Reposted from a few days ago, noting that jsalvatier (kudos to him for putting up the prize money, very community spirited)   has promised $100 to the winner, and I have decided to set a deadline of Wednesday 1st December for submissions, as my friend has called me and asked me where the article I promised him is. This guy wants his god-damn rationality already, people! 

My friend is currently in a potentially lucrative management consultancy career, but is considering getting a job in eco-tourism because he "wants to make the world a better place" and we got into a debate about Efficient Charity, Roles vs. Goals, and Optimizing versus Acquiring Warm Fuzzies

I thought that there would be a good article here that I could send him to, but there isn't. So I've decided to ask people to write such an article. What I am looking for is an article that is less than 1800 words long, and explains the following ideas: 

  1. Charity should be about actually trying to do as much expected good as possible for a given amount of resource (time, $), in a quantified sense. I.e. "5000 lives saved in expectation", not "we made a big difference". 
  2. The norms and framing of our society regarding charity currently get it wrong, i.e. people send lots of $ to charities that do a lot less good than other charities. The "inefficiency" here is very large, i.e. Givewell estimates by a factor of 1000 at least.  Our norm of ranking charities by % spent on overheads is very very silly. 
  3. It is usually better to work a highly-paid job and donate because if you work for a charity you replace the person who would have been hired had you not applied
  4. Our instincts will tend to tempt us to optimize for signalling, this is to be resisted unless (or to the extent that) it is what you actually want to do. Our instincts will also tend to want to optimize for "Warm Fuzzies". These should be purchased separately from actual good outcomes
  5. Our human intuition about how to allocate resources is extremely bad. Moreover, since charity is typically for the so-called benefit of someone else, you, the donor, usually don't get to see the result. Lacking this feedback from experience, one tends to make all kinds of gigantic mistakes. 

but without using any unexplained LW Jargon. (Utilons, Warm Fuzzies, optimizing). Linking to posts explaining jargon is NOT OK. Just don't use any LW Jargon at all. I will judge the winner based upon these criteria and the score that the article gets on LW. Maybe the winning article will not rigidly meet all criteria: there is some flexibility. The point of the article is to persuade people who are, at least to some extent charitable and who are smart (university educated at a top university or equivalent) to seriously consider investing more time in rationality when they want to do charitable things. 

New Comment
57 comments, sorted by Click to highlight new comments since: Today at 8:17 AM

Using a throwaway account to keep judging unbiased

Imagine you are setting out on a dangerous expedition through the Arctic on a limited budget. The grizzled old prospector at the general store shakes his head sadly: you can't afford everything you need; you'll just have to purchase the bare essentials and hope you get lucky. But what is essential? Should you buy the warmest parka, if it means you can't afford a sleeping bag? Should you bring an extra week's food, just in case, even if it means going without a rifle? Or can you buy the rifle, leave the food, and hunt for your dinner?

And how about the field guide to Arctic flowers? You like flowers, and you'd hate to feel like you're failing to appreciate the harsh yet delicate environment around you. And a digital camera, of course - if you make it back alive, you'll have to put the Arctic expedition pics up on Facebook. And a hand-crafted scarf with authentic Inuit tribal patterns woven from organic fibres! Wicked!

...but of course buying any of those items would be insane. The problem is what economists call opportunity costs: buying one thing costs money that could be used to buy others. A hand-crafted designer scarf might have some value in the Arctic, but it would cost so much it would prevent you from buying much more important things. And when your life is on the line, things like impressing your friends and buying organic pale in comparison. You have one goal - staying alive - and your only problem is how to distribute your resources to keep your chances as high as possible. These sorts of economics concepts are natural enough when faced with a journey through the freezing tundra.

But they are decidedly not natural when facing a decision about charitable giving. 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.

If you are to "love your neighbor as yourself", then you should be as careful in maximizing the benefit to others when donating to charity as you would be in maximizing the benefit to yourself when choosing purchases for a polar trek. And if you wouldn't buy a pretty picture to hang on your sled in preference to a parka, you should consider not helping save a famous painting in preference to helping save a thousand lives.

Not all charitable choices are as simple as that one, but many charitable choices do have right answers. GiveWell.org, a site which collects and interprets data on the effectiveness of charities, predicts that antimalarial drugs save one child from malaria per $5,000 worth of bed medicine, but insecticide-treated bed nets save one child from malaria per $500 worth of drugs. If you want to save children, donating bed nets instead of antimalarial drugs is the objectively right answer, the same way buying a $500 TV instead of an identical TV that costs $5,000 is the right answer. And since saving a child from diarrheal disease costs $5,000, donating to an organization fighting malaria instead of an organization fighting diarrhea is the right answer, unless you are donating based on some criteria other than whether you're helping children or not.

Say all of the best Arctic explorers agree that the three most important things for surviving in the Arctic are good boots, a good coat, and good food. Perhaps they have run highly unethical studies in which they release thousands of people into the Arctic with different combination of gear, and consistently find that only the ones with good boots, coats, and food survive. Then there is only one best answer to the question "What gear do I buy if I want to survive" - good boots, good food, and a good coat. Your preferences are irrelevant; you may choose to go with alternate gear, but only if you don't mind dying.

And likewise, there is only one best charity: the one that helps the most people the greatest amount per dollar. This is vague, and it is up to you to decide whether a charity that raises forty children's marks by one letter grade for $100 helps people more or less than one that prevents one fatal case of tuberculosis per $100 or one that saves twenty acres of rainforest per $100. But you cannot abdicate the decision, or you risk ending up like the 11,000 people who accidentally decided that a pretty picture was worth more than a thousand people's lives.

Deciding which charity is the best is hard. It may be straightforward to say that one form of antimalarial therapy is more effective than another. But how do both compare to financing medical research that might or might not develop a "magic bullet" cure for malaria? Or financing development of a new kind of supercomputer that might speed up all medical research? There is no easy answer, but the question has to be asked.

comments have a size limit? It seems they do. Continued in reply.

continued from above

What about just comparing charities on overhead costs, the one easy-to-find statistic that's universally applicable across all organizations? This solution is simple, elegant, and wrong. High overhead costs are only one possible failure mode for a charity. Consider again the Arctic explorer, trying to decide between a $200 parka and a $200 digital camera. Perhaps a parka only cost $100 to make and the manufacturer takes $100 profit, but the camera cost $200 to make and the manufacturer is selling it at cost. This speaks in favor of the moral qualities of the camera manufacturer, but given the choice the explorer should still buy the parka. The camera does something useless very efficiently, the parka does something vital inefficiently. A parka sold at cost would be best, but in its absence the explorer shouldn't hesitate to choose the the parka over the camera. The same applies to charity. An antimalarial net charity that saves one life per $500 with 50% overhead is better than an antidiarrheal drug charity that saves one life per $5000 with 0% overhead: $10,000 donated to the high-overhead charity will save ten lives; $10,000 to the lower-overhead will only save two. Here the right answer is to donate to the antimalarial charity while encouraging it to find ways to lower its overhead. In any case, looking for low overhead is helpful but not enough to answer the "which is the best charity?" question.

Just as there is only one best charity, there is only one best way to donate to that charity. Whether you volunteer versus donate money versus raise awareness is your own choice, but that choice has consequences. If a high-powered lawyer who makes $1,000 an hour chooses to take an hour off to help clean up litter on the beach, he's wasted the opportunity to work overtime that day, make $1,000, donate to a charity that will hire a hundred poor people for $10/hour to clean up litter, and end up with a hundred times more litter removed. If he went to the beach because he wanted the sunlight and the fresh air and the warm feeling of personally contributing to something, that's fine. If he actually wanted to help people by beautifying the beach, he's chosen an objectively wrong way to go about it. And if he wanted to help people, period, he's chosen a very wrong way to go about it, since that $1,000 could save two people from malaria. Unless the litter he removed is really worth more than two people's lives to him, he's erring even according to his own value system.

...and the same is true if his philanthropy leads him to work full-time at a nonprofit instead of going to law school to become a lawyer who makes $1,000 / hour in the first place. Unless it's one HELL of a nonprofit.

The Roman historian Sallust said of Cato "He preferred to be good, rather than to seem so". The lawyer who quits a sleazy law firm to work at a nonprofit organization certainly seems like a good person. But if we define "good" as helping people, then the lawyer who stays at his law firm but donates the profit to charity is taking Cato's path of maximizing how much good he does, rather than how good he looks.

And this dichotomy between being and seeming good applies not only to looking good to others, but to ourselves. When we donate to charity, one incentive is the warm glow of a job well done. A lawyer who spends his day picking up litter will feel a sense of personal connection to his sacrifice and relive the memory of how nice he is every time he and his friends return to that beach. A lawyer who works overtime and donates the money online to starving orphans in Romania may never get that same warm glow. But concern with a warm glow is, at root, concern about seeming good rather than being good - albeit seeming good to yourself rather than to others. There's nothing wrong with donating to charity as a form of entertainment if it's what you want - giving money to the Art Fund may well be a quicker way to give yourself a warm feeling than seeing a romantic comedy at the cinema - but charity given by people who genuinely want to be good and not just to feel that way requires more forethought.

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. Consider going to www.GiveWell.org and making use of the excellent resources on effective charity they have available.

Excellent. I was beginning to despair that we wouldn't have a submission at all.

Can you add some links to further reading on LW to it? E.g. all the articles cited in the OP?

Over the past few days I've been low on energy and can't seem to reach the level of concentration needed to write a good article of the desired length in the near term....I may make a solid attempt sometime over the next few months but not by Wednesday.

I'm pretty happy with the submission from throwaway_account_1. I suspect that it's possible to better optimize for the intended audience but can't think of an easy way to do this.

next few MONTHS?!?!!??!?!

Can somebody please pay multifoliaterose another $x if he actually writes something?!

Okay, okay, I'll give it a go, no money required; I just need to reach a sufficiently high level of energy. I overextended myself between August and November and am in the process of recovering; that's why I deferred. But I can probably write something sooner, will try to get something done over the next few days. :-)

Can somebody who lives close to multifoliaterose (east coast I believe?) Please go to his house and give him some energy?

I believe a well booted foot travelling at 3 m/s would deliver about 10-20 Joules of energy. It might be applied to his behind for maximum impact. ;-)

The lawyer who quits a sleazy law firm to work at a nonprofit organization certainly seems like a good person. But if we define "good" as helping people, then the lawyer who stays at his law firm but donates the profit to charity is taking Cato's path of maximizing how much good he does, rather than how good he looks.

If the law firm is sleazy, then he might be actively doing harm to people while working there, and this could justify a decision to quit, or to avoid law school in the first place.

For example, he might be an ambulance-chaser whose cases clog up the courts and drive up insurance premiums, to the point that some poor people decide to drive without insurance, go to jail, and can't feed their families.

It's hard to shut up and multiply when we haven't looked at the numbers.

I see this as a flaw in the argument that might well be repairable. Otherwise, it's a great article -- upvoted twice.

If you consider the lives of other people to be as valuable as your own life

This seems like massively too strong an assumption to make. Practically no-one actually values the lives of random strangers equally to their own. And you don't need to assume it anyway.

comments have a size limit? It seems they do. Continued in reply.

On the positive side that allows us to up-vote this twice.

Great idea Roko and jsalvatier!

Here's my entry.

are there any other good ones?

Giving What We Can has a similar approach (quality-adjusted-years-of-life are what count, not % spent on admin), and concurr with GiveWell on one of the top 4 charities IIRC (Stop TB Partnership).

Our norm of ranking charities by % spent on overheads is very very silly.

That sounds rather overstated. Are there any good charities that spend a lot on overhead? The vast majority of charities, especially as weighted by money, are best thought of as frauds. This is a very crude filter, since most well-intentioned charities still fail to accomplish anything, but hardly silly. The most serious criticism of this filter is that one is not going to give to many charities, so filtering 90% is of limited use.

I think it is no understatement to say that the norm is very, very, silly, though now we are in the territory of arguing about the mapping from real-world consequences to adjectives, i.e. we are arguing about connotations.

The only reason to chose the word "silly" is for the connotations.

Givewell started out asserting that this rule has lots of false negatives, with the real-world predictable consequence of ill-will. The denotation of their statement is far less important than the connotation, but they were wrong there, too.

Why is the denotation wrong? It does produce lots of false negatives.

and false positives and creates bad incentives.

Global poverty is too large a problem for any one person to solve, but each of us can still transform the lives of thousands of people. While it is difficult to help directly, we must not forget our most important advantage: on a world scale, we are very rich. We can thus pay for efficient services in health and education which, though desperately desired, are out of the reach of those in poverty. If the typical US citizen gave 10% of their income to the right NGOs, then each year they could:

* Distribute 700 mosquito nets, preventing 1,900 cases of malaria and 6 deaths
* Cure 170 people of tuberculosis, preventing 8 deaths
* Save 1,100 years worth of healthy life
* Provide 1,100 additional years of school attendance

The median personal income in the US is $35,500 (US Census 2008). Ten percent of this is $3,550.

Mosquito nets can be distributed for $5 each, cases of malaria prevented for $1.80, deaths from malaria prevented for $600 (see note 49 in this GiveWell summary).

Tuberculosis can be cured for $20, and deaths from TB prevented for $150-$750 (see the GiveWell page on the Stop-TB Partnership).

Disability Adjusted Life Years can be averted for as little as $3 each (see our page on neglected tropical diseases).

Treating children for neglected tropical diseases produces an extra year of school attendance for each $3 (see the J-PAL study , but note that this doesn't include the possible need for extra teachers if more class members turn up).

Read through that list again and consider that each of us could each achieve one of these great benefits every single year. Read it through and try to imagine the scale of those numbers: to see the individual names and faces in your mind. Pick out one of these individuals and try to imagine the huge effect this will have on his or her life. It is just staggering. In a single week we can perform something like a miracle: saving a life, or restoring sight to the blind. Over our lives, we can each perform thousands of these ‘miracles’, leaving behind a remarkable legacy. Moreover, we can do all of this without leaving our countries, without leaving our preferred jobs, and without even giving up any parts of our lives that are truly important to us.

We clearly have a duty to do at least this much. We can do something of extreme moral importance without sacrificing anything of comparative value. How could we look these people in the eye and justify our failure to give even such a small amount? Isn't this the least we could do?

Many people flee from these facts and try hard to forget them, but we needn't do so. Instead, we can embrace the facts and simply decide to give generously. This is what the members of Giving What We Can have done. We've each made a public pledge to give at least 10% of our incomes to where we believe it will do the most to fight poverty. Whatever our incomes, we will all have a tremendous effect on thousands of lives. We don't seek any praise for this as it seems to us to be the least we could do. What we do want is for others to join us in this endeavour: to share advice on the most effective ways to help, and to give what we can.

~GivingWhatWeCan.org

(if chosen for the prize, I will donate half to GivingWhatWeCan.org and half to Deworming the World.)

EDIT: Per the request of someone who appears to be heavily involved with GWWC, if chosen for the prize, I will donate the entire prize to Deworming the World.

It's worth mentioning that contra our intuitions, there are strong reasons for donating to a single charity rather than multiple charities. Unless you donate a lot of money you're not going to affect the marginal good of donating to that charity. This means that whatever you think has the highest expected marginal good then it will remain so and you should donate only to that charity until your expectation changes. You also should not be risk averse when it comes to charity; since the planets population is quite large, a 50% chance of saving 2 lives is just as good as a 100% chance of saving 1 life.

Er, I meant my comment about what you plant to do with the prize, not as a comment on your essay.

Unless you donate a lot of money you're not going to affect the marginal good of donating to that charity. This means that whatever you think has the highest expected marginal good then it will remain so.

Er, yeah, of course. That's totally right.

You also should not be risk averse when it comes to charity; since the planets population is quite large.

I'm not sure that conclusion follows as strongly as you think it does. Given a choice between an exactly 10% chance of saving exactly 11 lives and a perfect guarantee of saving 1 life, I guess I'd go with .10 * 11 = 1.1 expected lives. But given a variety of charities with uncertain effectiveness and uncertain error bars, it seems sensible to diversify my portfolio. I could be objectively wrong about whether Deworming the World is better than the best water-treatment charity, but it's unlikely that any one charity or combination of charities is significantly and knowably better than a small basket of some of the best charities in different categories. In short, I think there's a difference between diversifiying and hedging -- you don't need to hedge with individual charities, but you should still diversify.

Er, I meant my comment about what you plant to do with the prize, not as a comment on your essay.

Well, it's not my essay -- it's Giving What We Can's. I just plagiarized it, on the theory that they would see it as good publicity and wouldn't mind. I'm giving them half by way of guilt money, and in case they can think of something better to do with the money. If not, they can just pass their $50 on to Deworming the World.

It's important to understand why you normally diversify. The reason why diversification is a good idea is because you have diminishing marginal utility of wealth. This reason just doesn't apply to charity. Your contributions make such a small difference that even if "total utility" had diminishing returns to live saved (debatable), over your range of influence it has effectively constant returns.

I agree that it's emotionally attractive to diversify, but it's simply incorrect to do so (though if your choices are of similar quality, it's not the worst thing in the world).

I'm beginning to think that LW needs a series on finance. A lot of what seems intuitive to me, doesn't seem intuitive to others, and I think standard finance has several valuable insights like this.

So, I know it's wise to purchase warm fuzzies and utilons separately, but it just so happens that I get a significant quantity of warm fuzzies from saving hundreds of lives. I'm weird like that.

Anyway, suppose (against all evidence) that utilities are ordinally intercomparable. Suppose further that the relevant chunk of my utility function is U(charity) = U(fuzzies) + U(altruism), where U(fuzzies) = ln(# of lives saved), and U(altruism) = (net utility of saved life to owner) * (my discount rate for the utility of strangers). Let's say the typical life saved by charities is worth 30,000 utilons to its owner, and that my discount rate for strangers' utility is 1/100,000.

So, if I save 200 lives, I get ln(200) + (30,000 200 / 100,000) = 65 utilons for me. If I save 2,000 lives, I get ln(2000) + (30,000 2,000 / 100,000) = 607 utilons for me. My original point was going to be that I do get diminishing marginal returns to charity, but apparently given my assumptions they diminish so slowly as to be practically constant, and so I will shut up and pick just one charity in so far as I can find the willpower to do so.

Hooray for accidentally proving yourself wrong with back of the envelope calculations.

I'd love a discussion of finance accessible to a smart but finance-phobic lay audience (raises hand).

For example, I too have been under the impression that I diversify my investments, not because I'm concerned that my investible income will drive some specific company over an inflection point in its utility function, but because it makes me less vulnerable to a single point of failure. ( Indeed, I'm still under that impression; the alternative seems absurd on the face of it. )

I think the point is that investments pay a return to you, so a single point of failure really hurts you where it counts.

Charities, on the other hand, pay their return to the world. The world is not horribly damaged if a single charity fails; the world is served by many charities. In effect, the world is already diversified.

If the charity you gave all your donations to fails, you may feel bad, but you will get over it. Not necessarily the case if the investment you sunk all your own money into fails.

Sure, I get that. The thing I was responding to was jsalvatier's comment that "It's important to understand why you normally diversify. The reason why diversification is a good idea is because you have diminishing marginal utility of wealth." S/he wasn't talking about charity there, I don't think... though maybe I was confused about that.

"Diminishing marginal utility of wealth" means the same thing as "don't want to be exposed to a single point of failure".

Yes, I think we do need a series on econ/finance.

(blink)

Two investment strategies, S1 & S2. They have the same average expected ROI, but S1 involves investing all my money in a single highly speculative company with a wider expected variance... my investment might go up or down by an order of magnitude. So, S1 suffers from a single point of failure relative to S2.

You're saying that I could just as readily express this by saying "S1 involves diminishing marginal utility of wealth relative to S2"... yes?

Huh. I conclude that I haven't been understanding this conversation from the git-go. In my defense, I did describe myself as finance-phobic to begin with.

No - what's under question is the scaling behavior of your own utility function wrt money; if you exhibit diminishing marginal utility of wealth, that means you want to avoid S1.

Isn't anything worth doing worth troubleshooting for a single point of failure?

Only if the relevant utility function is sublinear. If the relevant utility function is linear or superlinear, you aren't risk averse, so you don't care about a single point of failure. You care about p(failure)x0 + p(success)xU(success).

E.g. assuming altruism is all you care about, if you could pay $10,000 for a bet which would cure all the problems in Africa with probability 0.01%, and do nothing with probability 99.99%, then you should take that bet.

So, the project "global charity" doesn't have a single point of failure even if all individuals choose exactly one charity each.

But I'm not sure that I'm permitted to take the global point of view -- after all, I only control my own actions. From my personal vantage point, I care about charity and I care about preserving my own solvency. To secure each of these values, I should avoid allowing my plan for achieving either value to suffer from a single point of failure, no?

Right, you should make sure your plan for personal solvency doesn't have a single point of failure. As for global charity, do you really have a plan for that? My model had been that you are simply contributing to the support of some (possibly singleton) collection of plans. With the objective of maximizing expected good. If the true goal is something different - something like minimizing the chance that you have done no good at all and hence miss out on the warm and fuzzies - then, by all means, spread your charity dollar around.

Has anyone worked out numbers (according to whatever axioms) for what the best donor behaviour is for charities to encourage? Will they do better if they encourage people to give all their money to one charity, or to split it amongst several?

I'm not sure how you'd design a model for "encouragement" and its effects. Care to elaborate on what sort of model you're thinking about?

Simply what would work out best for the charities rather than the donors. Would it be in the best interests of a given charity to have donors interested in giving all their charity money to one charity, or to have donors split it between a bunch of charities?

Can't say I've seen specific models. I imagine it depends on what the utility functions of charities are. If the charities are run by people just looking to keep their jobs/get status, I'd guess they want people to donate to them rather than other charities on the margin, and that would lead to people donating to lots of charities. If charities just want to help make the world better then that implies that they want people to donate to the single best charitiy. Since there are many charities in the world, that supports theory 1.

I don't think it depends on the motivations of those in the charities, based on my views of the insides of a few - some of which had succumbed thoroughly to the Iron Law of Institutions (where the people working there didn't believe any more), some of which were pretty solidly oriented to their stated purpose, and some of which were in between (some recoverable, some not).

I think we can reasonably just assume, without making this question meaninglessly hypothetical, that the charities want money and we don't need to go more deeply into it for the question to be applicable to the real world - and I am indeed asking because I am interested in the real world applications of this question:

What is the best giving strategy for charities to encourage: for people to spread their donations, or for people to give their year's charity spend to a single charity?

(For a start, I would guess this would vary with size and fame of charity. Large charities would be more confident of being the winner, small charities would be more pleased to get anything at all. Then there is the fact that not only are the donors not independent actors, the donors and charities aren't either. Can a donor and a charity be said to be conspiring to achieve their aim? I think they can. This complicates things, though I'm not sure if it's enough to make a difference.)

[And I would be flatly amazed if there wasn't considerable study on this subject already, as there have been enough large charities for long enough who would be considerably interested in the topic that anything claiming to be a radical breakthrough in thinking on the subject from outside the charity field will need to be assessed in the context of existing work, rather than being regarded as a completely new idea in an unexplored field. As I recall, there was no mention of any past work in this area at all. Is there actually none, or did the authors just not look?]

uncertain error bars, it seems sensible to diversify my portfolio.

no. Bayesianism. Integrate over the uncertainty distribution for the error bars.

Sorry, I'd like to, but I'm running on flawed hardware. The uncertainty distribution is itself plagued by error bars, and so on all the way down. As Dirty Harry would put it, "A man's gotta know his limitations." Or, if you prefer Sherlock Holmes, "Data, data, data! I cannot make bricks without clay."

You can still integrate. I doubt that the meta-errors are really important differentiators between villagereach and Oxfam.

It sounds like meta -errors are not your true rejection

I'd guess that your true rejection might be wanting to avoid the emotional pain of failure if you stake all $ on one particularly good-looking charity which then goes on to be exposed as a fraud.

Or possibly your true rejection is the emotional hit you'd take from worrying about whether you got it wrong.

There are many non-rational reasons people have for placing a certainty premium on charity.

I agree. I share Mass_Driver's emotional desire to diversify even though I know it's wrong.

I'd probably actually diversify since it seems like a positive sum game between the egoist in me and the altruist in me. The egoist more wants actual, real status/reward, which tends to only be gained when you pick an actual winner. People don't give you any praise for anything other than actual, real successes, I find. And if the expected marginal utilities of the top 5 causes are comparable (the same to within a factor of 2), the altruist isn't actually conceding very much.

It sounds like meta -errors are not your true rejection. I'd guess that your true rejection might be wanting to avoid the emotional pain of failure if you stake all $ on one particularly good-looking charity which then goes on to be exposed as a fraud.

That's a pretty good guess. Probably correct. I wonder, though, how many people manage to care about charity so directly as to value saving lives literally for the sake of saving lives, rather than for the emotional satisfaction associated with it. I think the odds of me suffering from plague, reincarnation, violent uprising, etc. that is partly caused by me donating to a slightly suboptimal basket of charities are basically negligible. What, then, is the moral or philosophical theory that says that I should privilege the act of donating my whole charity budget to one maximally efficient charity over the emotional satisfaction of donating to a basket of moderately efficient charities? I enjoy the latter more; I know because I have tried each method a few times in different years. Why should I personally do something that I enjoy less? I don't mean to be triumphant about this; possibly there is a very good reason why I should do something that I enjoy less. I just don't know what it is. And don't say something blunt like "it'll save more lives." I know it will save more lives on average, and I've noticed that I don't care. Should I work to change this about myself, and if so, why?

I see it as a "deal" between an egoist subagent and an altruist subagent.

The crucially important factor in this deal is just what the effectiveness ratio is between charity #1 and charities#2, #3, #4, #5, #6. If the marginal good done per $ is similar between all of them, then OK go ahead and diversify.

All right, well, let's consider the least convenient example. Suppose the estimated marginal good between charity #1 and #6 is off by a factor of 8 -- enough to horrify the altruist, but barely enough for the egoist (who primarily likes to think that he's being useful on lots of the most important problems) to even notice.

What can I tell the moderator subagent that would make him want to side in favor of the altruist subagent?

Well instead of spreading the money between all 6 charities, why not reduce your donation by 50% but donate all of it to #1, and then give the remaining 50% to the egoist subagent to buy something nice with?

It's good thinking, but this particular egoist primarily likes to think that he's being useful on lots of apparently important problems. He can't be bribed with ordinary status symbols like fancy watches. Is there a way to spend money to trick yourself into thinking you're useful? None immediately springs to mind, but I guess there might be one or two.

you could spend 90% of the money on cause1 and split the remaining 10% between the rest

Which actually isn't all that irrational if we think of it as a decision theory problem with diminishing returns on money - making sure that at least some of your money is used well becomes more important than gambling that all of it is used well.

Of course, given the nature of what's being dome with the money, the returns diminish much, much more slowly than we're used to; diversity shouldn't be a concern until you're Onassis-ish rich.

As clearly stated above, for small donors marginal returns don't diminish.

But in the end, there is an "all the way down," that being the information you started from. So the problem is solvable by good ol' statistics, usually with there being a single best option.

Giving What We Can does not accept donations. Just give it all to Deworm the World.