Net global welfare may be negative and declining

I appreciate the care and detail here, but would guess that wild animals dwarf everything considered here and present a much more difficult + important question.

How bad are forests per unit of land vs corn/soy/wheat fields or cattle ranches that have been replacing them seems like a key question. 

This sort of work is very sensitive to your choices for moral weights, and while I do appreciate you showing your input weights clearly in a table I think it's worth emphasizing up front how unusual they are. For example, I'd predict an overwhelming majority of humans would rather see an extra year of good life for one human than four chickens, twelve carp, or thirty three shrimp. And, eyeballing your calculations, if you used more conventional moral weights your bottom-line conclusion would be that net global welfare was positive and increasing.

While I agree that net global welfare may be negative and declining, in light of the reasoning and evidence presented here, I think you could and should have claimed something like this: "net global welfare may be negative and declining, but it may also be positive and increasing, and really we have no idea which it is - any assessment of this type of is enormously speculative and uncertain".

As I read the post, the two expressions that popped into my head were "if it's worth doing, it's worth doing with made-up numbers" and "if you saw how the sausage is made ...".

The problem here is that all of the numbers for 'animal welfare capacity' and 'welfare percentages' are essentially - and unfortunately - made up. You cite Rethink Priorities for the former, and Charity Entrepreneurship for the latter, and express some scepticism, but then more or less take them at face value. You don't explain how those people came up with numbers and whether they should be trusted. I don't think I am disparaging the good folk at either organisation - and I am certainly not trying to! - because you asked them about this, I think they would freely say "look, we don't really know how to do this. We have intuitions about this, of course, but we're not sure if there's any good evidenced-based way to come up with these numbers";* indeed, that is, in effect, the conclusion Rethink Priorities stated in the write-up of their recent workshop (see my comment on that too). Hence, such numbers should not be taken with a mere pinch of salt, but with a bucketload. 

You don't account for uncertainty here (you used point estimates), and I appreciate that is extra hassle, but I think the uncertainty here is the story. If you were to use upper and lower subjective bounds for e.g. "how unhappy are chickens compared to how happy humans are?", they would be very large. They must be very large because, as noted, we don't even know what factual, objective evidence we would use to narrow them down, so we have nothing to constrain the bounds of what's plausible. But given how large they would be, we'd end up with the conclusion that we really don't know whether global welfare is negative or positive.

* People are often tempted to say that we could look at objective measures, like neuron counts, for interspecies comparison. But this merely kicks the can down the road. How do we know what the relationship is between neuron counts and levels of pleasure and pain? We don't. We have intuitions, yes, but what evidence could we point to to settle the question? I do not know. 

I think this is one of those posts where the question is ultimately more valuable than the answer. And to be clear that isn't a criticism and I upvoted the post. I appreciate posts that push people to think about important questions, even if our best guess answers are not currently very compelling.

• Does the conclusion flip if you don't value 30 shrimps/shrimp moments the same as a human?  
• It might be more meaningful to present your results as a function, e.g., if you value shrimps and chicken at xyz, then the overall value is negative/positive 
• Particularly in uncertain domains, it might have been worth it to consider uncertainty explicitly, and RP does give confidence intervals.

Hi Kyle, 

This is a very interesting post! One quick and very small technical detail: Rethink Priorities' welfare ranges aren't capped at 1 for non-human animals. (It just happens that, when we adjusted for probability of sentience, they all happened to have 50th percentile estimates that fall below 1). They're instead a reflection of the difference between the best and worst states that the non-human animal can experience relative to the difference between the best and worst states that a human can experience (which is normalized to 1). In theory, this relative difference could be greater than 1 if the range in intensity of experiences that a non-human animal can experience is wider than that of humans. 

In fact, one of our welfare range models (the undiluted experiences mode) that feeds into the aggregate estimates tends to produce sentience-adjusted welfare range estimates greater than 1 under the theory that less cognitively complex organisms may not be able to dampen negative experiences by contextualizing them. As such, a few animals have 95th percentile estimates for their welfare ranges that are above 1 (octopuses, pigs, and shrimp). Here are some more details about the models and distributions: https://docs.google.com/document/d/1xUvMKRkEOJQcc6V7VJqcLLGAJ2SsdZno0jTIUb61D8k/edit?usp=sharing As well as the spreadsheet of results from all models: https://docs.google.com/spreadsheets/d/1SpbrcfmBoC50PTxlizF5HzBIq4p-17m3JduYXZCH2Og/edit?usp=sharing 

Again, this is a really thought-provoking and sobering post, thanks for writing it :)

Thanks for putting this together! I'm skeptical of putting too much weight on the conclusions (just given how much uncertainty there is), but I think this is valuable addition to the conversation on this subject. 

It's worth noting that MacAskill touched on this topic in chapter 9 of WWOTF, using neuron counts as a proxy for moral weight. He comes to a very different conclusion, which makes sense given that neuron count-based moral weights for basically all animals are much lower than the RP moral weights: 

To capture the importance of differences in capacity for wellbeing, we could, as a very rough heuristic, weight animals’ interests by the number of neurons they have. The motivating thought behind weighting by neurons is that, since we know that conscious experience of pain is the result of activity in certain neurons in the brain, then it should not matter more that the neurons are divided up among four hundred chickens rather than present in one human. If we do this, then a beetle with 50,000 neurons would have very little capacity for wellbeing; honeybees, with 960,000 neurons, would count a little more; chickens, with 200 million neurons, count a lot more; and humans, with over 80 billion neurons, count the most. This gives a very different picture than looking solely at numbers of animals: by neuron count, humans outweigh all farmed animals (including farmed fish) by a factor of thirty to one. This was very surprising to me; before looking into this, I hadn’t appreciated just how great the difference in brain size is between human beings and nonhuman animals.

If, however, we allow neuron count as a rough proxy, we get the conclusion that the total weighted interests of farm land animals are fairly small compared to that of humans, though their wellbeing is decisively negative.

This does not yet resolve whether the welfare of humans and farmed animals combined is negative. Even though, in totality, farmed animals may have fewer neurons, the vast majority of farmed animals (chicken and fish) live lives full of intense suffering, which could well outweigh total human wellbeing. If the intensity of the suffering of chickens and fish is at least forty times the intensity of average human happiness, then the combined wellbeing of humans and farmed animals is negative.

Are you willing to share your underlying source code? I might be interested in adding uncertainty to the analysis.  

In terms of digital sentience, I do think it is somewhat reassuring that we harm animals out of indifference, not malice.  There's no particular general reason I can see to think that when we pursue our interests with indifference rather than hostility to the well-being of other sentients, that the side-effects of this for the other sentients will systematically and non-coincidentally harm rather than help them. Like, it could be that what suits our interests just happens to be to bunch of digital minds that have no feelings, or that mostly enjoy the tasks we give them, just as much as it could turn out that the best way to exploit them involves net suffering for the minds. Factory farming is basically a single case, so I don't think it should move us much towards "actually exploiting other sentients for our own gain will have systematically negative, rather than basically random effects on their welfare". After all, it's not even completely clear the pre factory farming was net negative for the animals involved. 

Thanks for the post ! 

This is a question I have asked myself for a long time, and that I don't like asking myself, but that I find incredibly important. I'm less motivated to work on x-risks if in practice it means "supporting to continuation of factory farming for a long, long time".

I also think that including this possibility into other cause areas might be very important. This advice especially seems very crucial : modeling the animal welfare impacts of human health and global development interventions.

I'll note that while there are indeed massive uncertainties, I really have trouble seeing how the combined happiness of humanity might outweight the terrible suffering of farmed animals, since they are far more numerous and have far worse living conditions(although wild animal suffering might change everything). Especially since there are > 4x as many farmed chickens as humans, and > 10x more farmed fish and shrimp, with terrible living conditions.

Instinctively, of course, I think of humans as having more worth, and I like the idea of saving lives. But in practice, I don't find good arguments to justify that humans have a much higher moral weight than other animals.

This work draws heavily on the Moral Weight Project from Rethink Priorities and relies on the same assumptions: utilitarianism, hedonism, valence symmetry, unitarianism, use of proxies for hedonic potential, and more. Although I think the Rethink Priorities welfare range estimates are currently the best tool available for interspecies welfare comparisons, I do not necessarily endorse these assumptions in full, nor do I think the Rethink Priorities welfare ranges are the “correct” weights—only the best available.

These are some pretty big assumptions. In particular, when they discussed the sensitivity to the hedonism assumption, ReThink argued that this simplifying assumption wasn't a big deal, because relaxing it would only make a 3x difference or so, and this was not relevant for cause prioritization:

We suggest that, compared to hedonism, an objective list theory might 3x our estimate of the differences between humans’ and nonhumans’ welfare ranges. But just to be cautious, let’s suppose it’s 10x. While not insignificant, that multiplier makes it far from clear that the choice of a theory of welfare is going to be practically relevant. To see this, recall that Open Philanthropy once estimated that "[if] you value chicken life-years equally to human life-years, this implies that corporate campaigns do about 10,000x as much good per dollar as top [global health] charities."

But your human vs all-nonhuman lines are sufficiently close that a 3x, let alone a 10x, difference would reverse the bottom line conclusion. I think it does make a considerable difference that the argument they use for 'why we don't need to worry about this assumption' doesn't apply here, given that pure hedonism is basically the least-favourable theory of welfare to use here.

(As it happens I also think that their attempt to bound the importance of non-hedonic factors at 3x or 10x doesn't work, because their thought experiments about Tim only consider relatively weak non-hedonic goods).

Perhaps a dark thought, and coming from someone not being well versed in these types of high-level analyses: Could this work possibly lend some support to antinatalist/misanthropic points of view?  I should state for the record that I feel tremendous hesitation to these types of views. 

Thanks this is great.

I think it's worth the perhaps basic reflection that rethink priorities' animal welfare ranges are so close to humans, and the number of farmed animals so high that any global welfare estimates are going to look terrible using their data. This isn't a value judgement at all, just a perhaps-obvious observation.

From my perspective, for accuracy the next thing to include should be uncertainty ranges in the graph. For example a graph like this would need that before it could be considered for any serious peer reviewed literature.

Unfortunately even just using rethink's errors, ranges will be so large it will even make displaying them in a graph difficult.

Another huge source of uncertainty like you mention is wild animal welfare, an even harder area where most EAs I think would estimate their welfare as net negative, whereas some (like me) would estimate it perhaps positive.

As a side note I would love a non EA, non animal rights focused group to do some kind of counter - calculation on welfare ranges so I could see what different assumptions they might make, but that's probably unrealistic as why would a dispassionate group bother with that kind of difficult work? Maybe we ould just use neuron weights as a counter equivalent?

Again this is a really interesting graph nice one.

When doing rough analysis, there are virtues to having simple models simply laid out, so I commend this - but step 2 is looking at which analytic and other choices the simple model is most sensitive to, and laying that out, and  I think this post suffers from not doing that.

In this case, there are plausible moral and analytic assumptions that lead to almost any conclusion you'd like. A few examples: 

• Include declining total numbers of net-negative lives among wild animals. 
• Reject total utilitarianism for average utilitarianism across species. 
• Change your time scale to longer than 10m years, and humanity is plausibly the only way any species on earth survives. 
• Project species welfare on a per-species basis instead of an aggregate, and it may be improving, and this may be Simpson's paradox. 
• Change the baseline zero-level for species welfare, and the answer could reverse.

And other than the first, none of these is even considered in you future directions - even though the assumptions being made are, it seems, far too strong given the types of uncertainties involved. So I applaud flagging that this is uncertain, but don't think it's actually useful to make any directional claim, not would further modeling do that much to change this.

Finally, I'm struggling to see how and where this is decision relevant for people or organizations - but that's an entirely different set of complaints about how to do analyses.

On your future directions / tentative reflections (with apologies that I haven't looked into your model, which is probably cool and valuable!):

To the extent that we think this is relevant for things like lock-in and x-risk prioritisation we need to also think that current trends are predictive of future trends. But it's not at all clear that they are once you take into account the possibility of explosive growth a la https://www.cold-takes.com/all-possible-views-about-humanitys-future-are-wild/. Moreover, worlds where there is explosive growth have way more moral patients, so if their probability is non-negligible they tend to dominate moral considerations.

Once we focus on explosive growth scenarios as the most important, I find much more persuasive considerations like these: https://www.effectivealtruism.org/articles/the-expected-value-of-extinction-risk-reduction-is-positive

I've written up decently extended reflections on why we shouldn't give much weight to the fact that the history and presency of our world is an utter moral hellscape that I'm happy to share privately if these questions are important for you.

(All that said, I do think lock-in is undervalued in longtermism and I'm excited to see more work on that, and I do think the path to x-risk prioritisation is much more complicated than many EAs think and that these kinds of considerations you point out are exactly why.)

I was scared when seeing the title. Then I read a little bit more:

That is, the entire good of humanity may be outweighed by the cumulative suffering of farmed animals, with total animal suffering growing faster than human wellbeing is increasing, especially in recent decades

I already thought that was the case. It's really sad, but at least it's not another massive source of suffering to add to my list. Thank you for substantiating this with a calculation!

I’m skeptical of Charity Entrepreneurship’s estimates. Really, cattle have negative utility?

Cattle avoid being factory farmed. So I’d bet the average cow would much rather be born than not.

Really, are shrimp so picky about their environment that they feel negative, how they are farmed? I’d think any place where they are alive and fed may yield positive utility to a shrimp.

So I am distrustful of their ratings. Kudos to your analysis, though, I think that’s the right approach, assuming good data being fed into it

Executive summary: The total moral value of the world may be negative and declining, as growing animal agriculture could be causing more suffering than improvements in human welfare offset.

Key points:

1. The total moral value of the world includes both humans and non-human animals, with the rise of industrial animal agriculture significantly impacting the latter.
2. The author's analysis, focused on the period from 1961-2021, suggests net global welfare may be both negative and declining, due to the increased suffering of farmed animals outweighing human wellbeing.
3. The author notes several assumptions and limitations, including the exclusion of wild animals and insects, and the use of point estimates rather than confidence intervals.
4. The author used data from the Food and Agriculture Organization of the United Nations and human population data, along with estimates of welfare ranges across different species, to calculate the total welfare of a given species in a given year.
5. The analysis suggests that the suffering of farmed animals is so intense and their populations have grown so large that they may outweigh all of humanity’s progress.
6. The author concludes that humanity may have inadvertently set systems in motion that put the total welfare of the world on a steep downward trajectory, and calls for more rigorous research on animal welfare and the development of interventions aimed at improving total welfare.

This comment was auto-generated by the EA Forum Team. Feel free to point out issues with this summary by replying to the comment, and contact us if you have feedback.