You could alternatively have the best guess that either:
Soil nematodes, mites, and springtails are very likely not sentient.
The welfare of nematodes, mites, and springtails is very likely of negligible importance compared to the welfare of e.g. chickens.
Soil nematodes, mites, and springtails have very close to neutral lives on average (even if the welfare of each individual is significant).
The extreme uncertainty in the sign and magnitude of their welfare effectively âcancels outâ in expectation (even if still having a large variance).. or cancels out once combined with uncertainty around whether there are other considerations missing from your analysis completely (other species we should be thinking about, possibility of sentience in bacteria under some spiritual worldview, etc).
I think under any of these points of view, even the final one, recommending organisations working on both sides of your divide would be internally consistent.
Thanks for the relevant comment, Toby! I only covered the options I consider most reasonable, but the ones you mentioned crossed my mind, and I think they are worth discussing. As far as I can tell, they can be summarised into 2 objections:
The expected welfare of soil nematodes, mites, and springtails being much closer to 0 than I estimated. This could be due to their probability of sentience being lower, their welfare range conditional on sentience being smaller, or their expected welfare conditional on a given welfare range being smaller (for example, due to the welfare distribution being closer to symmetric around 0, as you said).
The effects on soil nematodes, mites, and springtails are practically neutralised by considerations I did not cover.
On the 1st objection, ACEâs cost-effectiveness analyses rely on Ambitious Impactâs (AIMâs) suffering-adjusted days (SADs). In this system, silkworms have a welfare range of 0.46 (you can ask Vicky Cox for the private sheet with the estimates), 230 (= 0.46/â0.002) times Rethink Prioritiesâs (RPâs) mainline welfare range of silkworms of 0.002. As a result, small invertebrates have a much greater welfare range in ACEâs cost-effectiveness analyses than under RPâs mainline welfare ranges. My estimates for the welfare of soil nematodes, mites, and springtails rely on RPâs mainline welfare ranges, so I believe these animals would have a welfare much further away from 0 under AIMâs, and therefore ACEâs, assumptions about welfare ranges.
On the 2nd objection, I think it would be a surprising and suspicious convergence if considering unmodelled effects much larger than the ones currently being modelled practically did not change anything in terms of ACEâs recommendations. My analysis of effects on soil nematodes, mites, and springtails significantly changed my cause prioritisation. At the very least, I would say ACE could explain why they think effects on wild animals are not worth considering.
I would not be surprised if effects on bacteria of changing cropland were much larger than those on soil nematodes, mites, and springtails. From Table S1 of Bar-on et al. (2018), there are 10^30 terrestrial deep subsurface bacteria, 10^9 (= 10^(30 â 21)) times as many as nematodes, and I guess the welfare range of bacteria can seasily be much larger than 10^-9 that of nematodes. However, the number of bacteria per unit area is correlated with the number of soil nematodes, mites, and springtails per unit area, as both are driven by net primary production (NPP), and I would guess bacteria to have negative/âpositive lives conditional on soil nematodes, mites, and springtails having negative/âpositive lives. So I believe my conclusion that the cost-effectiveness of interventions targeting vertebrates is driven by changes in cropland would hold accounting for bacteria.
You could alternatively have the best guess that either:
Soil nematodes, mites, and springtails are very likely not sentient.
The welfare of nematodes, mites, and springtails is very likely of negligible importance compared to the welfare of e.g. chickens.
Soil nematodes, mites, and springtails have very close to neutral lives on average (even if the welfare of each individual is significant).
The extreme uncertainty in the sign and magnitude of their welfare effectively âcancels outâ in expectation (even if still having a large variance).. or cancels out once combined with uncertainty around whether there are other considerations missing from your analysis completely (other species we should be thinking about, possibility of sentience in bacteria under some spiritual worldview, etc).
I think under any of these points of view, even the final one, recommending organisations working on both sides of your divide would be internally consistent.
Thanks for the relevant comment, Toby! I only covered the options I consider most reasonable, but the ones you mentioned crossed my mind, and I think they are worth discussing. As far as I can tell, they can be summarised into 2 objections:
The expected welfare of soil nematodes, mites, and springtails being much closer to 0 than I estimated. This could be due to their probability of sentience being lower, their welfare range conditional on sentience being smaller, or their expected welfare conditional on a given welfare range being smaller (for example, due to the welfare distribution being closer to symmetric around 0, as you said).
The effects on soil nematodes, mites, and springtails are practically neutralised by considerations I did not cover.
On the 1st objection, ACEâs cost-effectiveness analyses rely on Ambitious Impactâs (AIMâs) suffering-adjusted days (SADs). In this system, silkworms have a welfare range of 0.46 (you can ask Vicky Cox for the private sheet with the estimates), 230 (= 0.46/â0.002) times Rethink Prioritiesâs (RPâs) mainline welfare range of silkworms of 0.002. As a result, small invertebrates have a much greater welfare range in ACEâs cost-effectiveness analyses than under RPâs mainline welfare ranges. My estimates for the welfare of soil nematodes, mites, and springtails rely on RPâs mainline welfare ranges, so I believe these animals would have a welfare much further away from 0 under AIMâs, and therefore ACEâs, assumptions about welfare ranges.
On the 2nd objection, I think it would be a surprising and suspicious convergence if considering unmodelled effects much larger than the ones currently being modelled practically did not change anything in terms of ACEâs recommendations. My analysis of effects on soil nematodes, mites, and springtails significantly changed my cause prioritisation. At the very least, I would say ACE could explain why they think effects on wild animals are not worth considering.
I am wary of causing large harm to soil nematodes, mites, and spirngtails in the hope that other unmodelled undescribed effect will neutralise it. I estimated School Plates in 2023 increased 1.20 billion wild-animal-years (mostly nematode-years) per $, and nematodes seem to have pretty painful experiences. From FĂ©lix and Braendle (2010), âFrequently co-occurring predators [of nematodes] include fungi, which, depending on the species, invade the nematode through spores attaching to the cuticle or the intestine, or use trapping devices that immobilize the animal and perforate itâ. From FrĂ©zal and FĂ©lix (2015), âparasites infect their host via the two most exposed parts of the nematode, the cuticle and the intestine. Some non-invasive bacteria form a biofilm along the nematodeâs cuticle or directly stick to it (Hodgkin et al., 2013). Other bacteria proliferate in the nematode gut, which may induce constipation and likely impairs nutrient uptake (FĂ©lix and Duveau, 2012). The most intrusive parasites enter and proliferate inside the nematode body. Some pierce the cuticle (e.g., Drechmeria coniospora [Couillault et al., 2004], Figure 2J), while others enter intestinal cells via the apical membrane (e.g., microsporidia and Orsay virus [Troemel et al., 2008; FĂ©lix et al., 2011])â.
I would not be surprised if effects on bacteria of changing cropland were much larger than those on soil nematodes, mites, and springtails. From Table S1 of Bar-on et al. (2018), there are 10^30 terrestrial deep subsurface bacteria, 10^9 (= 10^(30 â 21)) times as many as nematodes, and I guess the welfare range of bacteria can seasily be much larger than 10^-9 that of nematodes. However, the number of bacteria per unit area is correlated with the number of soil nematodes, mites, and springtails per unit area, as both are driven by net primary production (NPP), and I would guess bacteria to have negative/âpositive lives conditional on soil nematodes, mites, and springtails having negative/âpositive lives. So I believe my conclusion that the cost-effectiveness of interventions targeting vertebrates is driven by changes in cropland would hold accounting for bacteria.