I'll suggest a reconceptualization that may seem radical in theory but is conservative in practice.

It doesn't seem conservative in practice? Like Vasco, I'd be surprised if aiming for reliable global capacity growth would look like the current GHD portfolio. For example:

1. Given an inability to help everyone, you'd want to target interventions based on people's future ability to contribute. (E.g. you should probably stop any interventions that target people in extreme poverty.)
2. You'd either want to stop focusing on infant mortality, or start interventions to increase fertility. (Depending on whether population growth is a priority.)
3. You'd want to invest more in education than would be suggested by typical metrics like QALYs or income doublings.

I'd guess most proponents of GHD would find (1) and (2) particularly bad.

Thanks for these ideas, this is an interesting perspective.

I'm a little uncertain about one of your baseline assumptions here.

"We're told that how to weigh these cause areas against each other "hinge[s] on very debatable, uncertain questions." (True enough!) But my impression is that EAs often take the relevant questions to be something like, should we be speciesist? and should we only care about present beings?  Neither of which strikes me as especially uncertain (though I know others disagree)."

I think I disagree with this framing and/or perhaps there might be a bit of unintentional strawmanning here? Can you point out the EAs or EA arguments (perhaps on the forum) that distinguish between the strength of these worldviews that are explicitly speciesist? Or only care about present beings? 

Personally I'm focused largely on GHD (while deeply respecting other worldviews) not because I'm speciest, but because I currently think the experience of being human might be many orders of magnitude more valuable than any other animal (I reject hedonism), and also that even assuming hedonism I'm not yet convinced by Rethink Priorities amazing research which places the moral weights of pigs, chickens and other animals extremely close to humans. Of course you could argue I think that because I'm a biased speciest human and you might be right - but that's not my intention.

And I do care about both present and future human beings, and am into longtermism as a concept, but am dubious right now about our ability to predictably and positively influence the long-term future, especially in the field of AI given EA's track record so far - with the exception of policy and advocacy work by EAs which I think has been hugely valuable.

Others may have very different (but valid) reasons than me for distinguishing between the importance of these worldviews but I'm not sure that you are right when you say "EAs often take the relevant questions to be something like, should we be speciesist? and should we only care about present beings?  Neither of which strikes me as especially uncertain (though I know others disagree)."

I probably agree with OpenPhil that there are indeed a range of important, uncertain questions different from the somewhat obvious ones you stated which can swing someone between caring more about current humans, animals or longtermism.

But even setting this issue aside, I see merit in your framework as well.

I would be interested to see if other people thing that those two not "especially uncertain questions" are what pushes people towards one or another worldview.

Nice post, Richard!

Global health & development is actually philosophically defensible, and shouldn't necessarily be swamped by either x-risk reduction or animal welfare. 

I think it would be a little bit of a surprising and suspicious convergence if the best interventions to improve human health (e.g. GiveWell's top charities) were also the best to reliably improve global capacity. Some areas which look pretty good to me on this worldview:

• AI governance and coordination.
• Global priorities research.
• Improving decision making (especially in important institutions).
• Improving individual reasoning or cognition.
• Safeguarding liberal democracy.

Maybe this is a nitpick, but I wonder whether it would be better to say "global human health and development/wellbeing" instead of "global health and development/wellbeing" whenever animals are not being considered. I estimated the scale of the annual welfare of all farmed animals is 4.64 times than of all humans, and that of all wild animals 50.8 M times that of all humans.

I agree that the "longtermism"/"near-termism" is a bad description of the true splits in EA. However I think that your proposed replacements could end up imposing one worldview on a movement that is usually a broad tent. 

You might not think speciesm is justified, but there are plenty of philosophers who disagree. If someone cares about saving human lives, without caring overmuch if they go on to be productive, should they be shunned from the movement? 

I think the advantage of a label like "Global health and development" is that is doesn't require a super specific worldview: you make your own assumptions about what you value, then you can decide for yourself whether GHD works as a cause area for you, based on the evidence presented. 

If I were picking categories, I'd simply be more specific with categories, and then further subdivide them into "speculative" and "grounded" based on their level of weirdness. 

Grounded GHD would be malaria nets, speculative GHD would be, like, charter cities or something

Grounded non-human suffering reduction would be reducing factory farming, speculative non-human suffering reduction looks at shrimp suffering

Grounded X-risk/catastrophe reduction would be pandemic prevention, speculative x-risk/catastrophe reduction would be malevolent hyper-powered AI attacks. 

Thanks for this! I agree that apart from speciesism, there isn't a good reason to prioritize GHD over animal welfare if targeting suffering reduction (or just directly helping others).

Would you mind expanding further on the goals of the "reliable global capacity growth" cause bucket? It seems to me that several traditionally longtermist / uncategorized cause areas could fit into this bucket, such as:

• Improving institutional decisionmaking
• Accelerating economic growth

Under your categorization, would these be included in GHD?

It also seems that some traditionally GHD charities would fall into the "suffering reduction" bucket, since their impact is focused on directly helping others:

• Fistula Foundation
• StrongMinds

Under your categorization, would these be included in animal welfare?

Also, would you recommend that GHD charity evaluators more explicitly change their optimization target from metrics which measure directly helping others / suffering reduction (QALYs, WELLBYs) to "global capacity growth" metrics? What might these metrics look like?

(Minor edits.)

Some possibilities to consider in >100 years that could undermine the reliability of any long-term positive effects from helping the global poor:

1. Marginal human labour will (quite probably, but not with overwhelming probability?) have no, very little or even negative productive value due to work being fully or nearly fully automated. The bottlenecks to production will just be bottlenecks for automation, e.g. compute, natural resources, energy, technology, physical limits and the preferences of agents that decide how automation is used. Humans will compete over the little share of useful work they can do, and may even push to do more work that will in fact be counterproductive and better off automated. What we do today doesn't really affect long-run productive capacity with substantial probability (or may even be negative for it). This may be especially true for people and their descendants who are less likely to advance the technological frontier in the next 100 years, like the global poor.
2. Biological humans (and our largely biological or more biological-like descendants) may compete over resources with artificial moral patients who can generate personal welfare value more efficiently per unit of resources. Saving human lives today and increasing the long-run biological(-like) population could mean fewer such more welfare-efficient moral patients. And that could be bad.
   1. OTOH, maybe with more humans, we exploit more resources than otherwise without competing, because we aren't exploiting resources as aggressively as possible or as aggressively as is not too objectionable to future moral agents (because future moral agents don't find it useful enough). So, this seems speculative either way.
   2. How much artificial moral patients matter is partly a normative/moral issue, not just an empirical one, so you could split across views for the relative weight of artificial moral patients. But those views could come to opposite conclusions on this issue and end up needing to cooperate to do something else to avoid acting against one another.

As a quick comment, I think something else that distinguishes GHD and animal welfare is that the global non-EA GHD community seems to me the most mature and "EA-like" of any of the non-EA analogues in other fields. It's probably the one that requires the most modest departure from conventional wisdom to justify it.

Thanks for this! I would be curious to know what you think about the tension there seems to be between allocating resources to Global health & development (or even prioritizing it over Animal Welfare) and rejecting speciesism given The Meat eater problem.

Quoting this paragraph and bolding the bit that I want to discuss:

Insofar as the GHD bucket is really motivated by something like sticking close to common sense, "neartermism" turns out to be the wrong label for this. Neartermism may mandate prioritizing aggregate shrimp over poor people; common sense certainly does not. When the two come apart, we should give more weight to the possibility that (as-yet-unidentified) good principles support the common-sense worldview. So we should be especially cautious of completely dismissing commonsense priorities in a worldview-diversified portfolio (even as we give significant weight and support to a range of theoretically well-supported counterintuitive cause areas).

I think the intuition here is that sometimes we should trust the output of common sense, even if the process to get there seems wrong. In general this is a pretty good and often underappreciated heuristic, but I think that's really only because in many cases the output of common sense will have been subject to some kind of alternative pressure towards accuracy, as in the seemingly-excessive traditional process to prepare manioc that in fact destroyed cyanide in it, despite the people in question having no ability to test for cyanide. The lesson is that sometimes heuristics and practices succeed, and are selected for success, by processes unrelated to the explicit justfications given for them.

But there has to be some kind of connection between the idea being good and it being preserved in common sense. If bad ideas and good ideas are equally "memetically fit", then common sense is no more likely to be true than any other hypothesis.

I think choices about how to give to charity are like this. The feedback mechanisms don't exist, either via obvious, conscious routes, nor any other route that I can think of. So I find it plausible that if the explicit argument for GHD fails, there's no other reason to privilege GHD in particular over other cause areas, no reason to expect an underlying unknown good principle to exist after all.

(Of course, just because I can't think of a route, doesn't mean there isn't one. But I think it merits skepticism.)

greater confidence in EEV lends itself to supporting longshots to reduce x-risk or otherwise seek to improve the long-term future in a highly targeted, deliberate way.

This just depends on what you think those EEVs are. Long-serving EAs tend to lean towards thinking that targeted efforts towards the far future have higher payoff, but that also has a strong selection effect. I know many smart people with totalising consequentialist sympathies who are sceptical enough of the far future that they prefer to donate to GHD causes. None of them are at all active in the EA movement, and I don't think that's coincidence.

why do you think that the worldviews need strong philosophical justification? it seems like this may leave out the vast majority of worldviews.

1. The massive error bars around how animal well-being/suffering compares to that of humans means it's an unreliable approach to reducing suffering.

2. Global development is a prerequisite for a lot of animal welfare work. People struggling to survive don't have time to care about the wellbeing of their food.

Love the post, don't love the names given.

I think "capacity growth" is a bit too vague, something like "tractable, common-sense global interventions" seems better.

I also think "moonshots" is a bit derogatory, something like "speculative, high-uncertainty causes" seems better.

This is an interesting and thoughtful post. 

One query: to me, the choice to label GHD as "reliable human capacity growth" conflicts with the idea you state of GHD sticking to more common-sense/empirically-grounded ideas of doing good. 

Isn't the capacity growth argument presuming a belief in the importance of long-run effects/longtermism? A common-sense view on this to me feels closer to a time discounting argument (the future is too uncertain so we help people as best we can within a timeframe that we can have some reasonable level of confidence that we affect).

Another relevant post not already shared here, especially for life-saving interventions: Assessing the case for population growth as a priority.

I have a piece I'm writing with some similar notes to this, may I send it to you when I'm done?

Relatedly, I have just published a post on Helping animals or saving human lives in high income countries seems better than saving human lives in low income countries?.

Summary

• I think the following will tend to be the best to maximise the cost-effectiveness of saving a human life:
  • Accounting solely for the benefits to the person saved, saving human lives in countries with low, but not too low, real gross domestic product (real GDP) per capita. Saving a human life is cheaper in lower income countries, but self-reported life satisfaction and life expectancy decrease with income. So saving human lives in the lowest income countries may be suboptimal.
  • Accounting solely for the benefits from economic growth, saving human lives in countries with high real GDP per capita. Saving a human life is more expensive in higher income countries, but these have greater productivity and life expectancy. So paying more to save human lives there may be worth it.
• In terms of effects on animals, I consider:
  • It is unclear whether saving human lives has a positive/negative impact on nearterm welfare accounting for effects on both humans and animals.
  • Saving human lives in high income countries is better than in low income countries if the benefits from economic growth dominate. The former increases animal suffering nearterm more, but makes it peak and end earlier, such that there is a greater overall reduction.
  • Eating less animals and more whole-food plant-based increases economic growth via decreased mortality.
• My overall view is that:
  • If improving nearterm welfare is the best proxy to maximise future welfare, helping animals seems better than saving human lives in low income countries.
  • If boosting economic growth is the best proxy, saving human lives in high income countries seems better than in low income countries.
  • I guess improving nearterm welfare is a better proxy than boosting economic growth. Nonetheless, I am mainly in favour of research on whether indirect longterm effects dominate, and, if so, on which are the best proxies for them.

(1) fetal anesthesia as a cause area intuitively belongs with 'animal welfare' rather than 'global health & development', even though fetuses are human.

It seems like about half the country disagrees with that intuition?

I like the general framing & your identification of the need for more defensibility in these definitions. As someone more inclined toward GHD, I don't do so because I see it as a more likely way of ensuring flow-through value of future lives, but I still do place moral weight on future lives. 

My take tends more toward (with the requisite uncertainty) not focusing on longtermist causes because I think they might be completely intractable, and as such we're better off focusing on suffering reduction in the present and the near-term future (~100 years). Similar thinking around wild animal suffering (although I wholeheartedly support improvements for farmed animals and think it could be tractable).

That feels different to your framing? But perhaps all that intractability is just a factor of my lower EEV estimates and/or the confidence in them?

I'd be curious if you have any thoughts on how your proposed refactoring from [neartermist human-only / neartermist incl. AW / longtermist] -> [pure suffering reduction / reliable global capacity growth / moonshots] might change, in broad strokes (i.e. direction & OOM change), current 

• funding allocation (proxies: Maule's funding by cause area, McDoodles)
• career advising (proxies: 80K problems and skills, Probably Good profiles)

Or maybe these are not the right questions to ask / I'm looking at the wrong things, since you seem to be mainly aiming at research (re)prioritisation? 

I do agree with 

we should be especially cautious of completely dismissing commonsense priorities in a worldview-diversified portfolio (even as we give significant weight and support to a range of theoretically well-supported counterintuitive cause areas) 

although I thought the sandboxing of cluster thinking (vs sequence thinking) handles that just fine:

A key difference with “sequence thinking” is the handling of certainty/robustness (by which I mean the opposite of Knightian uncertainty) associated with each perspective. Perspectives associated with high uncertainty are in some sense “sandboxed” in cluster thinking: they are stopped from carrying strong weight in the final decision, even when such perspectives involve extreme claims (e.g., a low-certainty argument that “animal welfare is 100,000x as promising a cause as global poverty” receives no more weight than if it were an argument that “animal welfare is 10x as promising a cause as global poverty”).

I really like this post, but I think the concept of buckets is a mistake. It implies that a cause has a discrete impact and "scores zero" on the other 2 dimensions, while in reality some causes might do well on 2 dimensions (or at least non-zero).

I also think over time, the community has moved more towards doing vs. donating, which has brought in a lot of practical constraints. For individuals this could be:

• "what am I good at?"
• "what motivates me?"
• "what will my family think of me?"

And also for the community:

• "which causes can we convince outsiders to go into?"
• "which spread in causes do we need to not look too weird and demonstrate that our principles work?"

It can be difficult to separate these from the more universal moral and epistemic considerations, once someone is in a committed career path.

Would you be up for spelling out the problem of "lacks adequate philosophical foundations"?

What criteria need to be satisfied for the foundations to be adequate, to your mind?

Do they e.g. include consequentialism and a strong form of impartiality?

(Minor edits.)

In terms of the underlying worldview differences, I think the key questions are something like:

(i) How confident should we be in our explicit expected value estimates?  How strongly should we discount highly speculative endeavors, relative to "commonsense" do-gooding?

(ii) How does the total (intrinsic + instrumental) value of improving human lives & capacities compare to the total (intrinsic) value of pure suffering reduction?

[Aside: I think it's much more reasonable to be uncertain about these (largely empirical) questions than about the (largely moral) questions that underpin the orthodox breakdown of EA worldviews.]

Shouldn't you just set a probability distribution^[1] over those stances^[2] and allow the obvious comparisons between them? We're estimating the impact of an intervention, and we could have different stances on how to estimate impact, including how much to discount and different priors, but as long as impact is measured in the same terms, e.g. the same utility function over which we maximize expected utility, as a function of the welfares of moral patients and/or other things, we can just treat this like uncertainty about the effects of our interventions. For expected utility maximizers, we could just take the weighted average across the stances when we take the expected utility.

If you can't put a probability distribution over those stances, then I don't know how you could ground any split of resources across them (although you could still pick portfolios that aren't robustly dominated by any other). Approaches for handling normative uncertainty generally require such distributions

I would say a better place to break intertheoretic comparisons (and form buckets, if going for a bucket approach) would be different attitudes towards risk and ambiguity, i.e. different degrees and versions of risk aversion, difference-making risk aversion and ambiguity aversion. And of course between moral stances, like variants of utilitarianism, alternatives to utilitarianism, the normative parts of moral weights, etc..

GHD looks most plausible as a priority with at least modest difference-making risk aversion or ambiguity aversion, or with certain person-affecting or non-aggregative views. Otherwise some x-risk work will likely end up beating it.

1. ^{^}

   Or multiple, with imprecise probabilities.

2. ^{^}

   Especially for (i), maybe less so for (ii), given moral uncertainty about the moral weights of nonhuman animals.

To help carve out the space where GiveWell recommendations could fit, considering flow-through effects:

Assuming some kind of total utilitarianism or similar, if you don't give far greater intrinsic value to the interests of humans over nonhuman animals in practice, you'd need to consider flow-through effects over multiple centuries or interactions with upcoming technologies (and speculate about those) to make much difference to the value of GiveWell recommendations and for them to beat animal welfare interventions. For the average life we could save soon, it seems unlikely that the instrumental value to other people during that life (assuming a life of <120 years) is >10x greater than the value to the person whose life is saved (or to their immediate family), unless they have many descendants over their lifetime or their life becomes dramatically different from the current lives of the adults around them (dramatic increases in the productivity of their labour^[1], dramatic life extension^[2]).

EDIT: Maybe GiveWell charity beneficiaries will have enough children and descendants over the next 100 years to make enough difference. Even if fertility rates decrease over time, they're still high in many low-income countries, some of which GiveWell charities work in. If you save one life, how many additional years of human life, potentially valuable in itself and with instrumental value for others, will there be over the next 100 years?

Also, you'd want to consider negative flow-through effects of GiveWell charity interventions, and positive and negative flow-through effects of animal welfare interventions.

1. ^{^}

    E.g. from rapid economic growth. However, AI could instead entirely replace their labour.

2. ^{^}

   Which could increase how much they would work over their lifetime and also directly increase the value of life-saving interventions to the person whose life is saved, because they could end up living much longer.