After rushing home from Utah I made it to a talk on efficient charity which St. Paul of Rational Altruist put together at the Berkeley Faculty Club.
(I should probably mention that from now on I will be preceding the names of leaders in the effective charity movement with the honorific “St”, because people’s social status and recognition ought to accurately track the level of effect they are having on the world. If through persistence and self-sacrifice someone manages to save the lives of a couple dozen people they never met, I don’t think even the harshest advocatus diaboli could object to an impromptu sainting or two.)
First, St. Elie of GiveWell talked about his organization’s efforts trying to research which charities did the most good and defended the idea of radical transparency – GiveWell’s procedure of meticulously recording everything they do and criticizing themselves ad infinitum in an attempt to become better and more evidence-based.
Then Robin Hanson of Overcoming Bias got up and just started Robin Hansonning at everybody. First he gave a long list of things that people could do to improve the effectiveness of their charitable donations. Then he declared that since almost no one does any of these, people don’t really care about charity, they’re just trying to look good. Then he told the room – this beautiful room in the Faculty Club, full of sophisticated-looking charity donors who probably thought they were there to get a nice pat on the back – that they probably thought that just because they were attending an efficient charity talk they weren’t like that, but that probabilistically there was excellent evidence that they were.
I have never seen a group of distinguished Berkeley faculty gain so sudden and intuitive an appreciation for the Athenians who decided to put Socrates to death. I spent the whole speech grinning like an idiot and probably scared Robin a little. And okay, some of that was because I woke up really early to get to the airport today and had become dangerously overtired and mentally imbalanced, but the rest of it was just that he sounds exactly like he does on his blog, he’s a great speaker, and it was just really funny in a train-wreck sort of way to watch a whole room of innocent and basically decent people get Hansonned. The man is one of a kind and his complete and obviously deliberate imperviousness to normal social niceties needs to be declared a national treasure.
But he made some genuinely unsettling points.
One of his claims that generated the most controversy was that instead of donating money to charity, you should invest the money at compound interest, then donate it to charity later after your investment has paid off – preferably just before you die, since donating money after death is legally complicated. His argument, nice and simple, was that the real rate of return on investment has been higher than the growth rate for 3000 years and this pattern shows no signs of changing. If you donate the money today, your donation grows with the growth rate, but if you invest it, it grows with the interest rate. He gave his classic example of Benjamin Franklin, who put his relatively meager earnings into a trust fund to be paid out two hundred years later; when they did, the money had grown to $7 million. He said that the reason people didn’t do this was that they wanted the social benefits of having given money away, which are unavailable if you wait until just before you die to do so.
And darn it, he was totally right. Not about the math – there are severe complications which I’ll bring up later – but about the psychology. On even the most cursory self-examination, my mind totally recoils at the thought of donating everything I’m going to donate to charity in a single lump sum just before I die. It just gibbers “But…but…you need to be a good person before then!” I’m not saying you can’t tear down Robin’s substantive argument in a bunch of good mathematical ways. I’m saying his ad hominem argument about my motivations seems to be true regardless.
Then he started talking about how you should only ever donate to one charity – the most effective. I’d heard this one before and even written essays speaking in favor of it, but it’s always been very hard for me and I’ve always chickened out. What Robin added was, once again, a psychological argument – that the reason this is so hard is that if charity is showing that you care, you want to show that you care about a lot of different things. Only donating to one charity robs you of opportunities to feel good when the many targets of your largesse come up and burdens you with scope insensitivity (my guess is that most people would feel more positive affect about someone who saved a thousand dogs and one cat than someone who saved two thousand dogs. The first person saved two things, the second person only saved one.) In retrospect this is absolutely true and my gibbering recoil at this problem isn’t just Yet Another Cognitive Bias but just good old self-interest.
Now that I have identified what this pattern feels like, I can look back in my memory and notice more examples. Probably the worst is that an efficient charity group was actually slightly interested in having me interview to work with them back in February-ish and I declined because I already had a career plan – and one from which I could make lots of money to then donate. This may end up being the correct decision and in fact probably is, but I already know that’s not why I did it. I did it because a cognitive paranoia I cultivated for pretty good reasons but can’t turn off at will tells me that doing anything not directly measurable is just my brain lollygagging about and inventing clever stories about why it’s so great, and I only get credit for direct obvious quantitative sacrifice. Working for an efficient charity organization, even if I was able to redirect other people’s money in valuable ways that ended up outweighing the utility of making lots of money myself, wouldn’t be enough to overcome my brain’s suspicion that I wasn’t actually doing any good after all.
Thankfully, Robin’s last point was that the most effective thing to do is to stop beating yourself up and be exactly as irrational as is necessary to convince your mind to go along with the whole “efficient charity thing” instead of freaking out and giving up in disgust. I have already measured about how irrational that is and I don’t see too much reason to change my decision now. Still, no sainthood for me just yet.
So let’s get to the fun part. How do we debunk Robin’s assertion that we should invest charitable givings and donate them only at the end of our lives?
St. Elie discussed this for a little while at the talk and gave what I thought was an unexpectedly good answer. He said that the world is getting better so quickly that we are running out of good to be done. After the initial burst of astonishment he explained: in the 1960s, the most cost-effective charity was childhood vaccinations, but now so many people have donated to this cause that 80% of children are vaccinated and the remainder are unreachable for really good reasons (like they’re in violent tribal areas of Afghanistan or something) and not just because no one wants to pay for them. In the 1960s, iodizing salt might have been the highest-utility intervention, but now most of the low-iodine areas have been identified and corrected. While there is still much to be done, we have run out of interventions quite as easy and cost-effective as those. And one day, God willing, we will end malaria and maybe we will never see a charity as effective as the Against Malaria Fund again.
St. Paul lists several other reasons on his blog. First of all, the US’ tax deduction laws favor spreading your donations out among as many years as possible. Second, you might become a worse person in fifty years and decide you don’t want to give your massive accumulated savings to the poor after all. He also lists a few other arguments, none of which in my opinion have quite the same power as those two.
Most of the people at the meeting today were not radical singularitarians – not all saints can be prophets. But if you believe, as I do, that we’re within about a century of a technological singularity of some sort or other, three new considerations come into play. First, affecting the singularity – either bringing it forward, pushing it backwards, or changing its nature – may be a unique and fantastically high-leverage charity target which will not be available (or may be less available) fifty years from now. Second, if a singularity goes wrong and kills us all, we lose our opportunity to donate to charity later. Third, if a singularity goes right, it’s a pretty good bet that people won’t need malaria nets anymore.
Suppose I will die at age 78 – ie fifty years from now. And suppose I want to donate $10,000 to charity. If I decide against Robin’s strategy, I donate $10,000 today, perhaps to Ethiopia. Ethiopia has a growth rate of about 7%, but let’s assume it can’t keep that up over the next 50 years and its average is 5% (this is still quite high). In 2063 Ethiopia ends up with $115,000 extra or so, which I round off to $150,000 because it can be enjoyed by a couple of generations rather than simply appearing at the end of the period.
Suppose I decide in favor of Robin’s strategy. A couple of investment sites say to expect 7% rate of return, so I can expect about $300,000 (all these numbers are adjusted for inflation, I think).
(if these numbers are right, then using the “efficacy of charity declines as things get better over time” argument means you have to believe in a 50%-in-fifty-years decline in charitable efficacy, which seems like a pretty high bar)
Okay. Now what if the world ends (or progresses beyond the need for charity) in 2062 – forty-nine years from now? In that case, giving now leaves Ethiopia with that $150,000, and waiting till later leaves them with nothing.
I don’t know enough math to do the integral properly, but I can do it at different points and then sum it up. If the world ends in ten years, saving loses $15,000. If the world ends in 20 years, saving loses $25,000. In 30, $45,000. In 40, $70,000. At 49.999, $150,000. If there’s an equal chance of the world ending at any one of those times, on average the world ending before you can donate loses you $60,000. But if the world doesn’t end, saving gains you an extra $150,000. So unless you think the world is more than 70% certain to end before you die, saving like Robin suggests is the best option.
And okay, there are so many problems with this analysis I don’t even know where to start (and I bet commenters will point out ones I missed). I hope someone can do some more rigorous math on this question. But the Fermi calculation gives me the opposite result from the one I was expecting and this is very awkward and I was totally intending to close up this essay with “and therefore, math tells us investing charitable donations is clearly a bad idea.” Instead I’m just going to keep recoiling and gibbering.
I’m reminded of an interview with the seti@home people when that was getting started (in the late 1990s, when dedicated video hardware in PCs was not universal). He pointed out that the pretty screensaver, which could be turned off, cut the CPU available for actually crunching numbers by about 50% – but going from the original blank-screen-only version to the screensaver version vastly more than doubled the number of people prepared to run it, so it was still a huge gain for them.
So don’t underestimate the value of “trying to look good” as a motivator. Should a hypothetical new charity try to appeal to the tiny community of self-conscious rationalists, or the huge community of people who want to look good in front of their friends?
This still leaves open the question of what we as individual donors should do, knowing all this. With the SETI example, maybe it makes sense to package your program with a screensaver, and then not use it when running seti@home on your own computer.
Interesting – I’m actually sort of relieved at the thought of a single big donation at the end of my lifetime being the most effective form of charity, because I like building up my savings, and spending those savings always feels a little painful.
Of course, I’m not at all certain that we’ll still have a civilization after that time. But I’m not 70% certain that we won’t, either. I hope that somebody’ll check your calculations and find them to be approximately correct.
“So let’s get to the fun part. How do we debunk Robin’s assertion that we should invest charitable givings and donate them only at the end of our lives?”
Huh – I’d approach it from the other end. There’s an Irish proverb “Live, horse, and get grass!” (You may more readily know Shakespeare’s version “While the grass grows, the steed starves”).
What is the ultimate purpose of giving to charity? To relieve some actual want or need. Let us say there has been a flood in the region and a proportion of the villagers have nothing but the clothes they are standing up in, which they wore as they fled the rising waters. A collection is being held in the public square and on the street corners of your town to get money to buy new clothing, food, and tents for them to live in until the flood waters have abated and they can go back and clear out the debris in their houses.
It may be more efficient to say “I could give you ten dollars right now, but if you prefer to wait forty years, the amount after investment will be much more and you can buy nicer clothes or better tents”, but does it achieve the end for which you intend to donate that money? In the meantime, while your ten dollars are earning interest for four decades – the people are living where? eating what?
Now, for certain long-term projects that are going to take a lot of time to achieve their purpose (e.g. cancer research charities, which I often see advertising in the backs of periodicals over here to ‘remember us in your will’ and asking for bequests), it may make more sense to adopt this method of donating money, but for others, well – “Live, horse, and get grass!”!
Though I do agree about the big flashy charity galas – if you’re going to buy a new dress and show off your family jewels at a glittering dinner and social function, why not instead total up the cost of the ticket etc., give that money directly, stay at home and if you really want your name and photo in the paper, take out an ad for pete’s sake?
Perhaps I am just prejudiced on the topic due to cultural biases, but one example why measuring charity by efficiency strikes poorly on Irish ears.
The (possibly bigger?) middle-to-upper-middle class version of those is the thing where they send young first-worlder college and high-school students (often promising professionals) to do gruntwork in reasonably safe third-world countries, sometimes utilizing oddly colonialist-sounding language.
Well, I guess the Hanson-type answer to this is that natural disasters happen all the time, and it doesn’t make sense to value a flood victim now over a flood victim forty years hence. And because investments grow faster than the growth rate, your choice is something like either saving one person now, or three people in forty years. If you give to a flood relief charity now instead, you are effectively damning two (net) people to starvation and homelessness, you monster! 🙂
Your post is saying basically, “Who cares about being efficient, I really want to do good.” But… doing the most good is what being efficient *means,* in this context.
Also, I read your link about the Doolough Tragedy. I don’t think you can blame that on efficiency-mindedness. Whatever caused the inspectors to get lost, and then insist the whole town come to them, it wasn’t a desire to do the greatest good for the greatest number. They had all the power, and they didn’t want to move, so they didn’t. That has basically nothing to do with thinking about how to help the most people per dollar. The actions of the officials in that story were highly inefficient, and GiveWell would not recommend them anytime soon.
If you are funding a charity to research how best to prevent floods, then the long-term strategy makes sense.
If you are trying to feed, clothe and shelter a flood victim right now, then the long-term investment sounds like the Epistle of James:
“15 And if a brother or sister be naked, and want daily food:
16 And one of you say to them: Go in peace, be ye warmed and filled; yet give them not those things that are necessary for the body, what shall it profit?”
“Don’t help in the current disaster because you’ll be able to help more in a future disaster” leaves out any possible contributions (including both paid and charitable) that the people who were helped in the first disaster might make.
But Delphi was about efficiency. There has been a lot of folkloric addition to the story, of the kind that naturally accretes when an event lingers in the memory of people and is repeated over and over as an exemplar or cautionary tale. The idea that the starving were turned away because the poorhouse officials were too busy eating their own meal is probably one of those curlicues that get added.
But the bones of it is that the inspectors were not wicked or selfish men; they were sent over by the government to investigate and carry out the most efficient use of relief. The government didn’t want to just throw money at the problem, it wanted to use in the best way the resources it had, to give the applicants for relief a sense of independence and foster the virtues of self-reliance and hard work in them, to get needed repairs and infrastructural work done to benefit the community, and to make sure fraud, scamming and the likes were not perpetrated at the expense of the public purse.
However, making sure that your starving, sick beggars really are genuinely starving and sick and not ‘pretending’ or ‘sponging’ or “welfare queens”, and that they should work for their dole of corn and money instead of just getting easy quick ‘food now’ into their hands which would be gone as soon as eaten and do them no good in giving them marketable employment skills, and that the business of government is not to interfere with the workings of the market which will sort all this out in time, and that disbursements from the public purse should get gains which will benefit the public, including the people claiming relief – well, sometimes they have unintended consequences.
Like bad weather meaning the plans of the inspectors to turn up at such a place on such a day at such a time being disrupted; like the starving sick being, well, starving and sick; like people who need food thinking that a public relief scheme means they will be given food to eat now, not that they will be assessed can they work for a wage at the end of the week which will let them buy a certain amount of cheap(er) grain.
Like people being so inconsiderate as to die, instead of living on air for ten years until the investment comes to maturity.
I think, if you fell into a lake and were drowning, you would prefer a ‘do-gooder’ who only wanted to ‘feel good’ who jumped in to pull you out, over someone who said “Hang on while I set up a foundation to train qualified life-savers!”!
I think you’re being confused here.
Suppose that I have $10 now, which will become $100 in fifty years. Suppose there are floods every year, and it costs $10 to save one flood victim.
Yes, if I spend the money now I can save one flood victim. If I spend the money in fifty years, I can save ten.
Although the one flood victim I save now will be very glad I did it now rather than later, the ten people who will die later will wish I had saved the money.
I’m not sure why you’re privileging the guy right now.
For one thing, we’d like to be helped if we need it now, not have helpers help someone else 40 years from now. We privilege the now because we live in it.
There’s also the lost economic output of the person you didn’t save. Which gets back to the fallacy of arguing from average growth rates.
More uncertainty: is it true that the interest rate has been higher than growth rates for 3000 years? How is that possible and what does it mean?
…so basically your argument is that because some people who claimed to be efficient were inefficient no one should try to be efficient?
This is a somewhat bad example, in that emergency disaster relief is thought to be “one of the least cost-effective health activities”, and so wouldn’t be chosen by someone trying to maximally optimize the efficiency of their giving.
Oh, God forbid we save the lives of the drowning instead of maximising our optimal donation strategy!
Er, yes, God forbid we cause people, on net, to die.
Eh, reading the Givewell post, it seems to be saying (maybe) that donating to help people in Disaster X is unlikely to save many lives; by the time the money can be used the people are saved or dead. Donating to the *capability* to help people in disasters is probably more robust. Donating to improving the quality of life of the victims, well, then there’s whether a Japanese or Kansan who’s lost their home is still better off than your everday Kenyan.
(And, conversely, whether we donate neither as charity nor to make ourselves feel good but as a social contract of mutual aid: I help you, you help me.)
Oh god forbid that we dare to check whether we are actually doing what we are trying to do?
“I should note that this chapter is less thoroughly referenced than most others in the report,” “We don’t at this point endorse the chapter’s conclusions” (and that’s a comment 1.5 years later). Seems more uncertain that your comment implies. Has there been more in the past 3-5 years? Also a strong implication that donating for *this* disaster may not save lives *now*, but donating to the organizations can help saves lives in disasters in general.
It’s not helped by the way disasters pull oxygen out of the room, so the charities have stuff earmarked and have to manage it. . . even perhaps more money than they need.
Never earmark a donation. If you don’t trust the charity to spend it wisely, don’t donate.
(I still remember the little pamphlet talking about doing marketing research for peasant handicrafts so they knew which ones they could sell the best.)
This articulates my objection pretty well.
Hanson’s point falls into the trap of thinking that more money later is better than some money now. If all we’re doing is shutting up and multiplying, this is certainly arguable (although I’m mildly suspicious of utility aggregation across time and very suspicious of utility aggregation across people).
But consider the (vastly simplified) case of it taking n billion dollars to cure malaria. As soon as we can gather n billion dollars, poof, no more malaria. In this case, we very much want as many people to invest money as soon as possible, because as soon as we cure malaria, that’s a huge chunk of disutility that is gone forever.
Obviously this oversimplifies–maybe we can invest for five years and then cure two diseases, and this could multiply out better. On the other hand, it reflects things that really have really happened in the world–we’ve essentially eradicated polio, and charitable contributions were huge there. At the very least, I think it points towards a flaw in Hanson’s reasoning; change now is better than change later, and this is particularly true with permanent change.
If charities think “invest money and use it later” is more efficient then I think they can do that. That is – I’ll give them the money and they can either invest or spend it. Extra bonus; you get better returns on bigger sums.
I’ve heard it claimed that charities are excessively cautious about investing, because donors so often dislike it their funds not being used for “something real”.
They are. They also object when charities spend money on PR and on salaries. People ought to work cheap for a charity, and the devil take the notion that better talent might make the charity even more efficient.
Maybe they want to invest, but they know they would face outrage from irrational donors if they don’t spend donations immediately. An individual, not subject to the same kind of scrutiny, can act more strategically.
Charities may have less choice in the matter than you think, as part of the long-standing legal effort to destroy perpetuities and long-lived institutions: remember that at least in the US, charities are required to disburse a percentage of funds every year. So to be investment vehicles realizing real returns, they need to not just beat inflation and the huge risk premia, they need to beat their disbursement as well or else eat the penalty.
This is just one reason among many that perpetuities are an *awful* idea for charities, and the reasons are sufficiently numerous and clear (some Hanson has blogged about in other contexts, some I’ve pointed out to him repeatedly) that I’m really surprised to hear from Yvain that Hanson is still pushing the perpetuity idea. It recently came up on LW again (http://lesswrong.com/lw/h3q/a_rational_altruist_punch_in_the_stomach/) and I posted a semi-comprehensive list of reasons against it (http://lesswrong.com/lw/h3q/a_rational_altruist_punch_in_the_stomach/8oq7).
Hanson seems to take into account the problems with long-term trusts, and instead suggests donating shortly before death. Mortality concerns are a lot weaker for individuals, and Scott addressed xrisk and opportunity cost above.
He did use a 7% rate of return, so reason 3 might still be a useful response.
> Hanson seems to take into account the problems with long-term trusts, and instead suggests donating shortly before death.
? How does this help much? The ‘great filter’ for perpetuities, if I may borrow a favorite phrase of Hanson which he has not seen fit to apply to his latest King Charles’s head, is not being contested in probate.
> Mortality concerns are a lot weaker for individuals
? I don’t know what you mean by this.
If you happen to die from non-immediate causes, you can personally wire money to the charity, and even if you don’t anticipate your death, a will that donates money is more likely to be carried out than one that creates a perpetuity.
>I don’t know what you mean by this.
People are less likely to die and be unable to transfer the money than organizations are.
Ah. Well, I suppose that helps a *little*.
Hanson agreed that it was a bad idea to try to save after your death, and instead suggested saving only until your retirement or some other point at which you were old but pretty sure you could reach before dying.
One notes that one’s own funding schema end with death and donation. The charity’s could go on in perpetuity without ever benefiting those intended.
Yeah, that’s an interesting point: simplistically the logic suggests we should never donate to charity; 40 years from now the logic would still apply, and we should keep investing that big lump sum so we can theoretically save even more people in the further future.
Donating money to charity makes you the kind of person who is more willing to donate to charity. Not donating to charity makes you the kind of person who isn’t as willing to donate to charity. Even if you want to donate the vast majority of your wealth right before you die, you’re more likely to do so if you regularly make donations.
And you can also use scope insensitivity to your benefit. If donating to lots of charities makes you feel good, it doesn’t really matter how much you donate to those charities. If you’re already giving $10M to the most efficient charity, giving a thousand dollars to the “save puppies from getting kicked fund” is more likely to make you feel good about yourself and do better financially/healthily and all those other nice knock-on effects than it is going to harm the most efficient charity by robbing it of .01% of your charitable efforts.
yeah, that’s totally a thing.
virtue ethics works.
The second paragraph looks like standard consequentialist reaosning to me.
Due to the idea of diminishing returns, I doubt that we are running out of good that needs to be done. For example malaria is more difficult to eradicate than smallpox because it has a non-human reservoir – mosquitos.
I do thinks it is an interesting questions though, whether it is better to invest X dollars in a world that has many urgent needs (the present), or to invest more than X dollars in a world that has few needs than us.
Also, I think the psychology argument is the most convincing – that few would be willing to part with a huge lump sum rather than giving it to their family and/or spending it.
That’s actually the point. We don’t “run out of good” completely, because of diminishing returns. But those same diminishing returns means that doing good becomes harder and more expensive. Someone who declined to give money to smallpox in order to save money so that the compound interest could do good in 2013 might find that their best option was malaria, and that their money actually saved fewer lives than if it had been spent immediately because malaria is a harder problem than smallpox.
I hadn’t though of that, thanks!
The categorical imperative should also perhaps be in play here. If everyone gives to charity now, some good gets done; if everyone saves it until just before they die, the charities shut down, their workers go away to do other things, and there’s no organisation to which you can give the money later.
Except then you can change the rule to “invest the money you want to give to charity, unless enough people are investing the money you want to give to charity that charities start shutting down” and in the vast majority of possible worlds you should be investing your money right now anyway.
That’s only a problem if everyone suddenly switches to this strategy all at once. In equilibrium, there would be a steady stream of donations-at-deaths, just like there’s currently a steady stream of donations-as-fraction-of-income. And even if charities did shut down, you could create new ones later.
I suspect you are overly credulous regarding the 7%+inflation for 50 years investment opportunities you’ve found.
Also, there’s more than just the singularity that can render your investment accounts moot, some with rather higher liklihood.
But then again, you are rounding up from 115,000 to 150,000 for rather subjective reasons, so precision obviously isn’t a priority. (Your saved money can also be used by a couple generations if you dole it out or give it to wise people or give it in ways that benefit multiple generations, so I don’t think that rounding is merited).
I think you’d want to think about the rate of return from charity.
I mean, let’s say you save someone’s life. That person grows up into an adult, an adult the world otherwise wouldn’t have had. What good will that person do, across his/her life? How much money will that person donate to charity, on average? How many children will that person have, and what good will they do? Will that number be more or less than the interest rate?
Let’s say you find a way to give startup cash to a legitimate small business, somewhere far away with a developing economy. How much utility will that business generate, for its customers and for its owners? Will that number be more or less than the interest rate?
Ummm… the person you save is saved not that he may save others, but that he may live?
Optimizing charity methods is a completely different problem. And it is important.
But you are trying to calculate how to save the most lives, the consequences of saving lives now instead of later also have to be taken into account to make the analysis rightly. If you save one child from malnutrition now, or ten in twenty years, the question arises of whether saving that one child could save twenty in twenty years.
Timeless decision theory may be important here. One might adopt a decision procedure such that, were every similar agent to use it, the singularity or any such massive hedonic event would happen and turn out well. I’m not sure that Hanson’s conclusion about donating to a single charity generalises automatically, should one for TDT reasons consider other minds than one’s own.
Supposing all other human donations are held constant, one might be wise to donate to a single best charity; but should one’s decision be instantiated in a plethora of similar minds, one might need to consider that smaller charities are systematically neglected in favour of big and obvious ones. I.e., might MIRI already be underfunded for precisely this reason—that other, similar minds with different beliefs about the singularity would never dream of chipping in a few dollars to some weird AI science institute, since this detracts from single best causes like Cancer Research?
Secondly, we are not free to behave as ideal decision agents; rather, human minds contain competing sub-processes. One’s mind is an unstable and incoherent willing, the better part of which (by its own standards) must cajole the rest. Should I-gestalt be motivated to donate or contribute to multiple good causes, that is something I-better cannot change by fiat. Should my better part attempt to trammel the worser, it might well lose—in which case all my income is lost to whiskey and cigars.
Hmmmm… My model has always been that the amount of funding to MIRI is unlikely to affect things much (also, I believe that there’s a maybe 20 percent chance of singularity in the next 200 years, otherwise status quo?
I admire and have donated a small sum to MIRI, but I too have my reservations.
Take this erudite piece.
I don’t believe that a superintelligent AI would form such ideas. Selfhood is a crude concept, which is intuitive to humans for evolutionary, and not decision theoretic reasons.
The AI is more likely to view “itself” as a set of intertwined dynamic processes, which keep one another in check. “It” doesn’t start “its” sentences with “I”. Rather,
“[Process G says]Process F assigns 30% probability to ‘XYZ’, thus [Process H resumes]manual extensor should rotate 3 degrees…”
Well…you get the idea. Eliezer & co., magnificent and humane men though they are, are like the proverbial drunk looking for his keys under the lampshade. It seems almost plausible that one might use a priori mathematics to design an AI that shares our folk intuitions about selfhood; it is implausible that such a theorist might design a messy Buddhist reasoner like the above.
I also find “CEV” implausible. Few humans, as they are now, would consent to technological eschaton, and would scarcely be placated by Eliezer’s promise to plug all their brains, too, into God.
So, long before any superintelligence is built, humans need to change. We need Philip K Dick’s empathy box, that we might become one cooperative superentity, which can delay the singularity as long it deems necessary. Collective consciousness, unlike CEV, can develop incrementally, and is a smaller threat of the unknown.
We, once conjoined, need not purely mathematical but rough-and-ready protocols, and not provably but probably safe AI experimentation and engineering, with which theorists keep pace.
Now having said this, MIRI deserve funds more than any other charity or institution. They can scarcely do harm, because the status-community they are creating is most valuable, even if they have the specifics wrong.
By the way Scott Alexander, I wish not to be impertinent but this comment box is a little faulty. One can stretch the box from its bottom right-hand corner, but should one pull it too far to the right it disappears behind the frame, at which point should one let go of the corner it cannot be recaptured. Ce n’est pas normal!
To expand on James’ latter point, if your preferences are best modeled by a parliament of subagents, some of whom are altruistic, you would in the best case donate to several charities, each of which is optimal for a different conception of the good. I believe this kind of thing is responsible for some people’s desire to donate to multiple charities, in addition to the less altruistic motives of wanting to look or feel good and a desire to maximize the chances of doing some good. Of course, if you donate to more than ten charities you’re either Bill Gates or you’re doing it wrong.
“each of which is optimal for a different conception of the good”
I like that.
All right, ladies, gentlemen and others: let me put it into terms you may understand better.
You – yes, you right there, not some hypothetical person in a foreign country – have a pain in your abdomen. It’s severe enough to have you waking up at four o’clock in the morning crying with the pain. It lasts several days. Over-the-counter painkillers, even at doses you worry will do damage to your liver, only dial it down for a couple of hours and then it comes back.
So you go to your doctor, who sends you to your local hospital to find out is it appendicitis, is it kidney stones, is it gall bladder, what is it? Which would you – real you, not hypothetical you – prefer?
(a) Doctor sees you, takes symptoms, sends you for a scan to see if it is your kidney or your gall bladder or your appendix or what is it that is making you cry with pain?
(b) Doctor sees you, takes symptoms, tells you this is fascinating since – for women, you know – there are so many things a pain in the lower right quadrant of the abdomen could be. It doesn’t look like appendicitis, so there’s no danger you are going to die in the night (it just feels like you are going to die in the night). Would you be interested in signing up for a research project where your first interview with the consultant will be in three years’ time? That is the maximally optimal efficient way to decide what this mystery pain is, not wasting money on ultrasounds and other hit-and-miss testing to see if it’s this or is it that!
What do you – real you, ladies and gentlemen, suffering in your real selves real pain – want to do? Maximise a strategy or get relief now? And then recommend your decision to the poor and sick of the world – so long as you are willing to abide by the same rules!
We understand what you’re saying. We know that every single person that drowns is a tragedy. We know that at some point we may be the people that need help, and when we need that help it will be no consolation whatsoever if the people in power are choosing to help other people far away, in the future.
But if we live in a world where we always choose to help 1 person now, here, instead of 10 people far away or in the future, then we are *condemning 10 people to death.* Real people, not hypothetical people. People’s pain doesn’t become less real when it is far away.
You keep giving examples of “efficiency” that are actually inefficient and poorly thought out.
Saying “I’m efficient” doesn’t mean you’re actually being efficient or effective. Saying “I’m compassionate” likewise doesn’t actually mean that you’re helping people. They’re just words.
What matters is, are you helping people, or not? Are you helping the most people you could be, or not?
We have medicine to treat sickness and dikes to prevent floods and engineered grain that can feed hundreds of times as many people as older methods of agriculture. We are able to accomplish these things BECAUSE compassionate people didn’t just sit around helping whoever was nearby, they thought about the future, they planned, they researched, and developed new tools, new drugs, new systems of agriculture.
If I get sick or hungry or am drowning and need help, I don’t want to live in the world where people do whatever is easiest to help people *now*, because then I’d be relying on people trying to treat my illness with leeches, save me from drowning with fishing nets instead of a coast guard boat, or feeding me a fish instead of teaching me to fish (or, better yet, going to school for a lucrative career in skilled, specialized labor).
(Since it sounds like you do have some background that leads you to be somewhat justifiably skeptical in people who claim they are ‘efficient’, I should also clarify: I don’t want to live in the world where everyone *says* they care about efficiency. There are a lot of ways that can go wrong. But I DO want to live in the world where people *actually are efficient* about helping)
Perhaps the one person you save now will save twenty people in the future.
Perhaps the ten people will not drown because a extinction-level-event-sized asteroid hits the earth and we all die.
Perhaps, but that doesn’t mean you shrug and save the person now, that means you actually think about what is likely to happen and make the best decision.
It is very plausible that helping other people to save lives is a viable strategy. That doesn’t mean saving random people and hoping they go on to save lives. It means systematically teaching people how to save lives. (Giving What We Can, Givewell, and the Center for Applied Rationality all do this in one fashion or another).
Extinction-level asteroid events are a real possibility – and we have, in fact, put together programs to keep a look out for asteroids. Those programs cost money to maintain, but that funding is already part of government budgets, so an individual altruist doesn’t need to worry about supporting it themselves.
These have real answers, “maybe X will happen” isn’t an argument all by itself.
But “Maybe X will happen” is the ENTIRE argument for deferring one’s charity until just before death.
But it’s not an “either/or”, it’s a “both/and” situation. Yes, by all means, invest so that you can leave a lump sum bequest to your favoured charity when you die. Yes, by all means, investigate which charity produces the best returns on the monies it receives (I am very sceptical that the professional fund-raisers hired to be bosses and run charities on business-lines do much more than bump up salaries for themselves and generate income mainly to cover the raised overheads).
What I am saying is that it is not balanced to rush from the right side of the boat to the left side. While you (in general, not any ‘you on here in particular’) are pluming yourself that you are not like those do-gooders over there getting their vicarious little thrills through conspicuous donation to buy tins of tuna right now for the starving kittens, you instead have soberly invested in a company researching ways that in ten years’ time will produce a way that kittens can eat grass and so don’t need tuna to live on – well, while you are patting yourself on the back so hard you’re in danger of breaking your arm, over all the future kittens who won’t die, remember the ones actually starving in front of you right now would appreciate even a tin of sardines.
I’m not sure what exactly you’re asking people here to do (generic people on this comment board or hypothetical people elsewhere).
Do you think I should be donating tin cans to cats, or do you think I feel insufficiently bad about the cats that exist that I’m not helping? (Metaphorically or otherwise). You seem to be assuming that people who self-identify as effective altruists don’t know or care that there are problems happening in the here and now.
If everyone else is on the right side of the boat and not in a hurry to move, you probably should rush to the left.
One efficient strategy, would be to invest for profit in companies that are trying to make their profit by creating something of permanent good: such as developing clean energy. Especially if they are doing this in some village that has no practical energy source now, and employing villagers who need the money now.
“Ethiopia has a growth rate of about 7%”
Which gets at a big flaw in Robin’s argument, comparing the interest rate to the average growth rate. What if the growth rate for charity targets is much higher, as with GiveDirectly? What if we care about helping people now, regardless of whether we can call it “growth”?
Given cognitive bias, we can be suspicious of stuff that makes us feel good, but also about arguments that justify our not doing anything.
“only ever donate to one charity – the most effective”
Does that hold if you’re not certain about what the most effective one is? Doesn’t uncertainty make it something of a multi-armed bandit problem?
There’s also “what if everyone did this, is it stable?” If everyone did this and agreed on the most effective charity, that one would get swamped likely with more money than it can use and the others would starve. Then the herd might go focus on some other victim.
So everyone spends money on the highest priority thing until it has enough money, and then they spend money on the second highest priority thing. I don’t see what the problem is there…?
“swamped with more money than it can use”. That’s very wasteful, not efficient. And if the other organizations close for lack of donations, then when you’re ready to turn your attention there’s no one to take your money.
That’s not what we’ve seen with GiveWell’s recommendations. They recommended Village Reach for a few years, people gave money to Village Reach, and after a few years Village Reach ran out of room for funding and GiveWell no longer recommended them as their top charity (although they still thought highly of their work). There was no swamping.
I worked at a larger charity where swamping did actually happen at the time of the Haiti earthquake. We took in money hand over fist until we decided we didn’t have room for more funding and stopped taking donations for that particular purpose. So people donated to our other work, or donated elsewhere. It worked.
The delay between the time something stops being top priority and the time people stop spending on it? 😉
The problem with donating only to the most effective charity is the same as putting all your retirement funds into your best bet for investment: what if you are wrong?
What if your most effective charity got its effectiveness from investing with Bernie Maddox? A good number of charities crashed and burned after the Ponzi scheme was revealed. While it seems that due diligence would have turned up the problem with Maddox, because a number of investors did — is that really feasible for a donor?
The two situations aren’t comparable. Your utility is highly nonlinear in money, and in particular you suffer substantial negative utility if your money gets too low for you to easily live. In such a situation it makes sense to reduce the variance of your money because that improves the expected value of your utility. But the world’s utility is probably approximately linear in your charitable donations, so as an effective altruist you can afford to be much more risk-loving (as viewed in terms of money) than in your daily life.
“Probably appromixately”? On what grounds do you make this assertion?
The world is big, and if you aren’t incredibly rich, your charitable donations don’t affect it much. Nice functions can be approximated by linear functions if the input is only allowed to vary a little bit; that’s the basic lesson of single-variable calculus.
The same thing occurred to me. If that lady in the second row had continued her indignant speech for too much longer, I was seriously going to stand up and tell her that some of us came there specifically to hear the corrupter of the youth…
I’d be interested in a more detailed description of what happened here.
The short version is that an audience member called Robin Hanson’s model of human psychology an insult to intelligence everywhere or something like that. The specific words escape me.
The audience member in question seemed to actually be saying that Robin was oversimplifying his model such that he was ignoring details that actually end up making a big difference. (I might be steelmanning her a bit, though. I have a tendency to do that. She was definitely speaking in a way that didn’t pattern-match to the way rationalists talk, but that doesn’t convince me that she’s wrong.)
I didn’t claim that she was wrong. She was obviously right. But she phrased her point in a way that made it impossible to respond to, which is a rhetorical move I dislike.
…wait, what? How did you know about that lady in the second row? I thought you were in Wisconsin or somewhere? How are you in Berkeley and I still haven’t met you? Did I meet you and have no idea who you were? Or is this something that needs to be fixed immediately?
Patrick is in town for the MIRI workshop.
My primary worry with the investment plan is that future selves may not be as altruistic as my present self is– particularly taking the outside view, in which idealistic college students with odd sociopolitical beliefs tend to become less idealistic and have less odd sociopolitical beliefs shortly after they graduate.
…It seems like the solution is some kind of legal “returns from this investment can only be spent on charity” thing, but IDK how to implement that.
Buy some stock now and give it to the charity so *they* can invest it?
To paraphrase a certain reactionary, one can go from liberal to conservative without changing a single opinion by the simple expedient of waiting thirty years.
Yes, but that sounds more like ‘dangerous radical to Rockefeller Republican’ and would not diminish interest in charity.
A bigger risk is that once you get older you may be needier, get into more costly status contests, etc.
One thing that came up in the post-talk discussion is that there’s a big opportunity for doing more marginal good by donating countercyclically: instead of saving up until the end of your life, you just save up until there’s an economic downturn and donate it then, because charities do in fact get way less in donations during downturns, precisely when many of them need to be doing the most activity. (Which is further support for Robin’s thesis that people donate in order to feel good rather than do good!)
Good post, but there’s a flaw in your Ethiopia calculations. You are effectively assuming that the internal rate of return from donations INSTANTLY falls to the rate of growth for the Ethiopian economy as a whole, i.e. that your intervention is no better than cash distributed evenly across the economy weighted by existing cash holdings. But the point of funding bednets, or even cash transfers targeted at the poor, is that you get much higher rates of return for a while, which multiply the impact of your spending before it settles down to local growth rates. If you have a good intervention that is generating $10 of value for locals per dollar you spend, before returns converge to local growth rates, that makes for a rather significant change.
Whether 7% is reasonable here has a big effect on the argument. I would have (with a similar level of not knowing much about this) guessed closer to 3%. Your $10K after 50 years is $300K at 7% but only $45K at 3%.
If my personal finance professor is to be trusted (and he has worked as a financial advisor for many years), then 8% is what people typically earn when investing (real growth of course).
Why does your personal finance professor think that? For example, Franklin’s long investment only earned about 2%/year.
Investment advisors in the US offer figures based on the US without taking into account survivorship bias (which is severe many other markets were confiscated, or collapsed), taxes, and transaction costs into account. The periods they use are also typically gerrymandered, and include a large increase in P/E ratios.
More looking turns up:
Yeah, ~7% before counting inflation is consistent with what I’ve read (thus the ~4% safe withdrawal rate for financial independence — assumption being that despite performance being a bit under that after accounting for inflation, you’ll still do bits of things for money, too).
Quick sanity checks:
A source for avg 3.2% US inflation rate.
Looking at vanguard TSM index for the last 20 years (sampled at start of 1993 and 2013, though I haven’t looked to see if they’re representative samples), it’s had a CAGR of 6.3%.
The survivorship bias issue is pretty concerning for these 3-4% beyond inflation growth rates estimates in this context.
[Robin Hanson says that] the real rate of return on investment has been higher than the growth rate…his classic example of Benjamin Franklin
Did he give any numbers? Did he say how much you should be able to beat the growth rate? Did he even say how much Franklin managed? Here he says Franklin achieved real returns of 1.8% in Philadelphia and 2.2% in Boston. But he got the cities to donate the labor of administering the investments, so this overestimates his returns. According to this site, growth 1790-1990 was 1.7%.
Franklin wrote as if he would achieve 5% growth. But that’s absurd, because he only charged 5% interest. I think he lost a point to defaults and two points to inflation.
>He said that the world is getting better so quickly that we are running out of good to be done.
There should be a graph of givewell’s best estimate for dollars:lives saved for the most effective charity over time. When the cost to save a life, or buy QALY’s goes up, we are doing well.
This got discussed, but St. Elie said that their changing estimates of effectiveness were more likely due to noise in how good they were at estimating than anything else. He told the story of how they had recently downgraded AMF from 1 life/$1000 to 1 life/$2300 just because of better calculations. With that level of noise, I don’t know how useful a graph like that would be.
I’ve been pursuing both strategies. Donating as much as I can today (currently $1000 a month), and putting money into retirement accounts, which I then sign to the same charity as the beneficiary on my death.
The current plan is to avoid end-of-life health measures (80% of lifetime health spending is in last 5 years?! No thanks) and give away whatever money I have left before people can steal it from me in the form of prolonging low quality-of-life experience.
Pingback: Does anyone give money to charity efficiently?
If you can show me any investment vehicle offering 7% above inflation, I will give you my life savings. That seems like a serious over-estimate.
My guess is that for now investing is probably better. You don’t have to make a global decision, invest for 50 years or donate now. You go invest for 1 year and reevaluate the decision in 2014.
1. Is charity becoming less effective? Low hanging fruit is getting picked, but GiveWell and others are discovering new low hanging fruit, maybe even lower hanging fruit. If you gave in 2003 odds are it would have been to something a lot less effective per dollar than AMF in 2013.
2. Waiting to decide gives you option value.
3. If anyone really believed all the low hanging fruit was getting picked they’d take out a loan to shift next years donations into the present when they are more effective. But no one does. (Note: this isn’t because of high interest on the loans, you could get a interest free card for 18 months and max it out.)
Pingback: When your name is a verb | Blunt Object
Pingback: Overcoming Bias : More Now Means Less Later
Pingback: Soup kitchens are for the soul, not the needy | oblivia
The problem as I see it is much more fundamental: is charity, in itself, an efficient way to effect meaningful change?
I was inspired to write a blog post about it here: http://obliviablog.wordpress.com/2013/04/08/soup-kitchens-are-for-the-soul-not-the-needy/
Pingback: Your Charitable Giving Is Irrational: You Care More About “Looking Good” Than “Doing Good” | Paul M. Jones
Pingback: Assorted Links | azmytheconomics
Pingback: Somewhere else, part 43 | Freakonometrics
I think the flaw in his argument is that the growth rate is not necessarily a single uniform thing – especially when dealing with kinds of utility without immediate financial equivalents. Other commenters have mentioned spending your money today to eradicate a disease, or saving a child’s life and considering the positive impact they and their descendants will make in the future.
You could also directly improve the growth rate with your contribution now. Suppose you dig a well and build a school in a given community. Then come back in fifty years and there’ll be two generations of educated people, all of whom have an extra couple of hours in their working day from not having to hike to the river and back to get water. That will have a massive impact on the local economy, and might be enough for them to pull themselves out of poverty so they don’t even need the extra aid you were going to give them just before you died.
Disutility also undergoes growth if left unchecked: you could repair a leaky dam now, or rehouse all the people whose homes were destroyed by flood later; you could educate people about safe sex and give them condoms now, or care for lots of AIDS orphans later.
Pingback: Charity for Corporations « Bayesian Investor Blog
Undeniably imagine that that you stated. Your favorite justification appeared to be
at the web the simplest factor to remember of.
I say to you, I definitely get annoyed even as other people
think about issues that they just don’t recognise about. You managed to hit the nail upon the highest and outlined out the entire thing with no need side effect , folks can take a signal. Will likely be back to get more. Thanks
Pingback: Don’t Donate Your Points to Support Tornado Victims! - View from the Wing
Pingback: Giving Now vs. Later: a Summary | The Centre for Effective Altruism
Pingback: Giving Now Currently Seems to Beat Giving Later | The Centre for Effective Altruism
Hello, couldn’t find you on cardsapp. how can I find you? thanks