I think interventions may overwhelmingly change the welfare of soil animals and microorganisms in expectation. So I believe it is useful to imagine what to do if they were certain to be sentient, and all other potential beings, including humans, had a probability of sentience of exactly 0. These are not my actual views. I am only certain that I am sentient right now (not moments ago), and I would say even a Planck volume in deep space vacuum has a probability of sentience higher than 0. However, the hypothetical helps internalise dominant effects on soil animals and microorganisms. The goal would be increasing the welfare of these in the real world, so the views of others should be considered as they are in reality. The vast majority of people would still not care about the subjective experiences of soil animals and microorganisms.

One can increase the welfare of soil animals and microorganisms focussing on:

• Increasing the welfare per organism-year, thus improving their lives.
• Increasing the number of organisms with positive lives.
• Decreasing the number of organisms with negative lives.
• Building capacity for the above, such as by:
  • Gathering or disseminating evidence that soil animals and microorganisms are more likely to be sentient than others assume (practically 0).
  • Researching the conditions under which they have positive or negative lives.
  • Finding which interventions cost-effectively change their population.
  • Conducting other analyses which can increase funding for work on soil animals and microorganisms.

I think the focus should be on building capacity. For example, I recommend research informing how to increase the welfare of soil animals over pursuing whatever land use change interventions naively seem to achieve that the most cost-effectively. Nevertheless, such research will only be a good fit for a small fraction of people. In this post, I am asking about what you in particular would do in the hypothetical I described above.

I believe work planned with soil animals and microorganisms in mind will tend to be more cost-effective than that optimised around other target beneficiaries. Likewise, I would target (optimise for increasing the welfare of):

• Humans in low and middle income countries (LMICs) instead of humans in high income countries (HICs) to increase the welfare of humans in LMICs.
• Humans in HICs instead of humans in LMICs to increase the welfare of humans in HICs.
• Shrimps instead of chickens to increase the welfare of shrimps.
• Chickens instead of shrimps to increase the welfare of chickens.
• Dogs instead of chickens to increase the welfare of dogs.
• Chickens instead of dogs to increase the welfare of chickens.
• AI systems instead of shrimps to increase the welfare of AI systems.
• Shrimps instead of AI systems to increase the welfare of shrimps.