Why the Open Philanthropy Project Should Prioritize Wild Animal Suffering

22

Criticism of effective altruist organizationsWild animal welfare
Frontpage

Cross-posted to my website.

Like the last time I wrote something like this, my suggestions here could apply to any large foundation. But most large foundations don’t care at all about what I say, and the Open Philanthropy Project cares at least a tiny bit about what I say, so I’m going to focus on Open Phil.

The Open Philanthropy Project ought to prioritize wild animal suffering (WAS). Here’s why:

  1. WAS is important and neglected.
  2. WAS is not tractable for most actors, but it’s tractable for Open Phil.

(Previously I discussed some of my issues with the importance/neglectedness/tractability framework, but I believe it works reasonably well for our purposes here.)

Why wild animal suffering matters

The problem of wild animal suffering has enormous scale. There exist far more sentient wild animals than there do humans or factory-farmed animals. Wild animal suffering dwarfs all other problems that currently exist. Some other problems (such as existential risk) may matter more, but WAS is certainly the biggest problem that’s happening right now.

Additionally, wild animal suffering is neglected: hardly anyone cares about this problem, and of the people who care, hardly any of them are trying to do anything about it. Animal Ethics is the only organization spending non-trivial time on the problem of wild animal suffering, and it’s a small organization with limited staff time and narrow focus–I see room for much, much more work on reducing suffering in the wild than what Animal Ethics does currently.

Why Open Phil should prioritize wild animal suffering

For people who care about animals, their biggest objection to reducing wild animal suffering is that it’s intractable. But this is mistaken: we can do lots of things right now to work toward reducing wild animal suffering. (If you doubt that we can do anything about wild animal suffering, please, please read my essay on this subject, and if you disagree, leave a comment explaining why.)

Even given the sad state of WAS research, we already have some concrete proposals for how to reduce wild animal suffering without risking big negative side effects. For example, Brian Tomasik has suggested paying farmers to use humane insecticides. Calculations suggest that this could prevent 250,000 painful deaths per dollar. This intervention alone looks much more cost-effective than GiveDirectly even if we heavily discount insects’ capacity for suffering. And this is just an initial idea; surely there exist much more effective interventions than this, and we could find them if we spent more time looking.

Reducing suffering in the wild is probably much more tractable than most people tend to think. That said, if you want to work on wild animal suffering, you either need specific relevant skills (which are rare and hard to develop) or you need to fund an organization doing relevant work; and right now Animal Ethics is the only such organization. We have something of a coordination problem here where people won’t work on wild animal suffering because they can’t get funding, and people don’t want to fund it because so few people are working on it.

What we need is a large, committed source of funding to jump-start the cause. If the Open Philanthropy Project began funding work on wild animal suffering, it could stimulate new research efforts or small-scale interventions by offering grants. Specifically, Open Phil should probably create a new focus area for wild animal suffering and possibly hire dedicated staff. This problem has such large scale, and so many possible interventions, that it absolutely deserves to be a dedicated focus area. Open Phil might consider lumping WAS under its farm animal welfare program, but this would excessively constrain its budget and limit the amount of staff time that it could receive. Wild animal suffering is a massive problem, and easily deserves as much attention as most of Open Phil’s other focus areas.

16 comments, sorted by Highlighting new comments since Today at 5:37 PM
New Comment

Did you see the presentation Lewis Bollard gave at the Sentience Conference? He mentions that wild animal suffering, insect suffering, and many other exciting cause areas are totally on the table at Open Phil (though under the auspices of the farm animal welfare department).

The Sentience Politics of the EA Foundation is also launching a research program on WAS led by Brian, so EAF is another great donation target for pushing WAS activism in addition to Animal Ethics.

The main reason why I think research on WAS is very tractable is the one described in The Attribution Moloch. We tend to think of research as the necessary drudgery that may uncover a highly effective intervention and whose value is measured by the effectiveness of that potential intervention times the probability of discovering it.

But J-PAL, for example, had to conduct a whole phalanx of experiments at all levels of scale and formality until they discovered that deworming was a great way of boosting school attendance. Each of those experiments that showed an intervention to have limited impact was not a failure but was highly valuable in that it informed the future research that led to the discovery of the effect of deworming.

Therefore I assign the same value to any research on WAS that produces results activists can update on that I would eventually assign to the implementation of the highly effective interventions that’ll be discovered. (If they are discovered – but WAS is sufficiently vast and neglected that I’m optimistic here.)

Thanks for writing this. One small critique:

"For example, Brian Tomasik has suggested paying farmers to use humane insecticides. Calculations suggest that this could prevent 250,000 painful deaths per dollar."

I'm cautious about the sign of this. Given that insects are expected to have net negative lives anyway, perhaps speeding up their death is actually the preferable choice. Unless we think that an insect dying of pesticide is more painful than them dying naturally plus the pain throughout the rest of their life.

But overall, I would support the recommendation that OPP supports WAS research.

Hi Michael :)

The "Humane Insecticides" article talks about using different insecticides that are equally lethal, rather than reducing insecticide use. (It expresses similar concerns as those you raise about the sign of reducing insecticide use.) The 250,000 number is an amount of pain equivalent to that many pesticide deaths.

That said, I'm somewhat skeptical about the number quoted in the article because it ignores a lot of costs (e.g., setup costs, identifying the right alternative chemicals, etc.). I first wrote it in 2007 when I was less attuned to the arguments for conservatism in cost-effectiveness estimates.

Still, some other estimates suggest similar orders of magnitude for how much expected insect suffering can be prevented per dollar, although these interventions are mostly more controversial (and more speculative).

I'm interested in

For example, Brian Tomasik has suggested paying farmers to use humane insecticides. Calculations suggest that this could prevent 250,000 painful deaths per dollar.

(which is honestly not about WAS)

and

Open Phil might consider lumping WAS under its farm animal welfare program

because they look like paths to circumvent the biggest red flag, which is the profoundly negative reaction that most people have to calm discussions about wild animal suffering. It seems intuitively like an idea which is still before its time relative to the general population. I think most people would agree that it's disliked to a greater extent than perhaps any other issue on the table.

I don't know how big of a problem it is for the EA movement if lots of people notice what Open Phil is doing. It might be a problem. But doing something like the above would not be very controversial, would begin to shift priorities, and would create a foundation of work that blurs the line between traditional animal welfare and WAS work.

(which is honestly not about WAS)

What would you call that kind of suffering if not WAS?

Farm insect suffering? It is insects, being deliberately killed, on farms. It's very different from the idea of intervening in natural ecosystems.

AI safety gets a similar negative reaction to WAS, but it's Open Phil's top priority for 2016. So I don't think this is a major concern.

I definitely don't think WAS should be part of the farm animal welfare program--it will almost certainly end up underfunded and won't do as much good as it would as a separate cause area with dedicated staff.

EA started pulling additional mixed or negative reactions after moving into AI safety, such as the Dylan Matthews article or all the people who had prior familiarity with LessWrong and thought the whole thing was kooky.

Also, people's reactions to wild animal suffering proposals seem to be substantially more negative than reactions to AI safety work (dataset: comment replies to McMahan and MacAskill's articles, comment replies to AI safety editorials, several thousands of Reddit comments).

I see more negative reactions to AI safety. I don't believe either of us has strong enough evidence to make a solid claim that one attracts substantially more negative PR than the other.

No one is actually opposed to the basic idea of researching AI safety. Some people just think it's silly. But people actually think that intervening in nature is actually ethically wrong. The issue also links to debates over meat consumption, where people are already wired to be irrational. For these reasons you see people call out the idea in stronger terms than they talk about AI.

People react more erratically and strongly to AI safety if they are already involved in computer science and AI. But that's not a representative reference class.

How many painful mosquito deaths would you have to be offered to prevent to choose that over causing one new human life (of quality equal to that of a typical person today) to be lived (all instrumental effects / consequences aside)?[1][2][3] (For my answer see [2].)

What would the distribution of EAs' answers look like? College graduates' answers? Everyone's answers?

What range of answers does the OP assume?

Or more broadly, for what range of moral theories can a case be made that WAS should be prioritized?

I ask these questions because, while I find the OP argument intriguing, my current values (or my current beliefs about my values, depending on how you want to think about it) are such that preventing mosquito suffering is very insignificant relative to many other things (e.g. there being more humans that live good lives, or humans living better lives) and is therefore far from being a high priority for me.

While I haven't dived deeply into arguments for negative utilitarianism or other arguments that could conceivably change my view significantly, I think it's unlikely (~10%, reported in [2]) that doing so would lead me to change my view significantly.[4]

It seems to me that the most probable way that my view could be changed to believe that (e.g.) OPP ought to prioritize WAS would be to persuade me that I should adopt a certain view on how to deal with moral uncertainty that would, if adopted, imply that OPP ought to prioritize WAS even given my current beliefs about how much I value the suffering of mosquitoes relative to other things (e.g. the lives of humans).

Is there a case to be made for prioritizing WAS if one assigns even a small probability (e.g. 1%) to a negative utilitarian-like view being correct given that they also subscribe to certain plausible views on moral uncertainty?

My views on how to deal with moral uncertainty are very underdeveloped. I think I currently have a tendency to evaluate situations or decide on actions on the basis of the moral view I deem most probable, however as the linked LessWrong wiki article points out, this has potential problems. (I'm also not aware of a less problematic view, so I will probably continue to do this until I encounter something else that appeals to me more. Bostrom's parliamentary model seems like a reasonable candidate, although I'm unsure how this negotiation process works exactly or would play out. Would have to think about it more.

Lastly, let me just note that I don't challenge the non-normative factual claims of the OP. Rather, I'm simply stating that my hesitation to take the view that OPP should prioritize WAS comes from my belief that I value things significantly differently than I would have to in order for WAS to be something that OPP should prioritize.


{1] A similar question was asked in the Effective Altruism Facebook group. My version gets at how much one values the life of a typical person today relative to the life of a typical mosquito rather than how much one values extreme pleasure relative to extreme suffering.

[2] Since I'm asking for others' answers, I should estimate my own answer. Hmm. If I had to make the decision right now I would choose to create the new human life, even if the number of painful mosquito deaths I was offered to prevent was infinite. Although note that I am not completely confident in this view, perhaps only ~60%. Then maybe ~30% to 10^10-infinity and ~10% to <10^10 mosquitoes, where practically all of that 10% uncertainty comes from the possibility that a more enlightened version of myself would undergo a paradigm shift or significant change in my fundamental values / moral views. In other words, I'm pretty uncertain (~40/60) about whether mosquitoes are net negative or not, but I'm pretty certain (~75%=30%/40%) that if I do value them negatively that the magnitude of their negative value is quite small (e.g. relative to the positive value I place on (the conscious experience of) human life).

[3] Knowing that my view is controversial among EAs (see the link at [1]), perhaps I should meta-update significantly towards the consensus view that not only is the existence of suffering inherently bad, but it's also a much greater magnitude bad than I think in the ~30% scenario that it is. I'll refrain from doing this for now, or figuring out how much I should update if I only think there's an X% that it's proper to update. (I'm also not sure how much my intuitions / current reported estimates already take into account others estimates or not.)

[4] The basis of my view that the goodness of a human life is much greater than the possible (~40% in my view) badness of a mosquito's suffering or painful death (and the basis of more general versions of this view) is my intuition. Thinking about the question from different angles I have been unable to shift my view significantly towards placing substantially more value on mosquitoes' significance or preventing mosquito suffering.

Even if you discount insects that heavily (which I believe is wrong), there's still a strong case to be made for trying to prevent wild vertebrates from suffering.

Hmm. I do believe I discount vertebrates much less than I discount insects, however I also think there's a huge difference between say chickens and chimpanzees or chimpanzees and humans. Even among humans (who have quite similar brains to one another compared to inter-comparisons), I think that the top 10% of Americans probably live lives that I value inherently (by which I mean ignoring the effects that they have on other things and only counting the quality of their conscious life experience) at least one order of magnitude (if not several) more than the bottom 10% of Americans. I believe this is an unpopular view also, but one consideration I might be able to give in support of it is if you reflect on how much you value your own conscious experience during some parts of your life compared to others you may find as I do that some moments or short periods seem to be of much greater value than others of equal duration.

An exercise I tried recently was making a plot of "value realized / time" vs "time" for my own conscious life experience (so again: not including the effects of my actions, which is the vast majority of what I value) and found that there were some years I valued multiple times more than other years and some moments I valued many times more than all years on net. The graph was also all positive value and trending upwards. Sleeping much less than awake. (I don't think I have very vivid dreams relative to others, but even if I did, I would probably still tend to value waking moments much more than sleeping ones.) Also, remembering or reflecting on great moments fondly can be of high value too in my evaluation. There's also the problem of not knowing now what certain experiences were like in the past to actually experience them since I'm relying on my memory of what they were like, which for all I know could be faulty. I think in general I choose to value experiences based on how I remember them being rather than how I think they were when I lived them (if there is a discrepancy between the two).

Also note that I'm a moral anti-realist and so I don't think there are correct answers, so to a certain extent how much I value some periods of my conscious life experience relative to others is a choice, since I don't believe that there are completely defined definite values that are mine that I can discover either.

A general thing I'd be really interested in seeing is peoples' estimates of how much they value (whether positively or negatively) the total life experiences of say, mosquitoes, X, Y, Z, chickens, cows, humans (and what that distribution looks like), oneself over time, a typical human over time, etc. And also "What would a graph of (value realized per unit time) vs (time) look like for Earth's history?" which would answer the question "How much value has been realized since life began on Earth?" (note: I'd ignore estimates of value realized elsewhere in the universe, which may actually be quite significant, for the sake of the question). If you'd like to indulge me on your own views on an of this I would be very interested, but of course no need if you don't want to. I'll estimate and write my own answers up sometime.