Summary
- I think organisations using Rethink Priorities’s (RP’s) mainline welfare ranges, at least Ambitious Impact (AIM), Animal Charity Evaluators (ACE), the Animal Welfare Fund (AWF), and RP, should consider effects on soil nematodes, mites, and springtails. I believe these are the driver of the overall effects of the vast majority of interventions.
- Effects on soil nematodes should not be neglected just because they may be beneficial or harmful. Mathematically speaking, “number which is certain to be barely positive” + “number which can be very negative, barely negative, barely positive, or very positive” = “number which can be very negative, barely negative, barely positive, or very positive”, which does not have to be barely positive in expectation. I would only disregard the effects on soil animals if I considered them much smaller in expectation than those being considered.
- Estimates of the expected effects on soil nematodes, mites, and springtails cannot be taken literally, but should still be considered. One should update less on uncertain evidence, but more given an uncertain prior.
- Many argue effects on soil nematodes, mites, and springtails should only be considered once society cares much more about animals than now. I suspect this results from implicitly supposing that helping animals more inside society’s moral circle, like farmed vertebrates, is more cost-effective. In contrast, I recommend supporting the interventions which at the margin increase welfare the most per $, even if to increase the welfare of animals which are highly neglected in society.
I think organisations using RP’s mainline welfare ranges should consider effects on soil nematodes, mites, and springtails
I think organisations using RP’s mainline welfare ranges, at least AIM, ACE, AWF, and RP, should consider effects on soil nematodes, mites, and springtails. If they are not planning to consider these effects, I think they should explain why not for reasoning transparency. I believe these are the driver of the overall effects of the vast majority of interventions. I estimate the following based on welfare ranges for those soil animals derived from RP’s mainline welfare ranges:
- Cage-free corporate campaigns benefit those soil animals 28.2 times as much as they benefit chickens.
- Broiler welfare corporate campaigns benefit those soil animals 444 times as much as they benefit chickens.
- Veganuary in 2024 harmed those soil nematodes 3.58 k times as much as they benefited farmed animals.
- School Plates in 2023 harmed those soil nematodes 5.42 k times as much as they benefited farmed animals.
- GiveWell’s top charities benefit those soil animals 87.6 k times as much as they benefit humans.
- Buying beef benefits those soil animals 1.82 M times as much as it benefits cows.
I also estimate feed crops replacing temperate grasslands, savannas, and shrublands increases the welfare of soil nematodes, mites, and springtails 9.39 (shrimp) to 11.7 M (dairy milk) times as much as it changes the welfare of directly affected animals.
I believe any organisation aiming to increase animal welfare should consider the effects on soil animals. I am focussing on the ones using RP’s mainline welfare ranges given the apparent contradiction between this, and neglecting effects on soil animals. In addition, neglecting effects on soil animals increases the probability of causing harm, so organisations which intrinsically value decreasing this, instead of just maximising expected impact, have an additional reason to consider them.
AIM may be the most open to considering these. Joey Savoie, AIM’s CEO, said on 15 May 2023 they “consider cross-cause effects in all the interventions we consider/recommend, including possible animal effects and WAS [wild animal suffering] effects”. However, none of AIM’s public reports covers effects on wild animal welfare of interventions targeting humans, or farmed animals.
ACE commented on 25 June 2025 they “intend to publish a blog post on the consequences of farmed animal welfare interventions for wild animals, after the busy work of charity evaluations is wrapped up for the season”.
Effects on soil nematodes should not be neglected just because they are unlikely to be sentient
I suspect many are only willing to account for effects on beings which are sufficiently likely to be sentient, even if the effects on them are very large in expectation due to them being very numerous. It is as if probabilities of sentience below a minimum arbitrary threshold are rounded to 0. Even Peter Singer seems to have too much binary thinking about which animals are considered. Peter seemingly advocates much more for vertebrates than invertebrates, and said in 2023 that “a reasonable place to draw the line is to say that there are some invertebrates that can feel pain”. I do not think one should be drawing lines defining a moral circle. Sentience, and welfare per animal-year are probabilistic, and this has to be multiplied by the number of individuals to get their total welfare. Invertebrates are less likely to be sentient, and have a welfare per animal-year closer to 0 than vertebrates, but there are many more of them. Rounding to 0 a probability of sentience, or welfare per animal-year close to 0 introduces an infinite amount of scope insensitivity. Regardless of the number of beings affected, the change in their welfare will be estimated to be exactly 0.
Furthermore, the probability of sentience of soil nematodes, mites, and springtails is not anything close to Pascalianly low. RP got a probability of sentience for nematodes, which have the least neurons among those soil animals, of 6.8 % under their preferred “#1 High-Value Proxies Model”, and it only ranged from 6.8 % to 6.9 % across the 5 models they considered, although I assume these are far from independent given the proximity of the estimates. I suspect the actual probability of sentience of nematodes is higher. In RP’s words, “Assigning proxies labeled “Unknown” zero probability of being present is certainly leading to underestimates of the welfare ranges and probabilities of sentience [all else equal]”. The probability of dying in a car crash is around 2.70*10^-9 per km (= 10^-6/370), and many people still consider it reasonable to fasten seat belts for increased safety on short trips, even if they would prefer it to be optional.
Effects on soil nematodes should not be neglected just because they may be beneficial or harmful
I expect the interventions I mentioned above to benefit soil nematodes for my best guess that they decrease their population, and that soil nematodes have negative lives. Nevertheless, I got a probability for this of 58.7 %, so I am uncertain about whether the effects on soil nematodes are beneficial or harmful. However, they should not be neglected just because of this. Consider an intervention aiming to decrease the consumption of animal-based foods which:
- Increases the welfare of farmed animals, the target beneficiaries, by 1 QALY with 100 % probability.
- Decreases the welfare of soil nematodes by 1 kQALY with 30 % probability, and by 0.001 QALY with 30 % probability.
- Increases the welfare of soil nematodes by 0.001 QALY with 20 % probability, and by 1 kQALY with 20 % probability.
There is lots of uncertainty about whether the effects on soil nematodes are very negative, barely negative, barely positive, or very positive, but I would not neglect them. They decrease welfare by 100 QALY (= 0.3*(1*10^3 + 0.001) - 0.2*(1*10^3 + 0.001)) in expectation, and therefore the intervention decreases welfare by 99.0 QALY (= 100 - 1*1) in expectation, thus being harmful.
You may well disagree with my numbers above. However, mathematically speaking, “number which is certain to be barely positive” + “number which can be very negative, barely negative, barely positive, or very positive” = “number which can be very negative, barely negative, barely positive, or very positive”, which does not have to be barely positive in expectation. I would only disregard the effects on soil animals if I considered them much smaller in expectation than those being considered.
Estimates of the expected effects on soil nematodes, mites, and springtails cannot be taken literally, but should still be considered
Joey said the following about my estimates of the effects on soil nematodes, mites, and springtails.
Hey Vasco, At a certain level of robustness, I do not take CEAs [cost-effectiveness analyses] as sufficient evidence to update on, and these estimates do not pass that bar. This post (https://blog.givewell.org/2011/08/18/why-we-cant-take-expected-value-estimates-literally-even-when-theyre-unbiased/ ) is the best articulation of how I think about evidence.
I also think about evidence as described in that post from Holden Karnofsky. Here is the summary of it I published in April 2022. I agree on not updating all the way from one’s prior estimate of the expected value to the new one, and on updating less on more uncertain evidence. Yet, as implied by inverse-variance weighting used in meta-analyses, one should also update more given a more uncertain prior. I believe any reasonable prior estimate of the effects on soil nematodes, mites, and springtails will be highly uncertain. So I maintain estimates of the expected effects like mine should still be considered.
I suspect many are misinterpreting Holden’s post due to conflating the prior about the effects on humans with the prior about the effects on all potentially sentient beings. Holden rejects impartiality, which means 1 unit of welfare is always worth the same, even in principle, so he can consider a more certain prior which neglects some effects. I would say one should fully endorse impartiality at least in principle, and therefore consider the prior about the effects on all potentially sentient beings. This prior is much more uncertain than the one about the effects on humans, thus enabling much larger updates towards new estimates of the expected effects on soil nematodes, mites, and springtails.
Effects on soil nematodes, mites, and springtails should be considered even if they are very neglected in society
Many argue effects on soil nematodes, mites, and springtails should only be considered once society cares much more about animals than now. I suspect this results from implicitly supposing that helping animals more inside society’s moral circle, like farmed vertebrates, is more cost-effective. In contrast, I recommend supporting the interventions which at the margin increase welfare the most per $, even if to increase the welfare of animals which are highly neglected in society.
Moreover, some of the most cost-effective ways of increasing human welfare, like GiveWell’s top charities, target people in low income countries, and I estimate they are also among the most cost-effective interventions accounting for effects on soil nematodes, mites, and springtails. My best guess is that these have negative lives, and decreasing human mortality without decreasing food consumption increases cropland, thus decreasing the population of those animals. Humans in low income countries are more inside society’s moral circle than animals, but are still widely neglected. So I wonder what would be the case for prioritising less neglected animals over humans in low income countries. Likewise, I would like to know why the same arguments do not imply prioritising farmed vertebrates over farmed invertebrates, or pets over farmed vertebrates. I suspect many bring up cost-effective moral circle expansion as a justification for why helping their target animals is very cost-effective even accounting for all animals although this did not factor into their initial reasons.
Acknowledgements
Thanks to Vicky Cox for feedback on the draft, and to Zuzana Sperlova for confirming the cost-effectiveness analyses ACE will do this year will rely on RP’s mainline welfare ranges. I listed the names alphabetically. The views expressed in the post are mine.
Thanks to everyone for the discussion here. A few replies to different strands.
First, I agree with Vasco that transparency matters. However, transparency isn’t the only good—and, unfortunately, it often competes with others. (Time is limited. Optics are complicated. Etc.) So, by Vasco's own lights, it's only plausible that organizations should devote scarce resources to answering this particular cause prioritization question—and then post their answer publicly on the Forum—if they think (or should think) that the expected value of so doing is positive. It isn’t obvious that anyone in these organizations thinks (or should think) that’s true.
Second, you can use our work on welfare ranges without buying into naive expected utility maximization. I assume that many people who use our welfare ranges are averse to being mugged and, as a result, adopt one of the many strategies for avoiding that outcome. So, it can be true that (a) the expected value of impacts on some group of animals is very large in expectation and (b) you aren’t rationally required, by your own lights, to care much about that fact (and, by extension, investigate it in depth or engage on it publicly).
Third, our models have a narrow theoretical and pragmatic purpose: we wanted to improve the community’s thinking about cause prioritization regarding a group of animals where we took there to be good evidence of sentience. We don’t think you can take our models and apply them generally, nor do we think you can ignore the specific purpose for which they were developed. Put differently, once some animals have crossed some threshold of plausibility for sentience, we support using our models with trepidation, largely because we don't have better options. But you shouldn't apply the model beyond that and, if you have any other principled ways to make decisions, that's probably better. (Principled: “We think that any theory of change for the smallest animals begins with key victories for larger animals.” Unprincipled: “We don’t like thinking about the smallest animals.”)
Fourth, we disagree with @NickLaing characterization of the Moral Weight Project as stacking the deck in favor of high welfare range estimates. There are two reasons why. One of them is that the MWP does not say, “Sum the number of proxies found for a species and divide by the total number of proxies to get the welfare range.” If that were true, then the number of proxies would straightforwardly determine the maximum difference in welfare ranges. But that isn't correct. We have models (like the cubic model) where you need to have lots of proxies before you have a "highish" welfare range. However, we have lots of models, with uncertainty across them. Predictably, then, more moderate estimates emerge rather than any extreme (whether high or low). Someone is free to say: "A better methodology wouldn't have been so uncertain about the models; it would have just included animal-unfriendly options." That's clearly tendentious, though, and we think we made the right call in including a wider range of theoretical options. That being said, we’ll reiterate that those who are interested in the details of the project should examine the particulars of each model and its conclusions rather than just taking the overall estimates straightforwardly. You can find each model’s results here.
The second reason we disagree with Nick’s characterization of the MWP is that, even if you isolate a particular model, you don’t automatically get high welfare ranges. Suppose, for instance, that there are 80 proxies total and that a model uses them all. If there were N that were as simple as "any pain-averse behavior," then, for the core models of the MWP, saying "likely yes" to each of them would give you a sentience-conditioned welfare score of 0.875*N/80 on average. We didn't consider animals as simple as nematodes in the MWP because we didn't think that the methods were robust for that type of animal. (See above.) But say you think there's a 0.5% chance of sentience for nematodes. Then, the sentience-conditioned welfare range would have been approximately 0.005*0.875*N/80. If the average model had 5 proxies that are as simple as "any pain averse behavior" and we gave "likely yes" to nematodes on all five, that would generate a mean welfare range of 0.005*0.875*5/80 = 0.00027. Again, we don’t endorse using the MWP for animals with that small of a probability of sentience, but 0.00027 isn’t a particularly high welfare range. (And as we've said many times, we’re just talking about hedonic capacity, not “all things considered moral weights,” which don’t assume hedonism. That number would be lower still.)
Should we find funding for a second version of the project, we’re likely to take a different approach to aggregating the proxies to produce welfare ranges, aggregating welfare ranges across models, and communicating the results. Still, we hope the first version of MWP contributes to more informed and systematic thinking about how to prioritize among different interventions.
Thanks for the comment, Bob!
I agree transparency is not the only good. However, I think there is a high bar for not commenting on effects which in expectation seem way larger than the effects being covered.
I am not sure what you mean by "naive expected utility maximisation". I agree my analyses of the effects on soil nematodes, mites, and springtails have lots of room for improvement, but at least I am trying to consider these effects instead of assuming they do not change prioritisation even if they look much larger in expectation than the effects on the target beneficiaries.
I fully endorse expectational total hedonistic utilitarianism (ETHU), but I think this is far from required for effects on soil nematodes, mites, and springtails to matter a lot. These are much larger than the effects on the target beneficiaries under RP's mainline welfare ranges, so they would still be driver of the overall effect even putting just 10 % weight on ETHU.
What is so different between RP's probabilities of sentience of nematodes and silkworms of 6.8 % and 8.2 %? RP seems quite worried about farming black soldier fly larvae and mealworms, which I guess are roughly as likely to be sentient as silkworms, and therefore only 1.20 (= 0.082/0.068) times as likely to be sentient as nematodes by RP's own lights. Why is the methodoly used to obtain RP's mainline welfare ranges supposed to apply to many invertebrates, but not necessarily to nematodes, mites, and springtails?
I think the following point from @NickLaing is spot on. With the methodology used to obtain RP's mainline welfare ranges, "if a creature have any pain averse behavior (like just withdrawing from anything), it is guaranteed a highish welfare range". A single behavioural proxy likely to be absent, meaning 12.5 % (= (0 + 0.25)/2) likely to be present, implies a welfare range conditional on sentience, and the rate of subjective experience of humans under the pleasure-and-pain-centric model of at least 0.00339 (= 0.125/37) regardless of the simplicity of the organism (including bacteria).
I noted the sheet I just linked above is no longer public. I encourage you to make it public again such that people can examine the assumptions underlying RP's mainline welfare ranges.
You say "0.00027 isn’t a particularly high welfare range [of nematodes]", but it is 41.7 (= 2.7*10^-4/(6.47*10^-6)) times my estimate, and this already implies the expected effects on soil nematodes are way larger in expectation that those on the target beneficiaries.
Thanks for all your work on comparing welfare across species. I have found it super valuable!
Hi Vasco,
I just want to make a few points:
Overall, I encourage you and others on the EA Forum to not view our first version of the welfare range estimates as our final word on this. The book version, Weighing Animal Welfare, is more systematic, and we hope to improve on the methods in the future. But even still, I don’t think that the original version commits one to the view that very simple animals should dominate our calculations absent other highly controversial normative and meta-normative assumptions.
Thanks for sharing your thoughts, Laura!
Could you clarify why your methodology is supposed to apply to silkworms, but not nematodes? I agree nematodes have a less complex brain and nervous system, but silkworms are also less complex than other animals, so I do not understand how you are deciding when your methodology is supposed to be applicable. Did you preregister the animals to which you thought your methods should apply to?
My understanding is that estimating RP's mainline welfare ranges involved tens of judgement calls similar to the one you made to get the probability of sentience of nematodes. My broader point is that I do not know what majorly distinguishes nematodes from silkworms for you to think only the latter are reasonably covered by your methodology.
I had understood the estimates in the sheet which is no longer public were preliminary. However, it is still the case that the welfare range conditional on sentience under the pleasure-and-pain-centric model is proportional to the sum of the probabilities of the respective proxies being present? If so, for RP's probability of sentience of nematodes of 6.8 %, a single behavioural proxy likely to be absent results in a welfare range of nematodes conditional on the rate of subjective experience of humans of at least 2.31*10^-4 (= 0.068*0.00339), which both me and Nick consider high.
The conversion from qualitative probabilistic descriptions to probabilities adds uncertainty, but I do not think it the driver of disagreement. In my mind, and I guess Nick's, the major issue is that the effect of the presence of behaviours on the welfare range is not moderated by neural complexity. RP's mainline welfare ranges consider "one-ninth weight to the possibility that an organism’s welfare range [conditional on sentience, and the rate of subjective experience of humans] is equal to the number of neurons it possesses relative to humans". So an organism having 0 neurons only decreases its welfare range conditional on sentience, and the rate of subjective experience of humans by 1/9. I understand having no neurons at all would also lead to a lower probability of sentience, but I think it should directly imply a much larger decrease in the welfare range conditional on sentience.
What is your best guess for the probability of sentience of nematodes? It could be lower than 6.8 %, but still very far from Pascalianly low. I think reasonable approaches to deal with meta-normative uncertainty (about how to aggregate the recommendations of different moral theories) should not dismiss a 6.8 % or slightly lower chance of causing huge amounts of suffering.
In the book, "10 percent [weight is assined] to the equality model", under which the welfare range conditional on sentience is 1. So the final welfare ranges conditional on sentience are at least 0.1 (= 0.1*1). Do you endorse the estimates presented in Table 8.6 of the book over RP's mainline welfare ranges?
Hi Vasco,
When Bob was selecting the species, he was thinking of adult insects as the edge cases for the model (bees, BSF). He included juveniles to see what the model implies, not because he really thought the model should be extended to them. You'll notice that, in the book, the species list narrows considerably partly for this reason.
On the points related to sentience-conditioned welfare ranges, e.g. "So an organism having 0 neurons only decreases its welfare range conditional on sentience, and the rate of subjective experience of humans by 1/9. I understand having no neurons at all would also lead to a lower probability of sentience, but I think it should directly imply a much larger decrease in the welfare range conditional on sentience."
I think it's a mistake to point to a hypothetical sentience-conditioned welfare range, which is an intermediate step in the calculations, for an animal that has zero neurons as indicative of an issue with the methodology overall for animals with complex brains.
Put straightforwardly, if an animal has zero neurons, it would have a welfare range of 0 overall, because I would give it a zero percent chance of being sentient, which affects all the models.
I also am not going to put a precise probability of sentience on nematodes, but I do think it's much much closer to zero and crosses the threshold of being Pascal's mugged.
I'm finding these discussions very draining and not productive at this point, so will not be engaging further in this debate.
Thanks, Laura.
I encourage you to disclaim in the post with RP's mainline welfare ranges that Bob does not think the methodology used to produce them is applicable to silkworms. In practice, what does this mean? Would it be reasonable to neglect beings to which your methodology is not supposed to apply? Why is the methodology applicable to black soldier flies (BSFs), but not silkworms? I understand a methodology can be more or less applicable, but I still do not understand which concrete criteria you are using. I also think the applicability of the methodology should ideally be taken into account in the estimates such that these are more comparable.
I suggest people account for the lower applicability of your methodology to less complex organisms by using welfare ranges equal to the geometric mean between RP's mainline welfare ranges, and the number of neurons as a fraction of that of humans. Does this seem reasonable?
I am not certain that neurons are required for an organism to have a non-constant welfare, so I think organisms without neurons have welfare ranges above 0. I guess you mean that organisms without neurons have a welfare range of roughly 0, but exactly how close to 0 matters. As I say in the post, "Rounding to 0 a probability of sentience, or welfare per animal-year close to 0 introduces an infinite amount of scope insensitivity. Regardless of the number of beings affected, the change in their welfare will be estimated to be exactly 0".
Could you elaborate on why you seem to believe the probability of sentience of nematodes is Pascalianly low, and therefore arguably much lower than RP's mainline estimate of 6.8 %? I feel like one can reasonably argue from this that the probability of sentience of silkworms is also Pascalianly low, and therefore not worry about improving the conditions of BSFs and mealworms, which RP estimates will be 417 billion in 2033.
Feel free to follow up later if you are finding this discussion draining, and not productive. I think it would be good for RP to write a post clarifying the extent to which the methodology used to produce RP's mainline welfare ranges apply to the animals covered and not covered, and why.
I would say the probability of sentience of nematodes is higher than 6.8 %. From Andrews (2024):
A few responses to @Bob Fischer and @Laura Duffy
Love points one to three from Bob! Perhaps unsurprisingly once he starts disagreeing with me I have some issues.
1. I think I've been misrepresented somewhat. I never claimed that the moral weights project did this “Sum the number of proxies found for a species and divide by the total number of proxies to get the welfare range.”
What I said in the comment was "BOTH their sentience ranges and their behavior scores rely heavily on the presence of pain response behavior".
And In a previous post I did comment that Median final welfare ranges are fairly well approximated by the simple formula Behavioural Proxy sum x Sentience (see graph).
So indeed headline numbers did actually turn out pretty close to the rsult your statement..... "If that were true, then the number of proxies would straightforwardly determine the maximum difference in welfare ranges" . I might be misinterpreting what you mean by this though.
(Behavioral proxy percent) x (Probability of Sentience) = Median Welfare range
2. I stand by (for the moment) my opinion that both the behavioral proxies and Sentient probabilities DO seem to guarantee pretty high final moral weight numbers, although we all will have very different opinions on what 'high' means.
I don't understand how you chose the 0.5% chance of sentience for your low-end calculation? Its far lower than any number in your model The lowest number in your sentience modelling for a nematode is 6.8%, and the silkworm which was included in the MWP is 8.5%. Why pick a number for the example 13x lower than you model actually generated? The 6.8% number from your model would bring a calculation of more like 0.068x0.875x 5/80 which equals 0.0037, or 0.37% as a low end number. This by my lights at least isn't a very low baseline moral weight, but I understand if some would consider that a decently low baseline.
I agree that you have individual models with a low baseline but I'm discussing your overall process. Using your original overall process I still think that high numbers are guaranteed. Also if your method decides to combine bunch of models where some of them are close to P = 1, balanced with other models which are P=0.00001, then you're going to get something in-the-middle-ish (say 0.2-0.8) which also seems high to me.
Also as a side note (less important) I think that 5/80 for behavioural proxies is pretty hard to get for anything that moves around. Anything that has evolved to move is likely from an evolutionary standpoint to be attracted to things, withdraw from things and have some kind of way to remember that - otherwise they wouldn't survive. Maybe that does mean that anything that has evolved to move has a high chance of being sentient though, its an interesting question I know has been discussed before (Can't remember where).
I was surprised to hear "We don’t really put much stock in the probability of sentience estimates, which weren't the focus of the project and are subject to much more uncertainty than the welfare range estimates themselves conditional on sentience". Given that the sentience number is half the final calculation for your headline numbers, which are used freely and widely for expected value calcluations, the sentience number seems pretty important. It also does seem like you put a lot of work into estimating them. Given this statement "On reflection, I think lower numbers are more appropriate than 6.8%, and I really would not anchor on that as “RP’s own lights” I wonder whether reasonable options might be
1. Review the sentience numbers from the project and adjust them to where your thinking is now
2. Not publish a sentience-adjusted moral weight - Instead publish your unadjusted welfare ranges and let people choose their own best-guess sentience multiplier.
Thanks for the great points, Nick. Strongly upvoted.
I'm right on the side of ignoring the effects on tiny creatures.
1. I don't think Rethink's calculation techniques work as well for smaller animals as for larger ones. As I've discussed before, BOTH their sentience ranges and their behavior scores rely heavily on the presence of pain response behavior. This means if a creature have any pain averse behavior (like just withdrawing from anything), it is guaranteed a highish welfare range. If it has a few of these behaviors the numbers get high fast. Their methodology doesn't really have scope for tiny welfare ranges.
2. By my lights I consider this a mugging: I don't consider probabilities this small (for me probability of sentience is <0.0001) worth considering in calculations. Everyone has a different "mugging threshold" as it were, and for me this falls below that. If I bought RP's sentience probability of 0.07, I wouldn't consider this a mugging.
3. Net Positive lives: On the off chance these creatures are sentient, I think they most likely have net positive lives for a few reasons.
First I buy many of the arguments in this article that both wild animal deaths and lives aren't as bad as are often claimed.
Second if we're going to index welfare ranges on human behavior, why don't we index animal wellbeing on human wellbeing? I feel like its a bit "choke on your cake an vomit it too" to suggest that animals are much like humans in their sentience and ability to experience pain, yet don't have similar positive and negative ranges of experience that usually end up net-positive for humans, even those who live in pretty tough situations? I get this is a bit basic...
Third soil nematodes mostly just go about their business, and aren't necessarily under the continuous stress that wild prey are like deer or mice are, where I think there are better arguments for wild animal net-negativity. Yes they have plenty of predators, but I don't think they don't spend a huge proportion of their efforts avoiding them.
4. I'm not just a consequentialist utilitarian. Enough said and on this note I appreciate the recent article by @Rethink Priorities and @Marcus_A_Davis which claims we should be more uncertain about our philosophical judgements.
I buy most of @Michael St Jules 🔸 top 3 arguments as well, which are mostly practical considerations downstream of mine. I suspect he would disagree with my 3 considerations above (but could be wrong).
I think its reasonable to talk about why they are excluded, but in many cases it might be better EV not to communicate it, even when the org has considered these creatures in its calculation. The general public might well distrust orgs that consider these creatures' welfare. Like @Henry Howard🔸 said it can look crazy, but not only that, for most people it can rightly seem offensive or cruel to consider microscopic creature's welfare taking precedent over the welfare of humans or even larger animals. I have graet sympathy for people who criticize our community for indulging in discussions like this.
I could be convinced in the other direction through more compelling research that disagrees with my consideratoins above, but I think its unlikely to be forthcoming in the near future.
Thanks for sharing your thoughts, Nick! I very much agree with your 1st point (not with the other points), but I guess effects on microorganisms, and soil nematodes, mites, and springtails are still the driver of the overall effect of the vast majority of the interventions. In any case, AIM, ACE, AWF, and RP use RP's mainline welfare ranges without adjusting downwards those of less complex species, and I believe this clearly implies that effects on microorganisms, and soil nematodes, mites, and springtails are the driver of the overall effect of the vast majority of interventions they assess. So I think AIM, ACE, AWF, and RP should still explain why they are not accounting for those effects.
Thanks @Vasco Grilo🔸 Yes I agree that if you are using RP's mainline welfare ranges, if you choose to ignore small creatures or even adjust downwards you need a reason to do so . Even one line "we think its a low probability and a mugging" would satisfy me (if that was the reason). But using RP's ranges for other animals while ignoring smaller creatures with zero explanation doesn't fly.
I think there are good reasons though as I outlined for not expressing their reasons publicly. I would suspect that those organisations you listed might have discussed this in-house, and have decent reasons why they aren't considering small creatures but just don't want to make it public because of potential bad optics
Maybe if you reached out to them they would share some of their reasons?
Also I think the statement "I guess effects on microorganisms, and soil nematodes, mites, and springtails are still the driver of the overall effect of the vast majority of the interventions" is not technically correct. I think what you mean is that based on RP's moral weights effects on those animals might carry the highest expected value. Even if their methods are reasonable, there's still a 93% chance that effects on those animals have no effect on any intervention right?
Thanks, Nick.
I guess AIM, ACE, AWF, and RP publicly explaining why they are not considering effects on soil animals would improve their reputation inside and outside the effective altruism community.
I guess the methodology RP used to obtain their mainline welfare ranges would imply higher welfare ranges for soil nematodes, mites, and springtails than the ones I estimated, so I think I am already adjusting downwards. My best guess is that I should adjust downwards even more, but that the expected change in the welfare of soil microorganisms, nematodes, mites, and springtails is still the major driver of the expected change in welfare caused by the vast majority of interventions.
Even just the effects on wild insects could be important, so anyone who is interested in insect welfare should be interested in these effects, too, or have good reason to ignore them, which I'd be interested in knowing (and possibly challenging).
I'd also tentatively put mites, springtails and farmed insect larvae at similar welfare ranges and probabilities of sentience, although quite uncertain. So, if you count farmed insect larvae as worthy of moral consideration at all, you should probably count mites and springtails, too, and to similar degrees on welfarist grounds.
I think some potentially good reasons to ignore the effects on wild terrestrial invertebrates are the following:
7% for crops for direct human consumption, 1% for urban and built-up land, including settlements and infrastructure, and 12-13% for logging for wood. I expect almost all of the rest of habitable land to be left alone, some possibly abandoned and allowed to rewild.
Thanks for the great points, Michael!
The only argument I can think of against this would be optics. To be appealing to the public and a broad donor base, orgs might want to get off of the train to crazytown before this stop. (I assume this is why GiveWell ignores animal effects when assessing their interventions’ impact, even though those swamp the effects on humans.) Even then, it would make sense to share these analyses with the community, even if they wouldn’t be included in public-facing materials.
I think most views where nonhumans are moral patients imply these tiny animals could matter. Like most people, I find the implications of this incredibly unintuitive, but I don’t think that’s an actual argument against the view. I think our intuitions about interspecies tradeoffs, like our intuitions about partiality towards friends and family, can be explained by evolutionary pressures on social animals such as ourselves, so we shouldn’t accord them much weight.
Thanks for the comment, Ariel! That makes sense to me.
I don’t know much about nematodes, mites or springtails in particular, but I agree that, when thinking about animal welfare interventions, one should be accounting for effects on wild animals.
(As Vasco says, these effects plausibly reverse the sign of factory farming—especially cattle farming—from negative to positive. I’m personally quite puzzled as to why this isn’t a more prominent conversation/consideration amongst the animal welfare community. (Aside from Vasco’s recent work, has ~any progress been made in the decade since Shulman and Tomasik first talked about the problem? If not, why not? Am I missing something?))
Thanks, Will.
There is a series from @Michael St Jules 🔸 on human impacts on animals.
1. Reductio ad absurdum: If we consider the lives of nematodes and mites meaningful, suddenly all human welfare questions become meaningless compared to the question of how our behaviour affects nematode/mite welfare. The conclusion will be that we either need to nuke ourselves or completely restructure society around maximising nematode wellbeing. This is impractical, and like many internally consistent but impractical philosophies (nihilism, antinatalism, Kaczynskiism) aren't conducive to a functioning society.
2. Poor analysis: The calculations are always the same: huge numbers multiplied by tiny numbers, all of which are highly uncertain and unlikely to become more certain with "more research" (highly doubt any study is going to illuminate the moral value of mite suffering)
3. Looks crazy: Even mentioning the issue to say why it doesn't matter has a significant cost: the fact that it was considered seriously enough to warrant rebuttal makes the organisation look crazy to normal people, in the same way that Rethink Priorities running an analysis on whether nuking Australia would be net good or bad would look crazy.
I think there is actually a reasonable middle ground here. If indeed the vast majority of all meaningful lives are those of soil organisms, I think an EA approach would imply:
Seems like we're far from a consensus even on whether more or fewer of these organisms is the goal. You suggest that biodiversity loss is bad but Vasco Grilo suggests more monoculture farms is better because that leads to fewer microorganisms and he considers their lives net negative.
Give 1000 researchers 1000 years to study nematodes and demodex mites and I don't believe they'll be able to tell you whether their lives are worth living, let alone exactly what interventions would improve them.
A road to nowhere with great reputational cost
Thanks, Henry. Upvoted.
I simply recommend donating more to GiveWell's funds. Killing humans would be counterproductive. It would mean less human-years, and therefore less agricultural-land-years, and more animal-years of soil nematodes, mites, and springtails, which I think is harmful given my best guess that they have negative lives.
I feel like the same could be said, although to a lesser extent, about caring about invertebrates, and AIM, ACE, AWF, and RP have supported interventions helping these.
Why not advocate for massive desertification efforts and spreading radioactive material to sterilise the soil.? Bring CFCs back to eradicate ozone.
Yep agree. Invertebrates is approximately the point on the moral consideration spectrum at which the huge numbers * tiny numbers with highly uncertainty makes the ethics too fuzzy and volatile to be fruitful.
Somewhere between lobsters and maggots the numbers shoot off towards infinities and the whole thing becomes not worth thinking about.
The cost-effectiveness of advocating for an intervention is the cost-effectiveness of the intervention times the money moved to the intervention as a fraction of the spending advocating for it. I think this fundraising multiplier would be very low for desertification efforts even if they decrease the living time of soil animals more cost-effectively than GiveWell's top charities, such that advocating for supporting these is more cost-effective.
RP's probability of sentience of crayfish (similar to lobsters) is only 1.54 (= 0.453/0.294) times RP's probability of sentient of black soldier flies (BSFs).
I think if you see desertification as good (you seem to be saying it is), you should have very high suspicion that your ethical framework has led you astray somewhere.
I think desertification is beneficial because deserts, and xeric shrublands is the biome with the least soil nematodes, mites, and springtails by far, and my best guess is that these have negative lives, such that decreasing their population is good (although I am highly uncertain).
This seems to me like a Pascal's mugging. Much has been written about why we should not concede to such. To me, it is enough to see that history has not been kind to those who, when faced with a speculative moral analysis in conflict with human values, chose the analysis.
To ask that others prioritize the well-being of nematodes over that of clearly sentient animals (including humans), I'd need far greater confidence in the ability for these small beings to suffer. To prioritize reducing their populations, I believe we need much more confidence that their lives are net negative, and that downstream effects could be avoided. (Even with those considerations, I think there's still some moral uncertainty. Beings with net-negative welfare can still want to live, and their lives have value in non-utilitarian moral perspectives.)
I've seen much written that takes it as a premise that you shouldn't concede to a Pascal's mugging, but I've seen very little about why not.
(I can think of arguments for not conceding in the actual Pascal's mugging thought experiment: (1) ignoring threats as a game-theoretic strategy and (2) threats of unlikely outcomes constituting evidence against the outcome. Neither of these apply to caring about soil nematodes.)
You may be right. I think a lot of us feel that it is intuitively wrong and take that as a premise.
I don't have a rigorous argument against biting the bullet of expected value in the abstract. But in my view, utility calculations will never fully account for 2nd order harms (let alone alternative moral perspectives), and I think that provides ample reason to not rely on numbers alone and err on the side of caution.
Specific risks that come to mind for me here (at least, in the unlikely scenario where the nematode-extinction movement enters the EA mainstream) risks that come to mind for me are reputational damage, intra-movement conflict, climate change exacerbation, biodiversity loss, and the possibility of redirecting evolution toward greater suffering. I'm sure there's plenty of other risks I haven't considered.
I'm all for caring about soil nematodes and researching their welfare! I just think we need more clarity to justify shifting unrelated charity spending.
Thanks, Ailanthus. I would say my recommendation of supporting GiveWell's funds is very much in agreement with human values. In which sense do you think there is a conflict?
Thanks, Vasco.
I would agree that supporting GiveWell (or HIPF) is in alignment with human values.
But if I understand your analysis correctly, you find the vast majority (> 99.9%) of the benefit of giving to these charities is received not by humans, but by soil life (specifically, mostly by nonexistent nematodes that would have existed counter-factually).
All is well so long as human impact and soil life impact are closely correlated, but I see no reason why that must always be the case. I suspect there are interventions that could produce event greater results by convert even more wildlands to cropland, but with no benefits to humans. It's these interventions that come into conflict with my values.
Specifically, I find it morally dubious to purchase animal products with the intention of reducing nematode populations. More broadly, I'm doubtful that speculative, uncertain benefits (even if possibly immense) can justify clear harm. I think this useful moral intuition, given the complexity of 2nd order effects and human tendencies toward motivated reasoning.
Similarly, I also find the idea that destruction of wildlands and their creatures is good to be in tension with my intuitive values. While I wish a positive life for all sentient animals, I also value the existence of wild habitats. If indeed most wild beings have negative lives, these values are in conflict. Nonetheless, I feel that they come from largely overlapping drives, and I expect this is true for most who care about animals. Considering the controversy surrounding killing animals even with very good reasons (e.g. invasive species control) I think a message of "Expand your moral circle to include these creatures... then kill millions of them!" is unlikely to land well.
Along more preference-utilitarian lines, I have a hard time imagining the nematodes getting on board with this. If an superintelligent AI finds that human existence is probably net-negative, does that entitle them to eradicate us?
I could also imagine cases where human welfare and nematode welfare could be actively in conflict. For instance, if one had an opportunity to increase human population more rapidly by installing an authoritarian government.
I'm curious about how you navigate these issues in cases where they're not so obviously aligned. Would you support charities with no other benefits if you found greater impact on soil life? How would you trade off harms to humans and other animals?
Yes, I estimated over 99.9 % of the benefits of donating to GiveWell's top charities come from reducing the number of soil animals.
I would say buying animal-based foods is in agreement with human values in the sense of increasing the welfare of both consumers and animals.
I would be surprised if there are cost-effective ways of advocating for decreasing agricultural land which decrease human welfare. The cost-effectiveness of advocating for an intervention is the cost-effectiveness of the intervention times the money moved to it as a fraction of the spending on advocating it, and this fundraising multiplier will tend to be much lower for advocy for supporting interventions which decrease human welfare.
I doubt that total human welfare is negative. I estimated only 6.37 % of people have negative lives. However, even if total human welfare was negative, I do not think it would make sense for superintelligent AI to kill all humans:
I think increasing human-years via decreasing human mortality is generally more cost-effective than through increasing human fertility. GiveWell's top charities save a life for around 5 k$, 10 % of the lowest cost per additional birth of 50 k$ I found with a quick search.
I think effects on humans are smaller than those on soil animals, so I would focus on these whenever there are conflicts, but I am sceptical about finding cost-effective ways of helping soil animals that significantly harm humans.
60%➔ 80% agreeI appreciate you championing this view Vasco, despite all the pushback. I found your reasoning pretty convincing, and it seems to me like if it's wrong, it will be because of more general philosophical problems with utilitarianism or expected value reasoning.
I think Jonathan Birch's precautionary principle is a reasonable, risk-averse (in all directions) way to go about weighing up these risks. On this framework, some suspicion that a being might be sentient is reason to treat it as a research priority - but not to change public policies, for which there is a great cost. So yes, please someone do research on the sentience of these creatures. But I don't think that charitable organizations should be changing policies just yet.
There are several reasons why they shouldn't:
1. People are very bad at thinking in small or large numbers. When told to name a percentage, instead of saying 0.00014%, they are more likely to say 1, or 5%, because they know it's not 0% and whole numbers are easier to think about. In other words: I take RP's sentience estimates with a large grain of salt.
2. If we did take a pure expected-value approach, I think we would find ourselves being mugged in in many situations, and our whole lives would be quite different. And we would likely make many mistakes, due to the bias mentioned above. For this reason it's pragmatic to not purely maximize expected utility, but to favor 'safe bets'.
3. As others here have mentioned, organizations need to think about their PR, and their PR relies on acting on reliable evidence, not small probabilities. Given that RP's welfare ranges are meant to inform such organizations, they too should err on the side of certainty.
4. RP's welfare ranges inform animal welfare interventions. But when trying to improve animal welfare, we need to think not only about proximate impacts, but also a long-term theory of change. Changing minds and ending speciesism is an integral part of a long-term theory of change for animals, whether that ultimately includes nematodes or not. In the short term being overly radical or strange may be more likely to alienate supporters, thereby harming the movement's progress in the long run.
I didn't 100% disagree with the first question because 1) merely "considering" is probably good 2) I do think it's possible these animals are sentient and 3) I'm not totally sure about my arguments
Thanks for comment, Tristan!
I think the cost from changes in public policies resulting from advocating for the consideration of effects on soil animals is realistically negligible. Researching the welfare of soil animals requires some reallocation of resources, but it currently receives basically no resources at all, and it has little societal appeal.
Could you give concrete examples? I fully endorse maximising expected welfare, but I am not aware of any seemingly badly crazy actions recommended by this approach. I simply recommend supporting GiveWell's funds to increase the welfare of soil animals.
RP's probability of sentience of nematodes is 6.8 %, which is not that small. The probability of dying in a car crash is around 2.70*10^-9 per km (= 10^-6/370), and many people still consider it reasonable to fasten seat belts for increased safety on short trips, even if they would prefer it to be optional.
I personally think the proximate impacts are the driver of the overall effect. However, I believe organisations thinking otherwise should deliberately consider effects on soil animals instead of assuming these correlate well with the effects on target beneficiaries. Helping farmed animals may promote wilderness preservation, thus being harmful to soil animals for my best guess that they have negative lives.
Thanks for your response!
I'm not sure why you say this - for animal advocacy organizations, you're potentially asking them to change their interventions, causing fewer vertebrate lives to be saved. Maybe I should have been clear that while Birch talks about public policies, I think we can apply the same reasoning to charitable policies here.
I suppose my thinking is that if we took all risks seriously that are 1) small in probability, and where we 2) largely just don't have the information to know, we'd be suffering from complex cluelessness. As Greaves points out, we face this kind of cluelessness in many ordinary situations: choosing a career, deciding whether or not to give up coffee, whether to have a kid. In each decision of this type there are many possible outcomes of small probability, but which could end up being quite important (e.g. your kid could end up becoming a dictator; a career might cause you to marry a different person, etc.). In personal decision-making, most people largely ignore all these possibilities, and focus on the more certain and/or likely outcomes. In public decision-making contexts, I think the sensible approach is something similar, but investing more resources into researching the different outcomes in order to lower uncertainty first.
I suspect you might say that in this case we have more evidence than the examples I've given, and I have to admit I've never looked into the evidence of sentience in nematodes, mites or springtails.
Seatbelts are a good analogy, because actually most people didn't think it important to fasten their seatbelts originally. It was only after policy makers became aware of the large number of casualties that they made laws and information campaigns to encourage the use of seatbelts. So this goes to show that when the risk is small and uncertain people tend to discount it; but when it becomes certain, and the cost of avoiding it is small, then people are willing to act. In the case you've presented, it is both small and uncertain. Hence my suggestion of focusing efforts on research first.
By 'proximate' you mean short-term, and by 'the overall effect', you mean the long-term outcomes, right? Could you explain why you think that?
It seems very likely that worldviews will change slowly, requiring us to focus primarily on changing people now, in order to help most animals later. I expect that empathy for some animals (e.g. farmed animals) will gradually lead to empathy for others (e.g. wild animals). It is hard to expect people to care deeply about all animals when they're still eating some of them. So my theory of change starts with efforts that increase caring for those animals that people are closest to, and gradually encourages more radical empathy - the expanding moral circle. And if sentience is probabilistic, that's fine, it's just a circle with fuzzy edges. I assume that most 'traditional' animal rights activists also believe in this vision of progress. And as I said in my last comment, changing interventions now, due to the effects on animals with a small probability of sentience, might mean switching to interventions which less effectively lead to a nonspeciesist future, e.g. by not encouraging people to become vegan (since eating meat has been shown to hinder empathy for animals), or by causing environmentalists to oppose the policies of animal organizations. So by focusing on these animals instead of focusing on changing minds or policies, we might help animals in the short term while harming progress in the long term.
Thanks for the follow-up, Tristan.
Got it. I thought you were referring to public policies from governments, and assuming these would not be influenced by effects on soil animals. I agree organisations considering these may result in less vertebrates being helped, but, even in this case, I would still support the interventions increasing welfare the most accounting for effects on all beings.
I agree that, for example, a random couple can neglect the probability of their children becoming dictators, but only because the probability of this is sufficiently low for its expected effects to be much smaller than the expected effects from other more likely possibilities. I guess the same applies to many of the situations you have in mind. I would be curious to know about examples of considerations being neglected despite their expected effects being estimated (not guessed, as philosophers often do) to be much larger that the expected effects of the considerations being covered.
Sorry for the lack of clarity. By "proximate impacts", I meant the effects I considered. By "the overall effect", I meant all the effects, regardless of whether I considered them or not, across all space and time.
I am suspicious of pursuing interventions which are worse for animals nearterm in the hope they are worth it due to poorly analysed longterm effects. ACE did a randomised controlled trial (RCT) in 2017 suggesting advocating for farmed animals results in greater support for habitat preservation, which I believe is harmful given my best guess that wild animals have negative lives. I am not confident the results are robust, but they illustrate one should not simply assume that helping farmed animals nearterm is beneficial to wild animals longterm.
Thanks, again, for the response.
I acknowledge the examples I gave were kind of bad. If the probability of sentience here really is 6.8%, then that is significant. It's prompted me to look into that evidence and it is truly more than I thought. So that's an update.
I still think, even if they do deserve consideration, there's an argument to be made for delaying that consideration. The argument is of the form "the world isn't read yet". I'm very aware that most vegetarians and vegans are also environmentalists. But that's precisely because they think that environmental protection protects these animals - most have never thought about suffering in nature. My own anecdotal experience is that when I actually talk to such people, and make them aware of the ways that wild animals suffer, then they do tend to be in favor of interventions that would help them, at least if they're not too environmentally disruptive.
So I feel quite confident that the pro-conservation attitude is an intermediary step. People need to care for animals -> then they need to become aware of wild animal suffering -> then they will favor intervention in nature.
If you only focus on the short term, taking such attitudes as fixed, then you can never hope to help very many of these animals.
Interventions promoting a pro-conservation attitude, arguably including the animal-rights movement, may be harmful even if that attitude is a necessary step to care about wild animals. If such interventions make a lot people care about preserving wilderness, but make only a few care about the welfare of wild animals to the extent of being willing to intervene in nature, they may still harm wild animals if these have negative lives.
I think one can help lots of soil animals without people caring about these. I estimate buying beef, and donating to GiveWell's top charities decreases the living time of soil nematodes, mites, and springtails by 89.3 M and 237 M animal-years per $, which is good for my best guess that they have negative lives.
Vasco, do you consider it evidence for or against a theory that it doesn't have recommendations that involve large amounts of destruction of humans? I feel like you have a tendency to dodge this issue when pressed on it by just repeating that your view doesn't in fact recommend supervillain type stuff given practical current constraints. But that leaves open the question of what you would do if those constraints were loosened somehow.
Thanks, David. I would still fully endorse expectational total hedonistic utilitarianism (ETHU) even if this implied some "supervillain type stuff" among the most cost-effective actions. However, in practice, I am not aware of any seemingly villainous actions that follow from fully endorsing ETHU. In general, "supervillain type stuff" increases one's risk of going to prison, and therefore decreases expected future working time and donations, which I think are the major ways one can contribute to a better world.
I think this is fairly fragile to future developments. What if you become convinced humans do more harm than good and also it's a short window of time after it becomes technically possible for laypeople to make a doomsday virus in their garage, and before the government regulates to prevent this? It seems like, even if you have a 50/50 chance of being caught or whatever, the value of succeeding on your credences will be so high, that the expected value on your credence of trying to kill everyone might well still be net positive.
I think I sometimes get an unreal vibe from some of your writing not because I actually think there is a danger you would kill everyone, but because I think its obvious you probably wouldn't, and so you don't really fully endorse ETHU in every feasible situation.
I think my position is super robust to future developments. Feel free to suggest bets. Given my empirical beliefs, I just do not see how the most cost-effective interventions can include "supervillain type stuff". For example, conditional on me having a 50 % chance of killing all life, I would be super powerful, and therefore have way better options available to increase welfare (even if I thought life was negative).
I would recommend killing everyone if this was implied by ETHU. I am less confident about killing everyone being bad than that negative conscious experiences are bad, and that positive conscious experiences are good.
I said a 50/50 chance of getting caught for trying to make a pathogen that brings about human extinction, not a 50/50 chance of successfully killing all life (far, far harder.).
Feasibility depends on that kind of biotech being achievable for small non-expert groups obviously, or at least small groups only some of whom are experts. But even if it is not feasible, and your position is in fact robust, I think the broader point that I don't really believe that you would actually kill everyone in that situation remains.
Sorry. I have removed " as you suggest" from my past comment. In any case, I think a greater feasibility of killing all life means there are more options available to increase welfare, such that is it increasingly unlikely that killing all life is the best one.
So long as expected welfare is appropriately discounted based on reasonable guesstimates of sentience.
Welcome to the EA Forum, biznor! RP's mainline welfare ranges already take into account the probability of sentience (here are RP's estimates for this). In other words, they are expected welfare ranges (as a fraction of the expected welfare range of humans).
Although I’m not an expert my current belief is that the welfare of these things doesn’t matter at all because their utility = 0. Also, the article suggests that the people supporting this are fine with eliminating much of the world’s biodiversity, and I think that would be terrible.
Thanks for the comment, Rick. I care about the effects on soil nematodes, mites, and springtails, but I simply recommend decreasing the uncertainty about the welfare of these, and donating to GiveWell's funds.
The crucial consideration is whether those small invertebrates have a positive or negative welfare significantly different from zero. My best guess is that for those animals, their average welfare is incommensurable with zero, where zero represents the welfare in the state of non-existence. That means a nematode with welfare X can be considered equally good as a non-existing nematode, and increasing that nematode's welfare to X+dX can still be considered equally good as non-existence. In this case, we can neglect population-changing effects on soil nematodes. The effects of agriculture on nematodes are such population-changing effects (agriculture decreases the population sizes of small soil invertebrates), and hence can be neglected.
Hi Stijn,
What makes you think the welfare of soil nematodes, mites, and springtails is practically 0? What is your best guess for the welfare range of soil nematodes and silkworms? My estimate for the welfare range of soil nematodes of 6.68*10^-6 is 0.334 % of RP's mainline welfare range of silkworms of 0.002.
The crucial consideration is not about the welfare range, but about the welfare sign: whether a nematode's welfare is positive or negative. According to your reasoning, (animal) agriculture is very good if nematode's welfare is slightly negative and very bad if it is slightly positive, even if the welfare range is small. Some people believe the welfare sign is negative, others say it is positive, I expect most people believe it is around zero. There does not seem to be any consensus about the welfare sign. And we have no clue how we can possibly answer this question, what research method we could possibly use to determine the welfare sign. This is even more difficult than determining whether a nematode is sentient. I don't expect a solution the next decades. Therefore, I would say the welfare sign of a nematode is indeterminate: the question whether the welfare is positive or negative is ill defined. According to one reference frame, it is negative, according to another equally valid reference frame, it is positive. There is no objective, unique, absolute sign of a nematode's welfare. As in Einstein's special relativity, the moment 'now' for another observer (like the welfare 'zero' for another sentient being) is ill defined and depends on your frame of reference (e.g. whether you move towards or away from that observer). So in a sense it is meaningless to say a nematode's welfare is negative. This implies that we can basically neglect effects that change the size of the nematode population. Saying that animal agriculture is good or bad because it decreases the size of the nematode population, becomes meaningless.
A small remark: your welfare range estimate has three significant digits, which I find very weird given the large uncertainty range. Instead of writing 6.68*10^-6, which gives a false impression of accuracy, I would rather write 'around 10^-5'. (I hate it when my students give overly exact estimates with many digits.)
Also: I tend to believe that, as with the welfare sign, a nematode's welfare range is not exact and has no unique, objective, absolute value. There is a range of valid values for the nematode's welfare range. The intensity of welfare experiences of a nematode and a human are intrinsically incomparable. Like asking the question whether 'i' (the square root of minus 1) is larger or smaller than 1: that is mathematically meaningless, because there are different, equally valid metrics one could use.
Thanks for the additional thoughts, Stijn!
I agree decreasing the uncertainty about the sign of the welfare per animal-year of soil nematodes, mites, and springtails is more important than decreasing the uncertainty about how close it is to 0. However, this is important too. If the welfare per animal-year of soil animals was sufficiently close to 0 for effects on them to be much smaller than those on the target beneficiaries, the sign of the welfare per animal-year of soil animals would not matter.
How do you go from uncertainty about the sign of the welfare per animal-year to an indeterminate sign? The sign of the expected value of one's best guess distribution for the welfare per animal-year has to be negative, 0, or positive. Moreover, it has to be negative or positive in practice, because there are infinitely more distributions with an expected value which is negative or positive than ones with an expected value of 0. Do you think sufficiently large uncertainty about whether a given quantity is below or above a certain value (for example, about whether welfare per animal-year is below or above 0) always implies indeterminacy about whether the expected value of that quantity is below or above that value? For instance, if you are sufficiently uncertain about whether transformative AI (TAI) as defined by Metaculus will happen before or after the community's current median of 2039, would you conclude it is indeterminate whether TAI will happen before or after that date in expectation?
I always present the results of my calculations with 3 significant digits in order to preserve information, and facilitate checks from myself and others. The number of digits is not supposed to reflect the uncertainty.
Effects on soil animals can be neglected if they are much smaller than the effects being covered, but this requires comparing the size of the effects, at least implicitly. Making the comparison explicit has greater reasoning transparency, facilitating the examination of assumptions, and further improvements.
I expect if someone's welfare range is very small, it is more likely that its welfare is incomparable to (incommensurable with, on par with, neither better nor worse nor equal to) non-existence. When I look at my welfare, I try to imagine the average welfare W I would have over a period of say a month, that would make me indifferent between living that month versus non-existence. That welfare level W corresponds to 0 (by definition of zero welfare). Now imagine that same month, but with a tiny bit of extra happiness, for example the taste of one extra piece of chocolate. The welfare is W+dW (with W=0 and dW small). However, I think I would still be indifferent between living that month (with the extra chocolate) and non-existence. Hence, W+dW is also 0. In other words, my indifference curve is rather an indifference band. I believe I have such small indifference bands when it comes to comparing my welfare with non-existence. Now, the width of my indifference band is easily larger than your estimated welfare range of a nematode. You see where this is heading? If the nematode would have an equally wide indifference band as me when it comes to comparing its actual welfare with non-existence (and I have no reason to believe it is otherwise), and if its welfare range is smaller than this indifference band, then all its welfare levels are incomparable with non-existence, or indifferent from 0.
The analogy with TAI timelines: there are many ways to define TAI, based on the set of capacities of the AI. Suppose I say that a TAI needs to have capacities S, and I believe those capacities are reached by the year 2030. Now you may have another definition of TAI, with another set of capacities S'. That will correspond with another timeline, say 2035. The crucial problem is: there is no objective way to determine which definition of TAI is the correct one. So there is no objective way to uniquely determine the timeline. There is rather a range of timelines.
Thanks for clarifying, Stijn!
I would agree that greater indiferrence between not existing, and exiting over a given period implies the expected value of the welfare realised over the period is closer to 0. However, I still think this expected value cannot be 0 for the reasons I described in my last comment. At the same time, I believe this is very much compatible with indifference across potential future states. Humans' capacity to assess their future welfare has a finite sensitivity. Total indifference between not existing, and existing over a given period only suggests the expected welfare over the period is below the sensitivity. I would get a reading of 0 kg putting a feather on a scale with low precision designed to weigh elephants, but this does not mean that the feather has no mass. It just means the scale is not sensitive enough to measure its mass. Likewise, the scale would output 0 kg for feathers of different masses, but this does not mean they all have the same or an incomparable mass. It just means the scale is not sensitive enough to measure the differences in their mass.
I do not think you can assume humans' and nematodes' assessments of their future welfare are equally sensitive. I assume the force needed to crush a nematode would be imperceptible by humans. So, if nematodes' were as sensitive as humans with respect to physical forces, they would not be able to avoid mechanical threats, which seems very counter to what would be evolutionarily advantageous.
I referred to TAI as defined by Metaculus. I understand this still involves some ambiguities. Yet, I would not say it is indeterminate whether TAI will happen before or after 2039.
To make the analogy with the scale more accurate: suppose mass could also be negative (because welfare can be negative), and most of all: suppose Newton's law would use the measured mass, as measured on that scale. The latter would be a bit like quantum mechanics, as if the feather is in a superposition of different masses, and doing a quantum measurement with that scale gives its mass that enters Newton's law. Then the question whether the mass of the feather is positive or negative, is ill-defined when the feather is in a superposition of both positive and negative masses.
A better physics analogy is Einstein's theory of special relativity. Suppose you and I have super accurate clocks. I determine that "now!" corresponds with time zero. Relative to my "now!", there are events in the past and future. But suppose you also determined your "now!" as time zero on your clock. And suppose the space-time event that corresponds with you saying "now!" lies outside the lightcone of my "now!" space-time event. The question is: is your "now!" in the future or the past of my "now!"? When the speed of light is finite, this question is always ill-defined, even if we had infinitely accurate clocks. You can always pick a reference frame according to which your "now!" exactly corresponds with time zero on my clock (yes, with infinite precision; what a coincidence!). And pick another reference frame according to which your "now!" is in the future of my "now!". In the analogy, the time of your "now!" on my clock, corresponds with the welfare of the nematode relative to non-existence. My "now!" corresponds with zero welfare of non-existence, your "now!" corresponds with the welfare of the nematode. If your "now!" is in the future (past) of my "now!", that corresponds with a positive (negative) welfare of the nematode. The different reference frames in special relativity correspond with different 'welfare frames' of the nematode. So even if a nematode could measure physical forces with infinite precision, and would be sensitive to the pain of the slightest increase in a physical force, it would still be impossible to say whether the nematode has a positive or negative welfare. I can compare my welfare with non-existence to a high degree, but not infinitely accurately. In the special relativity analogy, this corresponds with a space-time with a very high lightspeed. But for the nematode, the comparison with non-existence is much more difficult, which would correspond with a space-time with a very low speed of light. If the speed of light is extremely low, or say zero, almost all your space-time events are outside my lightcone, which means you can pick any moment along your timeline and I can pick a reference frame according to which that moment is in the future of my "now!". With the nematode: pick any full description of its experiences, the physical forces that it feels and so on, and I can pick a welfare frame according to which that nematode has a positive welfare, and you picked a welfare frame according to which that same nematode, with the very same experiences, has a negative welfare. There is no objective way to determine which of our welfare frames is the correct one. If a nematode's welfare range is very small, it is likely that all its experiences that it could possibly have always lead to a welfare level that is incommensurable with zero, even if the nematode could measure its own welfare with infinite precision.
As for the TAI: the point is that I can give another equally reasonable/valid definition of a TAI as the one of Metaculus, and there is no objective way to determine which of our definitions is the correct one. This intrinsic indeterminacy of the definition results in the incommensurability of the timelines. I can always pick a definition according to which the arrival of TAI is after 2039, even if according to Metaculus' definition of TAI, the arrival is exactly in 2039 and we can measure that arrival date with infinite precision. The different definitions of TAI correspond with the different reference frames in special relativity, and the different welfare frames of the nematode.
Thanks, Stijn. I think it is better to keep the analogies simple. Imagine there is a glass with alcohol, whose freezing point is -114 ºC, at a temperature of 0.2 ºC. Consider a thermometer which systematically underestimates temperature by 1 ºC, another which systematically overestimates temperature by 1 ºC, and another which is well calibrated. In addition, assume the thermometers can only display integers. The thermometers would output temperatures of -1, 1, and 0 ºC. However, despite the different calibration of the thermometers, and their outputted temperature, it is not indeterminate whether the temperature of the alcohol in the glass is below or above 0 ºC. Subjective measurements of the hedonistic welfare of nematodes are much more uncertain than measurements of the temperature of ethanol with thermometers. However, what is the fundamental difference that makes you conclude there is no fact of the matter about whether nematodes have a positive or negative welfare per animal-year? Why does it mean one can neglect the welfare of nematodes, but not of farmed insects?
I still prefer the special relativity analogy, because its mathematical structure is similar to what I have in mind about welfare of nematodes. Your alcohol temperature analogy seems to be misleading, because it speaks about over- and underestimations and being well calibrated, but that is not the issue. The clocks of observers in special relativity are all well-calibrated, no clock overestimates time. And yet there is this weird kind of intransitivity in special relativity, where space-time events X and Y happen simultaneously (at time 0), space-time event Z is in the future of Y, and yet events X and Z also happen simultaneously (at time 0, but according to another reference frame). Same goes for nematode welfare: nematode in state X (non-existence) and Y have the same welfare (equal to 0), the nematode in state Z is strictly happier than in Y, and yet there is a welfare frame according to which states X and Z are equally good (welfare 0).
I believe (farmed) insects have a wider welfare range than nematodes, which means it is less likely that all their positive and negative experiences are incommensurable with non-existence. An insect can be so miserable that its welfare is negative according to all welfare frames, in which case it is objectively/absolutely true that that insect has a negative welfare. A nematode cannot reach such intense levels of misery.
Thanks, Stijn.
There is only one right perspective to assess the welfare of a being under hedonism. The perspective of the being whose welfare is being assessed. When I say my best guess is that the (hedonistic) welfare per animal-year of soil nematodes is negative, I am effectively guessing that I would prefer not existing over being born as a random soil nematode.
I am missing a quantitative argument distinguishing nematodes from insects. I feel like someone else could just as reasonably argue for neglecting effects on insects because their welfare range is still too small. For reference, "My estimate for the welfare range of soil nematodes of 6.68*10^-6 is 0.334 % of RP's mainline welfare range of silkworms of 0.002". Why is 0.002 not small enough? What is the minimum welfare range for effects to be considered? Why?
The crucial issue is that a nematode may not be able to compare its welfare with non-existence. Even for me it may be hard to compare welfare with non-existence, i.e. imagining a level of welfare at which I would be indifferent with non-existence. I believe there is not always a straightforward answer, even under hedonism, when one's welfare is at a similar level as non-existence. Just as in special relativity there is not always a straightforward answer when a space-time event happens now. Still, in special relativity, time is a continuous variable that can in principle be measured with infinite accuracy (i.e. clocks could in principle be infinitely well-calibrated in the theory of special relativity). So the notion of time makes perfect sense in special relativity, but the notion of "now" does not. Similarly: the notion of welfare makes perfect sense in hedonism, but the notion of zero welfare does not.
Yes, we are missing quantitative estimates for determining zero welfare of nematodes and insects. Like we do not know the speed of light. All I can say is that in the special relativity analogy, a nematode has a "lower speed of light" than an insect. And sure, I consider it likely that even insects have a welfare range that is smaller than the neutral range. This neutral range is the range of welfare levels that are incommensurable with zero, or the range of welfare levels for which there always is a valid welfare frame such that the welfare is zero. If the welfare range is smaller than the neutral range, it is always impossible to objectively/absolutely determine whether a welfare is positive or negative: there is always a valid welfare frame for which the welfare is positive, and another equally valid welfare frame for which the welfare is negative. This is all compatible with hedonism. I guess you assume in your hedonic theory, the speed of light is infinite, which means there is one absolute welfare frame (as in Newtonian physics there is one absolute reference frame). The more I think about it, the more skeptical I am about that assumption of an absolute welfare frame.
Do you think newborns can compare their (hedonistic) welfare to non-existence? I assume they cannot. However, I still think there is a fact of the matter about whether any given newborn has a positive or negative (expected) welfare over a certain period of time. Likewise for nematodes or any other being.
I do not quite understand what you mean by neutral range, but it looks like you would consider effects on nematodes if the neutral range of these was smaller than their welfare range. If so, you can only neglect effects on nematodes if you are certain their neutral range is larger than their welfare range. I assign this a probability of 0, not 1, as I see welfare per animal-year (as assessed by the being experiencing it) as a continuous distribution.
Yes, I think newborns can compare their welfare with non-existence, to a small degree, but I'm uncertain about it. That is why I think it is so difficult to estimate whether my newborn sons have a positive or negative welfare. I tend to believe that my first son had a negative welfare the first few weeks and positive now, and my second son (who is a month old and lying next to me now) has an average positive welfare these days. But it could easily be indeterminate. For a nematode I'm much more confident that it is indeterminate. A question I would ask is: would the most empathic veterinarians prefer to euthanize a nematode, like they prefers to euthanize a dog when that dog has a negative welfare? I doubt it. Would the most empathic total utilitarians prefer to breed more nematodes, like they prefer the existence of more individuals with positive welfare? I doubt it. Would those people believe that a nematode happens to have exactly 0 welfare? I doubt it. So a nematode's welfare is not clearly negative, not clearly positive, and not clearly zero. Then what is it? It's incommensurable with 0.
I wrote some ideas about that neutral range here:
https://stijnbruers.wordpress.com/2021/10/16/person-affecting-neutral-range-utilitarianism/
https://stijnbruers.wordpress.com/2020/08/26/relativistic-welfare-farm-animal-abolitionism-and-wild-animal-welfarism/
Yes, if a nematode's neutral range was smaller or zero, I would say that there is an objective fact of the matter whether the nematode has positive or negative welfare. just like when speed of light is infinite, there is an absolute reference frame, and every space-time event is either in the future, the present or the past of this space time-event I call "now!" That's Newtonian physics.
Your case would be like assigning a probability of 0 to the possibility that the speed of light is finite. Note that the fact that welfare is continuous, is irrelevant: also time in special relativity is continuous.
Thanks for the discussion, and best wishes for your newborn son, Stijn!
yes, this was a fruitful discussion; thanks! I summarized my arguments here: https://forum.effectivealtruism.org/posts/udcCBBwGnCneLRjkH/should-we-consider-the-welfare-of-small-soil-animals-on-the