Which charity does the most good?

What can you do that would have the best chance of making the world a better place? As Scott Siskind puts the question:

Most donors say they want to “help people”. If that’s true, they should try to distribute their resources to help people as much as possible. Most people don’t.

In the “Buy A Brushstroke” campaign, eleven thousand British donors gave a total of £550,000 to keep the famous painting “Blue Rigi” in a UK museum. If they had given that £550,000 to buy better sanitation systems in African villages instead, the latest statistics suggest it would have saved the lives of about one thousand two hundred people from disease. Each individual $50 donation could have given a year of normal life back to a Third Worlder afflicted with a disabling condition like blindness or limb deformity.

Most of those 11,000 donors genuinely wanted to help people by preserving access to the original canvas of a beautiful painting. And most of those 11,000 donors, if you asked, would say that a thousand people’s lives are more important than a beautiful painting, original or no. But these people didn’t have the proper mental habits to realize that was the choice before them, and so a beautiful painting remains in a British museum and somewhere in the Third World a thousand people are dead. […]

It is important to be rational about charity for the same reason it is important to be rational about Arctic exploration: it requires the same awareness of opportunity costs and the same hard-headed commitment to investigating efficient use of resources, and it may well be a matter of life and death.

Holden Karnofsky of GiveWell notes (in this video) that it isn’t easy to spot an ineffective charity. Many popular charities are “not even failing to do good, but doing harm”. At the same time, the positive difference you can make with a carefully targeted, empirically vetted charitable donation is extraordinary. Philosopher William MacAskill voices his excitement:

Imagine you’re walking down the street and see a building on fire. You run in, kick the door down—smoke billowing—you run in and save a young child. That would be a pretty amazing day in your life: That’s a day that would stay with you forever. Who wouldn’t want to have that experience? But the most effective charities can save a life for $4,000, so many of us are lucky enough that we can save a life every year through our donations. When you’re able to achieve so much at such low cost to yourself…why wouldn’t you do that? The only reason not to is that you’re stuck in the status quo, where giving away so much of your income seems a little bit odd.

GiveWell is the top organization investigating the impact charities have upon the most disadvantaged people in the world. If you want to be confident you’re really improving the world in a concrete way, really saving lives, it’s hard to do better than following GiveWell’s new annual giving recommendations (updated December 2014). The new recommendations are that each $100 you give to charity over the next 4 months break down as follows:

$60 – Against Malaria Foundation (AMF)

$12 – GiveDirectly

$12 – Schistosomiasis Control Initiative (SCI)

$10 – GiveWell

$6 – Deworm the World Initiative (DtWI)

(The $10 to GiveWell is an operating expenses donation GiveWell is requesting separately. I’m including it in the breakdown on the assumption that if you trust GiveWell’s expertise enough to base your decisions on their research, you probably also want to support GiveWell’s ability to keep those recommendations up to date.)

The above breakdown is intended to minimize the risk that, say, AMF keeps getting swamped with donations long after it’s reached its yearly target, while donors neglect DtWI. GiveWell’s goal is that AMF receive $5 million from individual donors over the next 4 months; GiveDirectly between $1 million and $25 million; SCI $1 million; and DtWI between $500,000 and $1 million. If everyone donates in the above proportion, then every top-effectiveness charity will be equally likely to hit its minimum target.

If you want to follow this breakdown exactly, go to https://givewell.secure.nonprofitsoapbox.com/donate-to-givewell and select “Grants to recommended charities (90%) and unrestricted (10)%” under “How should we use your gift?”. If you’d rather just donate to one organization and not split it up in this way, GiveWell suggests giving to the Against Malaria Foundation; you can do so by setting “How should we use your gift?” to “Grants to recommended charities” and writing under Comments “all to AMF”.

Edit 12/31: More specifically, Elie Hassenfeld of GiveWell writes:

For donors who have a high degree of trust in and alignment with GiveWell, we recommend unrestricted gifts to GiveWell. For donors who want to support our work because they value it but are otherwise primarily interested in supporting charities based on neutral recommendations, strong evidence, etc., we recommend giving 10% of their donation to GiveWell.

What do these charities do?

holdenin
GiveWell staff make a site visit to a charitable organization in western India.

AMF, GiveDirectly, SCI, and DtWI all focus on combating poverty and disease in poor regions of Africa and Asia. This isn’t an arbitrary choice; your dollar can go orders of magnitude farther in the developing world than in developed nations. Dylan Matthews of Vox writes:

GiveWell actually looked into a number of US charities, like the Nurse-Family Partnership program for infants, the KIPP chain of charter schools, and the HOPE job-training program. It found that all were highly effective, but far more cost intensive than the best foreign charities. KIPP and the Nurse-Family Partnership cost over $10,000 per child served, while deworming programs like SCI’s and Deworm the World’s generally cost about $0.50 per child treated.

AMF distributes insecticide-treated bed nets in the Democratic Republic of the Congo and other countries. This prevents transmission of malaria by mosquito bite, reducing child mortality and anemia and improving developmental outcomes. (General information on insecticide-treated nets.)

GiveDirectly makes secure cash payments to poor households they’ve vetted in Kenya and Uganda. Recipients may then use this money however they wish. This generally results in improved food security and investments with high rates of return. Direct cash transfers are a good way to avoid the common mistake of trying to micromanage the lives of people in the developing world. Impoverished individuals usually have much more robust and fine-grained knowledge of their own needs than any philanthropic organization or donor does, and they have clearer incentives to make sure every penny gets used wisely. (General information on cash transfers.)

SCI works with governments in sub-Saharan Africa to distribute deworming pills to schoolchildren, improving nutrition and developmental outcomes. DtWI does similar deworming work in India, Kenya, and Vietnam, with more focus on improving existing programs than on creating and scaling up programs. (General information on deworming.)

How do these charities compare to each other?

GiveWell publishes its evidence and reasoning process publicly so others can examine it in as much detail as they’d like and identify points of disagreement. That gives you a chance to deviate from GiveWell’s recommendations in an informed way, if you disagree with GiveWell about the tradeoffs involved. To summarize GiveWell’s take:

  • Cost-effectiveness: GiveDirectly is probably the least cost-effective, in spite of transferring 87 to 90 cents per dollar donated directly into the hands of poor individuals. This is because it still appears to be cheaper to cure the worst widespread diseases than to directly alleviate the poverty of otherwise healthy people. AMF and SCI are maybe 5-10 times as effective as GiveDirectly, and DtWI may be twice as effective as SCI.
  • Strength of supporting evidence: We can be relatively confident GiveDirectly is having the impact it intends to. The case for AMF is weaker, and the case for SCI is weaker still. DtWI has the weakest case, because its political focus places it more causal steps away from its goal. On the other hand, DtWI’s transparency and self-monitoring is much better than SCI’s, so there’s more likelihood we’ll notice in the future if DtWI has gone wrong than if SCI has.
  • History of rolling out more program: GiveDirectly and SCI have a strong track record. AMF and DtWI have an adequate track record.
  • Room for more funding: GiveDirectly is scaling up amazingly well, and could continue to make use of tens of millions more dollars this year. AMF has had difficulty finding enough places to distribute bed nets to use its funds effectively; however, it now appears to have fixed that problem and has a lot more room for funding it can use to leverage more distribution deals. DtWI and SCI have relatively little room for funding.

In their personal charitable donations, GiveWell staff generally followed the above recommendations, though several staffers gave substantially more to GiveDirectly (to reward its transparency and self-monitoring, and to be sure of having a positive impact), and less to the deworming charities. Other people who have explained how they’re factoring in GiveWell’s new recommendations include philosopher Richard Chappell, blogger Unit of Caring, consultant Chris Smith, and economist Robert Wiblin.

What are other contenders for the best causes out there?

If you’re interested in credible but less thoroughly vetted efforts to combat global poverty, you may want to look at GiveWell’s second tier of promising charities:

Following GiveWell’s recommendations is probably the best way to measurably improve the lives of human beings who are suffering and dying today. However, the same evidence-based approach should allow us to identify relatively effective and ineffective causes in the developed world too. GiveWell is in the early stages of looking for the most urgent and tractable projects in U.S. policy, and one of their top contenders is prison reform. If you live in the U.S. and are more interested in local issues, you may want to follow the work of:

On the other hand, there are some local, activism-oriented charities that may have a much larger impact than any I’ve listed so far — charities focused on non-human animal welfare. If you aren’t just worried about human suffering, you may want to give to:

  • The Humane League, a top-notch animal welfare nonprofit that discourages factory farming through outreach and advertising. They attempt to test the efficacy of their methods at Humane League Labs.

Another excellent way to try to outdo GiveWell’s recommended charities is to help fund scientific research into the life-saving innovations of the future. Historically, scientific and technological progress has had a vastly larger effect on human welfare than any philanthropy has, and this is another major area the Open Philanthropy Project hopes to investigate in the future. For now, the main scientific institute I can recommend donating to is:

  • The Future of Humanity Institute,  an Oxford-based research center that investigates social and technological changes that may impact our future as a species, as well as the effects of systematic uncertainty and bias on our attempts to predict such developments.

If there are interesting developments over the next year, I’ll update this advice December 2015. For now, the main organizations I recommend giving to are GiveWell and its top charities (donation page), the Humane League (donation page), or the Future of Humanity Institute (donation page), in increasing order of ‘uncertainty about the organization’s real effects’ and ‘probability of having a large positive impact’.

Edit 12/28: GiveWell has updated their donation page to include a “Grants to recommended charities (90%) and unrestricted (10)%” option. I’ve modified my above advice to make use of that new option. I’ve also started a birthday fundraiser to give to the charities I covered above.

Rejigger triggers?

From tumblr:

The reason why people on tumblr over-use the concept of “trigger” rather than just “thing I don’t like” or “thing that makes me angry” or “thing that makes me sad” is that, literally, in the political/fandom part of tumblr culture are required to establish your right not to read a thing, and you only have rights if you can establish that you’re on the bad end of an axis of oppression. Hence, co-opting the language of mental illness: trigger.

i.e. trigger warning culture is a rational response to an environment in which media consumption is mandatory. It’s not hypersensitivity so much as the only way to function.

There is a secondary thing, which is, here we are all oppressed, which ties into the feeling that you only have rights if you can establish that you’re at the bad end of an axis of oppression, but I’m not sure I can totally articulate that thing.

The idea that oppression confers legitimacy does seem to be ascendant, and not just on tumblr. Hostile political debates these days often turn into arguments about which side is the injured party, with both claiming to be unfairly caricatured or oppressed. This is pretty bad if it displaces a substantive exchange of ideas, though it may be hard to fix in a society that’s correcting for bias against oppressed groups. The cure isn’t necessarily worse than the disease, though that’s a question worth looking into, as is the question of whether people can learn to see through false claims of grievance.

On the other hand, I don’t think ‘I will (mostly) disregard your non-triggering aversions’ implies ‘you only have rights to the extent you’re oppressed’. I think the deeper problem is that social interaction between strangers and acquaintances is increasingly taking place in massive common spaces, on public websites.

If we’re trapped in the same common space (e.g., because we have a lot of overlapping interests or friends), an increase in your right to freely say what you want to say inevitably means a decrease in my right to avoid hearing things I don’t want to hear. Increasing my right to only hear what I want to will likewise decrease your right to speak freely; at the very least, you’ll need to add content warnings to the things you write, which puts an increasing workload on writers’ plates as the list of reader aversions they need to keep track of grows longer. (Blogging and social media platforms also make things much more difficult, by forcing trigger warnings and content to compete for space at the start of posts.)

I don’t know of any easy, principled way to solve this problem. Readers can download software that blocks or highlights posts/websites using specific words, such as Tumblr Savior and FB Purity. Writers can adopt content warnings for the most common and most harmful trigger and aversions out there, or the ones that are too vague to be caught by word/phrase blockers.

But vague rules are hard to follow. So it’s understandable that people would gravitate toward a black-and-white ‘trigger’ v. ‘non-trigger’ dichotomy in the hope that the scientific authority and naturalness of a medical category would simplify the problem of deciding when the reader’s right-to-not-hear outweighs the writer’s right-to-speak-freely. And it’s equally understandable that people who don’t have ‘triggers’ in the strictest sense, but are still being harmed in a big way by certain things people say (or ways people say them), will want to piggyback off that heuristic once it exists.

‘Only include content warnings for triggers’ doesn’t work, because ‘trigger’ isn’t a natural kind and people mean different things by it. Give some groups an incentive to broaden the term and others an incentive to narrow it, and language will diverge even more. ‘I’ll only factor medical information into my decisions about how to be nice to people’ is rarely the right approach.

‘Always include content warnings for triggers’ doesn’t work either. There are simply too many things people are triggered by.

If we want rules that are easy to follow in extreme cases while remaining context-sensitive in mild cases, we’ll probably need some combination of

‘Here are the canonical content warnings that everyone should use in public spaces: [A], [B], [C]…’

and

‘If you have specific reason to think other information will harm part of your audience, the nice thing to do is to have a private conversation with some of those audience members and consider adding more content warnings. If it’s causing a lot of harm to a lot of your audience, adding content warnings transitions from “morally praiseworthy” to “morally obligatory”.’

The ambiguity and context-sensitivity of the second rule is made up for by the very clear and easy-to-follow first rule. Of course, I only provided a schema. The whole point of the first rule is to actually give concrete advice (especially for cases where you don’t know much about your audience). That project requires, if you’re going to do it right, that we collect base rate information on different aversions and triggers, find a not-terrible way of ranking them by ‘suffering caused’, and find a consensus threshold for ‘how much suffering it’s OK for a random content generator to cause in public spaces’.

That wouldn’t obviate the need for safe spaces where the content is more carefully controlled, but it would hopefully make movies, books, social media, etc. safe and enjoyable for nearly everyone, without requiring people to just stop talking about painful topics.

Chaos Altruism

Impulse buying is a thing. We have ready-made clichés for picking it out. Analogously, ‘impulse giving’ is a thing, where you’re spontaneously moved by compassion to help someone out without any advance planning. The problem with most impulse giving is that it gives you the same warm glow and sense of moral license as high-impact giving, without making as much of a difference. Peter Singer puts it best:

 

 

My experience with the effective altruism community is that they don’t do much to encourage impulse giving of any kind. If you can give to low-impact charities in the heat of the moment, you should be able to do the same for high-impact charities; yet I think of ‘giving effectively’ as affectively cold, carefully budgeted.

This is probably mostly a good thing. We want people to think carefully about their big decisions, if it improves decision quality. However, the stereotype has its disadvantages. If people think they need to go through a long process of deliberation before they can give, they can end up procrastinating indefinitely. Borrowing Haidt’s analogy, we’re discouraging the elephant (our system-1 emotions and intuitions) from getting passionate and worked up about the most important things we do, while encouraging the elephant’s rider (our system-2 reasoning and deliberation) to overanalyze and agonize over decisions.

Effective altruism as it exists today is aligned with the legions and principalities of Order. I’d bet we can change that in some respects, if we so wish, without giving up our allegiance to Goodness.

 

Affective altruism

Eliezer Yudkowsky suggests that we “purchase fuzzies and utilons separately“. Better to spend some of your time on feel-good do-gooding and some on optimal high-impact do-gooding, rather than pursuing them simultaneously and doing a terrible job at both. In “Harry Potter and the Fuzzies of Altruism“, I noted that there are different kinds of fuzzies people can get for doing good.

One of these varieties is particularly valuable, because it doesn’t need to be purchased separately. I speak of the slytherfuzzy, that warm glow you get from being especially efficient and effective. Do-gooders who find cool, calculated pragmatism strongly motivating in its own right have an obvious leg up. I myself am more motivated by narrative, novelty, and love-of-neighbor than by Winning, but I’d love to find a way to steal that trick and bind my own reward center more tightly to humanitarian accomplishment.

If you’re trying to make yourself (or others) more enthusiastic about purchasing utilons, it may be helpful to make the way you buy utilons as fuzzy-producing as possible. This needn’t dilute the outcome. Select a charity based on a sober cost-benefit analysis, but give chaotically, if chaos happens to gel with your psychology. Impulse giving and effective altruism don’t have to be placed in separate mental boxes forever; we can invent new categories of behavior that wed Chaos Altruism’s giddy spontaneity to Order Altruism’s focus and rigor.

 

 

I’d expect mixed approaches to work best. E.g., you can settle on a fixed percentage of your income to give to a high-impact cause every year, but build a habit of giving bonus donations to that cause when the mood strikes you. I’m a big fan of using specific benevolence triggers. For example: ‘When someone on the street asks me for money, and I feel an urge to give them $X, give $X to a high-impact charity (whether or not I also give money to the individual who asked).’ Leah Libresco and Michael Blume make good use of this kind of ‘nudged giving’.

But I think we should also normalize whimsical, untriggered high-impact giving. If we start thinking of evidence-based humanitarianism as the kind of thing you can splurge on, I suspect we’ll come to see do-gooding as more of a fun opportunity and less of a burden.

Some people think of their philanthropy as a personal passion that drives them to excel, as in Holden Karnofsky’s “excited altruism“. Others think of their philanthropy as a universal moral obligation they’re striving to meet, as in Eliezer’s “one life against the world“. Try to fit all philanthropists into the ‘passion’ box, and you’ll get a contingent that feels cut off from what makes this work important; try to fit them all into the ‘obligation’ box, and you’ll get a contingent that feels burdened with a dour or guilt-inducing chore.

Likewise, there are important points of divergence between do-gooders who are motivated by different kinds of warm fuzzy (or hot blazing, or cool gliding, or wiggly sparkling…) feelings. I’m more of an obligation-based altruist, but I still find the ‘excited altruist’ framing useful. That I think in moralistic terms doesn’t say much about the specific feelings that drive me to do good in the moment.

 

Infectious altruism

My moralism also leaves open what feelings I should emphasize if I want to transfer my enthusiasm to others.

The ice bucket challenge is an example of memetically successful Chaos Altruism. Ditto today’s Giving Tuesday event, though an annual event is relatively compatible with the reign of Order. Will McAskill and Timothy Ogden have criticized these memes as possibly counterproductive, but it’s not obvious to me that the ineffectiveness of these events stems from their viral or ad-hoc character. Instead, those same attributes could be very valuable if they were targeted at more urgent causes.

McAskill and Ogden draw attention to the fact that charitable donations have been stuck at 2% of U.S. GDP for 40 years now. People (on average) seem to change where they donate, but not how much they donate. One approach to doing better, than, would be to redirect that 2% to worthier interventions.

At the same time, the success of the giving pledge shows that some people can be inspired to increase their donations. Perhaps we haven’t been able to rise above 2% because charities are too busy competing with each other to focus their advertising ingenuity on growing the pie. Perhaps some deep change in people’s mindset is needed; I’ll note that households giving to religious nonprofits donate twice as much. Relatedly, the key may be to shift entire (small) communities to giving more, so giving more is the norm among everyone you know. Then expand those supergiver tribes into neighboring social networks.

Experimenting with playful, unorthodox, and personalized modes of altruism seems like it could be useful for finding ways to make inroads in new communities. Over the next few years, I think we should place more focus on self-experimentation and object-level research than on outreach; but we should still keep in mind that we need a better handle on human motivation if we’re going to completely restructure the way charity is done. For that reason, I’m eager to hear whether any aspiring effective altruists find Chaos approaches attractive.