As a spin-off off my research ideas on prioritization, here more object-level research ideas for wild animal suffering alias welfare biology.

If you’re interested in researching one of these questions, please leave a comment or ask me to edit the post so we don’t duplicate effort.

Taxonomy of resilience of wild animal welfare improvements after civilizational collapse


  1. Which interventions against WAS are most permanent even without sustained human effort, especially in the case of civilizational collapse?
  2. Are there downsides to focusing completely on such interventions, especially with regard to option value?


Interventions (1) whose effects subside when they are not actively carried out anymore, (2) whose effects may be undone by a catastrophe, or (3) that are premised on structures that may be destroyed by a catastrophe are unlikely to be cost-effective compared to interventions that reduce more event-like suffering risks because interventions that are not robustly self-sustaining will have an effect over only a very short period of time in expectation. (Some people put the probability of an existential catastrophe before 2100 at 19%.)

My intuition is that this (the inferior cost-effectiveness) will remain true even if we find that “shmingletons” are relatively volatile after all or that the tractability of such risk reduction efforts is rather low. In combination, these may be stronger challenges. The greatest single challenge I’m aware of stems from the simulation hypothesis because most simulations are likely to be short-lived.

These three categories may not be the only failure modes, and there may also be different paths to success, so it would be helpful to create an exhaustive taxonomy of interventions that classifies them according to their cost-effectiveness based only on this one important factor.

A consideration that could be taken into account that pushes in the opposite direction is option value. If we build highly resilient systems not even we may be able to change them later on.


The resilience of the positive change of an intervention should be one major criterion for at least shallow analyses of its cost-effectiveness. Such a rubic could speed up the assessment and help us avoid overlooking important factors.

Invertebrate deaths per crop yield


  1. Are greenhouses only suitable or economically viable for some vegetable species but very low in suffering?
  2. Which crops promise the greatest absolute suffering reduction given current farming methods and optimal, economically viable methods?


Even though the net primary productivity (NPP) of cultivated land is not maximal compared to plausible counterfactuals such as forests (greater NPP) or grassland (similar NPP), the expected insect suffering caused by crop farming may be many times as great as that in forests because the animals attracted by the plants are continually killing off by farmers before they can influence the NPP of the land, so that it remains to attract more animals.

This may be different in organic farming systems that are limited in terms of the insecticides that they can use, but they may just use animals that attack the animals that try to feed on the crops, so that the effect on the aggregate suffering may be the same. If the yield is lower but demand is unchanged, the aggregate suffering may even be greater.

I’m more optimistic that this is different in greenhouse systems that physically shield the crop NPP from any animals. Insofar as the shielding is effective, this approach should make the NPP virtually disappear (ignoring the waste products that its human consumers will leave behind). Such systems can also be highly efficient and economically viable.

These system may only be suitable or economically viable for some species of plants, so it’ll be important to determine what the counterfactual is for the production of these particular species, for which species the potential for suffering reduction is greatest, and how this compares to other interventions to reduce insect suffering such as malaria prevention or buying grassland to cover it with gravel.


This area has potential for social enterprises, e.g., for-profits porting technologies that have been proven to be economically viable in the Netherlands or Japan to other markets. A for-profit could also produce products needed by companies implementing the most impactful greenhouse system if that approach is easier to bootstrap.

Maximum amount of exoplanet surface area reachable over time


  1. What formula best describes the volume of the three-dimensional shape of the expansion of earth-originating life into space over shorter timelines – thousands or tens of thousands of years rather than billions?
  2. What formula best describes the energy output per time of the stars in that volume?


I model the expansion into space as spherical or at least as third-degree polynomial on the basis that space is homogeneous at the largest scale. (See also this image.) But that only starts “at scales between 30 and 200 megaparsecs,” or 100–700 million lightyears, so 50–350 million lightyears in either direction. Assuming we can’t travel quite at light speed, it may be a billion years before the approximation becomes quite justified. (Maybe our local group or supercluster are already homogenous enough inside, but there we’re already at a scale of millions of lightyears.)

My observations, even the most optimistic ones, don’t promise that we can influence much more than the next 10,000 years. These observations may be flawed because they can’t observe any counterfactuals, but combined with the probably common intuition that influences don’t last forever, it may be valuable to have a better idea of the shape of earth-originating life in the universe over just the next 10,000s of years.

Additionally, whole-brain emulation may make biological life redundant at some point, so that environments in space will cease to be optimized for it, so that unintentional suffering will be minimized anyway at some point. This may happen even sooner than 10,000 years from now.


If our expansion into space will be much more circular than spherical or have some other shape with orders of magnitude smaller volume, then efforts to influence the aggregate suffering from space colonization will be radically less important than it currently seems. Those are some of the most effective interventions in the wild animals suffering space, indicating that this is an important consideration.

But this consideration is likely to be irrelevant in cases where space colonization is not run by biological life (either because of whole-brain emulation or because we’ve been superseded), which seems increasingly likely the further technology advances.

A related consideration may remain relevant, namely that the suffering of silicon-based beings will be upper-bounded by the available energy, which will be correlated with the number of solar system reachable over time. The expansion model could cover this case as well by including the energy output of the reachable suns.

Accidental Exodus: Will it be simple not to inadvertently export invertebrates into space


  1. Is it hard to avoid accidentally carrying invertebrates (such as small insects or their eggs) into space? How hard is it?
  2. What animals can survive and procreate in environments that exist aboard uncrewed spacecrafts?


Michael Dello-Iacovo has written about the risks from intentionally and accidentally exporting nonhuman animals to space in the course of space colonization (with a focus on terraforming, which may not be necessary in the long run, e.g., if we move to real or simulated environments in Dyson spheres).

It seems to be hard but important to keep bacteria out of spacecrafts, and if there are astronauts on board, it may get even more difficult to keep life out, perhaps even insects or their eggs.

But how hard is it exactly? Once a large exodus from earth starts, even a small risk may lead to lots of insects being exported and then multiplying at high rates.


If animal farming is superseded by a more efficient technology and we don’t prioritize terraforming, or if we move away from biological life entirely, the remaining risk to wild animals (that I’m aware of) is from being exported into space by accident. This scenario is not as conjunctive as that sentence may imply as it is an additional consideration in all other cases as well.

If these animals are unlikely to get onto spacecrafts or unlikely to survive in space, we can be somewhat less concerned about space colonization.

Are suffering levels resilient?


  1. Is it plausible that there are any feedback mechanisms that increase suffering again, be it indirectly, if we reduce it or reduce it the wrong way?


If we want to reduce insect suffering, we might think that we should euthanize insects (regardless of actually relevant factors such as whether they’re about to lay eggs or are in great pain), but if their population is bottlenecked by their sustenance, we just free up edible biomass to be eaten by another insect some generations later that wouldn’t have existed otherwise. (This relationship may be less than one-to-one.)

Similarly, subsidies may lead to export of the subsidized product, and thus to market distortions and costs disproportionate to the benefits.


I’m worried that there may be some mechanisms by which efforts to minimizes suffering may be systematically less effective than we think because they (the mechanisms) respond to the existing level of suffering or its rate of increase or decrease. This would make these efforts at suffering minimization greatly less tractable.

For example, per individual, there are probably adaptive mechanisms whereby (1) they adapt to a safer environment by becoming more sensitive within their lifetime or (2) they adapt to a safer environment by becoming more sensitive evolutionarily because more sensitive individuals can perform better within their culture.

But unless these changes span orders of magnitude of individual suffering over ones lifetime, it may be more relevant to consider population-level dynamics, e.g., are there any mechanisms that cause the number of animals to increase more quickly the lower it is?

Academic Movement Building: Publishing Sweet Spots


  1. What attributes make species, topics, and attributes (such as welfare) interesting for researchers?
  2. What makes a species or topics are interesting for welfare biology?
  3. Where are the sweet spots where a hypothetically easy-to-study species that is highly relevant for welfare biology has remained understudied?


To build more momentum for welfare biology research, it may be important to understand what makes a species and behaviors of a species attractive for academic researchers. (E.g., what makes drosophila so popular?)

Then we can determine what the main criteria are that make a species or topic interesting for welfare biology – e.g., being r-selected, likely to become more frequent due to global warming, hard to keep out of space missions, short lives and violent deaths, and reinforcement learning behavior.

Finally, we can put it all together and look for publishing sweet spots where a hypothetically easy-to-study species that is highly relevant for welfare biology has remained understudied.


Incentivizing research in welfare biology has been a priority of several charities in the space, and for good reasons, I think, because of my considerations surrounding the robustness of research impact.


comments powered by Disqus