into the next silver bullet: a broad, exciting, underdefined category that attracts more confidence than the evidence can yet support. This, in turn, is liable to become a bubble.
The lesson from past animal advocacy bubbles is not “don’t pursue promising new interventions.” It is: don’t confuse a promising bet with a solved problem. And don’t let clean theories of change crowd out messy but necessary long-term work. Don’t build a movement monoculture around whatever currently feels most exciting.
I don't really understand this post; I suspect the LLM has made it a little confused.
You say:
> When we mistake an early-stage research or field-building problem for a last-mile delivery problem, we set ourselves up for disappointment.
Cultivated meat was, for years, implicitly treated as if it was close to a last-mile problem: a few years from cost parity, just needing scale and distribution. It wasn’t. Something similar could happen with welfare tech.
Isn't most of the focus already on development?
> Leafleting was perhaps the original example. In the early-to-mid 2010s, it was often treated as one of the most effective things people could do for animals, based largely on evidence that later looked much weaker than initially believed. The story was wonderfully legible: distribute leaflets, convert people to vegetarianism, reduce animal suffering. Simple, measurable, scalable. However, the reality was messier. The foundational studies were methodologically flawed (e.g., self-reporting on diets leading to social-desirability bias) and long-term effects were unreliably extrapolated. When these flaws were found, there was a rapid downward adjustment, and now leafleting has effectively been fully removed from EA animal work.
I don't see why this was named as a "bubble", my impression is that this was basically good. People thought leafletting was good; people tried leafletting; it wasn't very good; people stopped doing it. That seems very functional to me.
My impression is that bubbles are bad because they are ex-post indications of overcorrections (on entry or exit or both) that have a bunch of harmful externalities - but I think one needs to make an actual argument at the object level for why there could be an overcorrection towards welfare tech rn (e.g. maybe you think that too much funding is going into it too quickly, but are you sure that money would actually be counterfactually spent on other animal stuff? Or maybe you think that people are moving away from other interventions to work on this stuff, but the main people I know doing AW tech dev are unsurprisingly technical people and weren't working on other AW interventions before that, so there's not much of a tradeoff there).