One purpose of this blog is to be explicit about some of the deep value judgments and worldviews that underlie our analysis and recommendations. As we raise the priority of expanding our research into new causes, this seems like a good time to lay out some of the things we believe – and some of the things we’re unsure about – on topics that could be of fundamental importance for the question of where to give.
In general, the below statements broadly describe the values of the GiveWell staff who have final say over our research. There may be cases in which different individuals would give different levels of weight/confidence to the various statements than I have, but at a high level we expect these statements to be a reasonably good guide to the values underlying GiveWell’s research.
Values
We don’t believe it would be productive to try to produce a complete explicit characterization of the fundamental values that guide our giving recommendations, but we think it’s worth noting some things about them.
- We are global humanitarians, believing that human lives have equal intrinsic value regardless of nationality, ethnicity, etc. We do believe there may be cases where helping some people will create more positive indirect effects than helping others (for example, I stated in 2009 that I preferred helping people in urban areas for this reason, though this represents my view and not necessarily the view of others at GiveWell). However, we do not agree with the principle that “giving begins at home”: we do not assign more moral importance to people in our communities and in our country than to others.
- The primary things we value are reducing suffering and tragic death and improving humans’ control over their lives and self-actualization.
We also place value on reducing animals’ suffering, though substantially less than on human suffering.We also place value on reducing animals’ suffering, though our guess is that the type of suffering animals experience is of a kind that we would not weigh as heavily as the type of suffering that humans experience. (We do not have clear consensus views on how to weigh these things against each other.) This bullet point edited for clarity on Sep. 5, 2013. - We do not put strong weight on “achievements” (artistic endeavors, space exploration, etc.) as ends in themselves, though these may contribute to the things we do value (details above). We also don’t put strong weight on things like “justice,” “equality,” “fairness,” etc. as ends in themselves (though again, these may contribute to the things we do value).
- We are broadly consequentialist: we value actions according to their consequences.
- We are operating broadly in an “expected value” framework; we are seeking to “accomplish as much good as possible in expectation,” not to “ensure that we do no harm” or “maximize the chance that we do some good.”
There are many questions that we do not have internal consensus on, or are individually unsure of the answers to, such as
- How should one value increasing empowerment vs. reducing suffering vs. averting deaths?
- How should one value
animal suffering in comparison to human sufferinghelping animals in comparison to helping humans? This line edited for clarity on Sep. 5, 2013. - Is it better to bring someone’s quality of life from “extremely poor” to “poor” or from “good” to “extremely good?”
- Is creating a new life a good thing? Can it be a bad thing? How “desirable” or “undesirable” must the life be for its creation to count as a good/bad thing? Should we value “allowing future lives to exist that would never come into existence otherwise” similarly to “lives saved?”
- Is it better to save the life of a five-year-old or fifteen-year-old?
We don’t believe it is practically possible to come to confident views on these sorts of questions. We also aren’t convinced it is necessary. We haven’t encountered situations in which further thought on these questions would be likely to dramatically change our giving recommendations. When we have noticed a dependency, we’ve highlighted it and encouraged donors to draw their own conclusions.
Worldview
We view the questions in the previous section as being largely “fundamental,” in that empirical inquiry seems unlikely to shift one’s views on them. By contrast, this section discusses views we have that largely come down to empirical beliefs about the world, but are very wide-ranging in their consequences (and thus in their predictions).
There are two broad worldview characteristics that seem, so far, to lie at the heart of many of our disagreements with others who have similar values.
1. We are relatively skeptical. When a claim is made that a giving opportunity can have high impact, our default reaction is to doubt the claim, even when we don’t immediately see a specific reason to do so. We believe (based partly on our experiences investigating charities) that most claims become less impressive on further scrutiny (and the more impressive they appear initially, the steeper the adjustment that happens on further scrutiny). As a result, we tend to believe that we will accomplish more good by recommending giving opportunities we understand relatively well than by recommending giving opportunities that we understand poorly and look more impressive from a distance.
We have written about this aspect of our worldview previously, and have done some rudimentary work on formalizing its consequences:
- A Conflict of Bayesian Priors? lays out the basic fact that we have a skeptical prior (by default, we expect that a strong claim will not hold up to scrutiny).
- Why We Can’t Take Expected-Value Estimates Literally does some basic formalization of this aspect of worldview and explores some of the consequences, defending our general preference for giving where we have strong evidence that donations can do a lot of good rather than where we have weak evidence that donations can do far more good. It also explains why we put only limited weight on formal, explicit calculations of “expected lives saved” and similar metrics.
- Maximizing cost-effectiveness via critical inquiry expands on this framework, laying out how it can be vital to understand a giving opportunity “from multiple angles.”
- We will likely post more in the future on this topic.
2. We believe that further economic development, and general human empowerment, is likely to be substantially net positive, and that it is likely to lead to improvement on many dimensions in unexpected ways. This is a view we haven’t written about before, and it has strong implications for what causes to investigate. While we see great value in directly helping the poorest of the poor, we’re also open to the viewpoint that contributing to general economic development may have substantial benefits for the poorest of the poor (and for the rest of the world). And while we are open to arguments that particular issues (such as climate change) are particularly important to the future of humanity, we also believe that by default, we should expect contributions to economic development and human empowerment to be positive for the future of humanity; we don’t feel that one must necessarily choose between improving lives in the short and long term. (This view is part of why we put more weight on helping humans than on helping animals.)
Because of this view, we are open to outstanding giving opportunities across a wide variety of causes; we aren’t convinced that the best opportunities must be in developing-world aid, or mitigation of global catastrophic risks, or any other particular area. Even if a particular problem is, in some sense, the “most important,” it may be possible to accomplish more good by working in another cause where there is better room for more funding. We will discuss this view more in a future post.
Comments
Re: economic development. How much have you investigated free trade/free immigration?
Nice summary!
Sam, we’re potentially interested in both issues but so far have done only preliminary investigation of the latter and no investigation (aside from informal conversations) about the former. In both cases we’d prefer to learn more about the general practice of political advocacy first.
I mostly agree with your positions. But definitely not about the following point.
You ask : Is it better to bring someone’s quality of life from “extremely poor” to “poor” or from “good” to “extremely good?”
And you say : We don’t believe it is practically possible to come to confident views on these sorts of questions.
On this specific question, I wonder what prevents you from coming to the confident view that it is incomparably better to go from hell to purgatory than from first to seventh heaven!
You later add : While we see great value in directly helping the poorest of the poor, we’re also open to the viewpoint that contributing to general economic development may have substantial benefits for the poorest of the poor (and for the rest of the world).
Again, I agree. But for the same price, I’d rather contribute to general alleviation of extreme poverty than general economic development.
I should have added that contributing to alleviation of extreme poverty may have substantial benefits for the richest of the rich, and for the rest of the world.
I agree that it is important for charity evaluators to be explicit about the value judgments that underlie their recommendations, and I thank you for writing this post.
When you say that you value all human lives equally, do you mean that you attach equal intrinsic value to all members of the species Homo sapiens? Or are you using ‘human’ to refer to a class of beings defined by properties other than species membership?
Great post Holden!
“We also place value on reducing animals’ suffering, though substantially less than on human suffering.” – Since animal charities are so many orders of magnitude more effective than human ones, it would be awesome to have GiveWell clarify this area. My guess is that even if you weight the suffering of non-human animals 50 or 100 times less than humans, animal charities are still the best.
(Peter has some calculations here: http://www.greatplay.net/vegan-outreach-cost-effectiveness-calculator)
Ben makes a good point. It is, of course, understandable to want to favor your group (nationality, race, species). But effective animal advocacy can prevent so much suffering for so little money that it is almost impossible to ignore. How many animals would you condemn to a life of factory farming (with a significant percentage suffering to death — http://bit.ly/10zid75 ) to help prevent one human from getting malaria? 10? 1,000? 1,000,000? At some point, it seems to me, we need to answer that question.
I agree with Robert and Ben above.
An analogy: Earth is like a house. We ought to repair the leaky roof before we remodel the outdated yet functional bathroom.
Similarly, we should concentrate on “fixing” the root causes of suffering rather than on relatively superficial improvements in happiness or wellbeing. The degree of difference between extremely poor to poor is greater than the difference between good and extremely good.
First off, thank you for the wonderful post, and for the amazing blog here!
(1) Do you use or believe in any time discounting of deaths averted? Discounting leads to higher valuations of interventions that saves lives today, and arguably there isn’t enough funding going toward long-term development projects like infrastructure development, in part because the benefits are further in the future, but also due to other challenges, like capital requirements, political will, and difficulty of evaluation.
(2) Do you have good sources of reading/learning on the discussion of broad economic development policies vs targeting the poorest of the poor for equity reasons? This is a big debate that I am unresolved on and would love to get more insight into.
As someone who differs with many of your value judgments (I’m non-consequentialist, think animals are far more comparably valuable to humans, and care about justice and equality), I’d just like to note that I still gain a lot from following your work. Thanks for the post.
When you publish your top charities is there a score associated? i.e. How much is #1 better than #2? Or are they very close.
Is there an estimate as to how much you could be in error?
A certain kind of cynic would be counting “deaths postponed” rather than “lives saved”…
Thanks for the comments, all,
Pablo, I don’t think I’m entirely clear on what you’re asking. I don’t feel it’s productive to try to define “human” in a completely precise way that handles all corner cases, and I’m not aware of any major and frequently-relevant debates over e.g. what species count.
Ben and Matt, I have looked into animal welfare a small amount, including the figures you cite, and I don’t agree with your claims. I believe that if there are good opportunities in animal welfare, they are likely to be concentrated in high-risk, very-difficult-to-quantify domains. I don’t believe the analysis you cite is of comparable robustness to the cost-effectiveness estimates we’ve done for our top charities.
Brandon,
(1) We believe there are multiple good reasons to use discount rates, even if one believes the intrinsic value of future lives to be equal to that of present lives. (For example, good done today compounds more.) What specific discount rate one should use depends on the context. We are certainly open to long-term projects, however, and don’t believe future lives to be intrinsically less valuable than present lives.
(2) I haven’t found any literature that I feel sheds much light on the question of whether it’s more effective to help the poorest of the poor directly or to help indirectly by contributing to broader economic development.
Rahul, we publish suggested allocations (see the right-hand bar of our top charities page) and we have also published a good deal of discussion of the relative merits.
Holden – thanks for the response. I am glad to hear you have investigated the efficacy of helping nonhumans.
Have you posted the reasons why you disbelieve the apparent results anywhere? If not, would you consider doing so?
I agree that the data isn’t as firm as with, say, AMF, (otherwise we wouldn’t need you to look into it!) but it surprises me to hear you think it’s so bad it shouldn’t be considered for future investigation, even givewell labs.
Ben, we advised an individual who looked into these estimates and who is planning to write up his work publicly in the near future.
Holden, may I ask you again : what prevents you from coming to a confident view on the question “Is it better to bring someone’s quality of life from “extremely poor” to “poor” or from “good” to “extremely good?”” ?
Robert, what reasons do you perceive for believing the answer is clear?
First, I am pretty sure that anyone who has experienced like me extremely poor, poor, good and extremely good quality of life knows clearly that answer. A way to express it is to say, for instance, that one is much more anxious to get out from the worst, even for a somewhat bad situation, than to reach the best from a good position. Putting oneself into excessive suffering, even for a few seconds, should convince everyone of that.
Second, as Elaine says here above, such priorities are mere common sense. They should be as clear when it is question of economy, ecology, eco-bio-psychology, etc. as they are (or as analogous priorities are) when speaking of a house.
Third, in many instances of philosophic or political or moral literature, the answer is more or less taken for granted. For instance, Karl Popper in “The Open Society and Its Enemies” : “”I believe that there is, from the ethical point of view, no symmetry between suffering and happiness (…) human suffering makes a direct moral appeal for help, while there is no similar call to increase the happiness of a man who is doing well anyway.”
For what it’s worth, I agree with Robert at the level of values, though I don’t think the dispute is amenable to empirical (or perhaps even philosophical) resolution. I think Holden’s expressing (openness to) a roughly utilitarian view while Robert is arguing for a variant of prioritarianism. I identify more with the prioritarian strand of thinking but I agree with Holden that the choice is non-obvious.
Robert, I don’t agree that the answer is so clear. I don’t think it’s safe to extrapolate from your own preferences, or from the statements of philosophers, to the preferences of – or the proper ethics over – the population as a whole.
As a side note, Popper’s quote refers to “happiness” whereas my line refers to overall quality of life, a broader concept that may include things like “empowerment” as well as happiness. I find the idea of helping someone to reach their full potential more appealing than the idea of making someone extremely happy. But I don’t believe any of these decisions have obvious right answers.
Holden, I understand that you don’t find MY reasons compelling but I am asking what prevents YOU from coming to a confident view on the question. We are not talking about the population as a whole (whose preferences or ethics are seemingly the cause of a whole lot of problems in the world) but about individuals in GiveWell or GiveWell as a group. I am all the more puzzled by this question because your recommended top charities are all ‘prioritarian’, aren’t they?
I fully agree with you about ‘happiness’ (or hedonic utilitarianism, or even utilitarianism) being a too narrow concept. But again, why is it not obvious to you that it is better to save the one lost sheep (as says the parable), i.e. to help someone to get out of a potentially disastrous hellish depression to reach a state of slight disability, than to help someone, who is slightly well functioning, to reach a gratifying do-gooder full potential?
If there is agreement among the personnel at GW that it is a reality that the worse conditions and experiences of human life are more INTENSE than the best conditions and experiences, it could simplify the rethinking about your future.
If that inherent intensity factor doesn’t settle the value question: “Is it better to bring someone’s quality of life from ‘extremely poor’ to ‘poor’ or from ‘good’ to ‘extremely good?'” Consider that there’s a third possibility. To bring an individual(s) or community “across the point” between the bad and into the good no matter how small or large the number of degrees of change is the best thing to do.
Robert, by default, any given claim isn’t obvious to me. The burden of argumentation is on the person who would claim that a question has a clear answer. I don’t see any of what I consider compelling reasons to consider the answer to this particular question to be clear.
While you may have a dominant preference for avoiding the worst life over pursuing the best, I don’t have that preference, and many of the people I know don’t (based on their actions) seem to either.
Thanks, Holden, what you say is very clear.
Obviously, I would prefer to be in a situation where I am pursuing the best life rather than avoiding the worst! But if it is question of helping someone,I would obviously prefer to help the worse off, all other things being equal. Why? Because I perceive that the need is greater and the effort more worthwhile.
If I understood you well, you are telling me that you, like most people, don’t have that preference. I will try to assimilate that information which changes a lot my outlook on why there is so much suffering in the world.
Robert, I’m not sure we disagree. Given the present state of the world, I have the intuition that there are greater returns to be had from helping the worst off than from helping others – because the worst off benefit more from the sorts of things (health, wealth) that philanthropy can help provide. GiveWell has indeed focused on helping the worst off and I think there’s a good argument for doing so.
The question of this blog post is a different one; it’s a question of “fundamental” values. If we put aside the returns to effort and asked whether we’d rather help people attain the best lives or avoid the worst – for the same amount of effort – I think the answer would not be clear. But this is the kind of question we generally don’t face in practice.
In practice we agree because fundamental values are left mostly unexamined. As a result however, and because IMO the very question that we are discussing here is not answered clearly, our practice is far from being sufficient. It will take a major shift in thinking for people to realize that reaching the best life for the farmer is less urgent than getting the sows out of the crates, to take an example that speaks to those here above who are concerned with animals. At Algosphere we are working on such a transformation of culture, and soon we hope to invite you, Holden, and others who might be interested, to consider what we are proposing.
Holden:
I think it would be great that you answer to this:
“How many animals would you condemn to a life of factory farming (with a significant percentage suffering to death ? http://bit.ly/10zid75 ) to help prevent one human from getting malaria? 10? 1,000? 1,000,000?”
I think you intentionally avoid to reply. But this question is very important.
I did intentionally refrain from responding to that question. We have very little internal consensus or even intra-person stability (at least in my case) regarding that question. A year from now, I could imagine myself answering “less than 1” or “More than a billion” or anything in between. To have a firmer view would take substantial thought and investigation; our general approach is to avoid getting into major investigations and debates regarding questions that aren’t highly action-relevant.
Hi Holden. Thank you for your sincerity.
It’s a bit shocking to me that you think that isn’t highly action-relevant, as you suggest. Billions of animals are horribly suffering each year, and a lot of suffering could be avoid giving to animal charities.
Daniel, it’s the second part of your statement that I have more questions about. I don’t believe there is any animal-welfare intervention comparable to our current top charities by our traditional criteria. It’s possible that political advocacy or other action could be effective based on a less rigorous case, and we are open to this possibility as we work on GiveWell Labs.
Holden writes:
I think it’s important to recognize, even in disputing it, that the “giving begins at home” principle is about the moral duty of the donor not (or at least not necessarily) the intrinsic worth of the recipient.
If that doesn’t seem obvious, consider for example the literal meaning of the expression: a moral duty towards one’s own family, which obviously doesn’t entail that one’s family is objectively better than others. Or consider the point of view of, say, a Belgian towards a wealthy Mexican who made his fortune in Mexico and feels that the Mexican therefore has a special duty to help poor Mexicans (even though it is more efficient for him to help poor people in Sub-Saharan Africa). Here, the Belgian’s moral intuition is not based on intrinsic superiority (in this case of Mexicans). Now of course one may think that the “charity begins at home” precept is wrong or even absurd, but my point is that one can nonetheless recognize that it is a common view and not one that presupposes that one group is superior to another.
There’s a second, epistemological, argument for a home bias in giving: in giving in one’s own society one better understands (or at least it’s plausible to believe that one better understands) the ultimate effect of one’s donations. I think that’s relatively uncontroversial (though of course one can argue semantics whether taking imperfect knowledge in account in this way is an example of bias to “beginning charity at home”).
GiveWell can get to their key conclusion here — that giving in say the US doesn’t make sense compared to the Third World — without necessarily denying the appropriateness of a charitable home bias. As you’ve convincingly argued, money given to say AMF has multiple orders of magnitude more impact per dollar than interventions in the US. This should be plenty to overwhelm all except the most extreme forms of home bias (both moral duty and epistemological aspects). In my view, that’s a more robust way to argue it. As it is now, if someone believes that “charity begins at home” and sees this, then it can be easily dismissed as starting from different moral intuitions.
On the question of whether it’s better to contribute to overall economic development or direct funds toward the specific needs of the very poor, William Easterly writes quite a bit on the subject in “White Man’s Burden” and with greater rigor in “The Elusive Quest for Growth.” The general conclusion is that efforts to transform whole economies have generally failed, while grassroots efforts have some successes that hold up to the sort of rigor GiveWell shines on them.
The question is about efficacy. Overall economic development would be preferable, we just don’t know how to how to impose it. We do know some specific grassroots interventions that work.
Colin, thanks for the comment. I appreciate your pointing out that the “charity begins at home” sentiment need not involve a belief in the objective moral superiority of one’s community. What I was trying to express is that we do not incorporate an extra moral obligation to people “at home” into our research and recommendations, either for reasons based on objective worth or for reasons based on moral duty. We do agree with the epistemological point, which is distinct from the “moral preference” issue.
Colin Rust wrote:
And Holden Karnofsky replied:
I’ve had similar thoughts.
Thinking about an organization such as Literacy Partners one can see the advantage of having teachers and tutors who are in the same city as the learners.
The instructors teaching people how to read can draw upon common experiences as New Yorkers.
They can help with the subway system, public transit, New York City social services, employment, medication, etc.
Cultural commonality between the donors, volunteers, staff members, mid-level management and high-level management could come into play at multiple levels: on-the ground execution, mid-level planning, high-level planning, etc.
Comments are closed.