{"items": [{"author": "Mac", "source_link": "https://www.facebook.com/jefftk/posts/373797882638271?comment_id=373854435965949", "anchor": "fb-373854435965949", "service": "fb", "text": "\"Cost effectivity\" is a very painful concept when it so brutally quantizes human suffering in terms of dollars. And it is the most effective means of aleviating that suffering. As long as you have your metrics right.", "timestamp": "1330705437"}, {"author": "Ivan", "source_link": "https://plus.google.com/101147004225363019038", "anchor": "gp-1330716425807", "service": "gp", "text": "i've often seen you make this claim, and while i'm open to the notion, i remain unconvinced.\n
\n
\nif i understand you correctly, you're trying to maximize the impact of your contribution on the overall good in the world; i don't know how you formulate your valuation of good, but it appears to be inversely proportional to suffering at least in most cases. so you examine the known problems that exist in the world, and you try to determine a probabilistically optimal contribution that you can make in the scope of each problem. on this level i think we're on the same page.\n
\n
\nbut i'm not convinced that single issue charities are actually optimal in the sense of having the most impact on the sum of all problems per dollar or hour that you contribute.\n
\n
\ni totally agree that they are optimal in an extremely localized context. there are many groups of people around the world who have various medical or economic or social ailments: AIDS, Guinea worms, malaria, poverty, slavery, oppressive governments, oppressive communities, and so on. they are facing consequences of these problems every day. pick a problem, an affected group, and an effective remedy; estimate the efficacy of your contribution. repeat for all combinations of problems, groups, and remedies, then determine the winning combination. as long as your efficacy estimates are reasonably close to the true efficacies, you will certainly do the most \nimmediate\n good by making all of your contributions to the winning combination.\n
\n
\nbut what about contributions that have long term impacts on large groups of people? there are many ways to contribute to needy groups that will, in the long run, effectively prevent them from facing many of the problems that they would otherwise face. if the village plagued by Guinea worms had better access to clean water and sanitation, then it would likely have never taken hold (this has been effectively done on the scale of entire countries) and they wouldn't need to be dewormed. if the family of the malaria-stricken child had access to a clinic that provides community-funded mosquito nets, then they wouldn't need donated ones. if the family who sold their child into slavery had access to a series of microloans to set up a market stand, then their child wouldn't need to be found and saved. and so on.\n
\n
\nthere are a few other things that i find tenuous about your valuation of single issue charities. for instance, it's rarely clear just how valuable it is to provide a community with the tools to improve itself in a sustainable way. how many families are you keeping in poverty while providing them with vaccines and mosquito nets? i don't know the answer in the short term, but in the long term, it's clearly better to help build the community, even when it comes to their access to vaccines and mosquito nets.\n
\n
\nit's the same argument as the one over remedial versus preventative medicine. sure, you can help a lot more people now if you spend their collective funds on remedies for their current ailments. but you can help countless people in much greater ways for decades to come if you spend their funds on enabling them to access preventative care and to have healthier diets and lifestyles.\n
\n
\nthe definition of \"cost effective\" is not so clear cut unless we're willing to severely restrict our scope. i'm not willing to do that, so i continue to believe that the most effective contributions i can make are the ones that empower communities and individuals to improve their own situations and to set up self-sustained feedback loops that continue to provide broad, long term improvements.\n
\n
\nbut there's more short term certainty in your approach, and that's admirable about it. and i have no doubt that in some cases, it might even be more effective in the long term. however, if the goal is to maximize overall good, it seems pretty clear that most of our investment should go into long term improvements that impact many groups in sustainably positive ways.", "timestamp": 1330716425}, {"author": "Julia", "source_link": "https://plus.google.com/102487727783123805341", "anchor": "gp-1330742960244", "service": "gp", "text": "I'm thinking about some sort of preventative thing that might be costly but prevent something even more expensive down the road. I'm thinking something like, suppose we'd identified and contained HIV before it was a worldwide pandemic, so it never was. Something like that. I guess what I'm trying to get at is how you quantify all the people whose lives and suffering you're saving by preventing them from getting sick in the first place. I don't really have a point here, just thinking.", "timestamp": 1330742960}, {"author": "David German", "source_link": "https://plus.google.com/111229345142780712481", "anchor": "gp-1330745791844", "service": "gp", "text": "\"Taken together with the difficulties in accurately predicting the long-term impacts of systemic changes, this definitional fuzziness makes it hugely difficult to evaluate progress.\"\n
\n
\n@Lucas\n I think this is exactly right, and it's why I \ndo\n agree with the \n@Jeff Kaufman\n approach. Longer-term development strategies may have a shot at bigger payoffs, but how do we tell which such programs are effective and worth funding? It's especially challenging because the charities with the most plausible-sounding development portfolios (\n@Oxfam\n, Partners in Health, etc.) are so opaque about their own metrics.", "timestamp": 1330745791}, {"author": "Jeff Kaufman", "source_link": "https://plus.google.com/103013777355236494008", "anchor": "gp-1330783002725", "service": "gp", "text": "@Ivan\n I agree with you that we should take the approach that does the most good over the long term. The idea of a development project that successfully lifts people out of poverty so they can pay for their own mosquito nets, afford shoes to avoid parasites, and have enough of a reserve to limit the huge stress of insecurity is incredibly appealing to me. But as \n@David German\n says, these programs are much harder to evaluate. It's not just that they are opaque about their metrics: you basically can't tell if a strategy works without doing controlled experiments, and to evaluate whether an approach \"breaks the cycle of poverty\" you need to run experiments that last decades. Which means that while your paragraph that continues from \"there are many ways to contribute to needy groups that will, in the long run, effectively prevent them from facing many of the problems that they would otherwise face\" makes a lot of sense to me, I don't think investing a given amount in clean water actually does better than the same amount in parasite-treatment, even when evaluated over the long term. (You put in wells, they break, no one can afford to fix them, people go back to doing what they were doing. Your charity says it does better? How do they know?)\n
\n
\n(This is actually a separate question from whether we should pick the most cost effective interventions and just do those. I think we agree on this and you just think that \"long term improvements that impact many groups in sustainably positive ways\" have higher, though more uncertain, cost effectiveness. Is this true?)", "timestamp": 1330783002}, {"author": "Boris", "source_link": "https://www.facebook.com/jefftk/posts/373797882638271?comment_id=374475059237220", "anchor": "fb-374475059237220", "service": "fb", "text": "I think you are right.
I need to think things through better and put them more eloquently, but better something than nothing:
One way to see the issue is to simplify things extremely: let's imagine a world where there are 2 things that can kill people, and NGOs are working on fixing both. As in the real world, organizations working on fixing both spend more money on interventions than organizations that work on one alone (simple principle from economics - specialization wins; I know this is too fast and it could be another way, but to be short, it looks like that's what it is in the real world in most messy cases).
Continuing the analogy, an argument some people make is that addressing only one of the two issues means the people can die of the other - and your help is for naught. But clearly many people only would have died of one thing (and not encountered the other). Here it matters how much more effective a specialized charity is than the other, and how often people encounter the two things that kill in their lives.
Super-sweet math summary:
I wish I could add a diagram: a 2x2 table (2 columns = chance of not dying from cause 1, chance of dying from cause 1; 2 rows - same but for cause 2). If the probability of NOT dying from cause 1 than dying from cause 1 (and same for cause 2) then targeting intervention 1&2 will mean you are targeting a smaller area - which is inefficient!
And a concrete example just in case:
Building a school is beneficial to children who don't have access to it (though costs a lot), but if there is a school but students are sick and can't attend - they don't get the benefits. One reason students skip school on some days is due to parasitic worms (costs $0.50/child/year to fix). Instead of giving money to build a school, it's better to give to charities that treat children for parasitic worms. There are two barriers to attending school - and one barrier is so much cheaper to eliminate. When you have limited money to give, it's best to concentrate on whatever is cheapest to fix (more fixing for the $).
It's not in principle that NGOs can't be effective at doing 2 things at once, but in practice it looks like that's the case.", "timestamp": "1330788882"}, {"author": "Todd", "source_link": "https://plus.google.com/112947709146257842066", "anchor": "gp-1330799342580", "service": "gp", "text": "One thing we could do that would help \"break the cycle of poverty\" is end subsidies for US producers of various goods, particularly agricultural ones. It's hard to produce wealth when your primary economic advantage is being neutralized by foreign governments.", "timestamp": 1330799342}, {"author": "David German", "source_link": "https://plus.google.com/111229345142780712481", "anchor": "gp-1330799479148", "service": "gp", "text": "@Todd\n Unfortunately, individual philanthropy can't do much to accomplish that.", "timestamp": 1330799479}, {"author": "Todd", "source_link": "https://plus.google.com/112947709146257842066", "anchor": "gp-1330799537176", "service": "gp", "text": "@David German\n =(", "timestamp": 1330799537}, {"author": "Ivan", "source_link": "https://plus.google.com/101147004225363019038", "anchor": "gp-1330808756559", "service": "gp", "text": "@Jeff Kaufman\n '''This is actually a separate question from whether we should pick the most cost effective interventions and just do those. I think we agree on this and you just think that \"long term improvements that impact many groups in sustainably positive ways\" have higher, though more uncertain, cost effectiveness. Is this true?'''\n
\n
\nalmost. actually my uncertainty in the effectiveness of such long term improvements leads me to diversify my social investment portfolio among several organizations. the rationale for this is the same as for portfolio diversification in personal investments. as good as you might be at selecting the charities that are most likely to be effective, there are really too many factors at play in the end results. even if you are perfectly optimal in your assessment of each charity's effectiveness with the information that's available, you still face a substantial risk of inaccuracy simply because the available information is quite limited. so while it's important to maximize your contribution to each charity (because each additional contribution tends to be more effective than the previous one), at some point that must be balanced against the need to diversify in order to defend against the inherent risk that your assessment of effectiveness won't pan out.", "timestamp": 1330808756}, {"author": "Todd", "source_link": "https://plus.google.com/112947709146257842066", "anchor": "gp-1330812363241", "service": "gp", "text": "@Ivan\n That's a good argument for diversification, but I'd guess that \n@Jeff Kaufman\n's current contribution level is not large enough to hit the point on the curve where the increase in marginal utility is offset by the increase in risk. I'd think you'd need to be contributing millions of dollars before that became an issue, if you're a good evaluator (and Jeff is if anyone is).", "timestamp": 1330812363}, {"author": "Julia", "source_link": "https://plus.google.com/114588710186521489410", "anchor": "gp-1330833357447", "service": "gp", "text": "@Ivan\n If all donors focused on a single intervention, maybe we'd wipe out malaria but lots of people would suffer or die from something else. However, I know most donors are haphazard and will fund random interventions. They're already diversifying plenty, so I may as well fund the one I think is best. (Unlike in personal investing, it doesn't matter whether any individual chooses right - it matters what we as an aggregate do.)", "timestamp": 1330833357}, {"author": "Jeff Kaufman", "source_link": "https://plus.google.com/103013777355236494008", "anchor": "gp-1330834188276", "service": "gp", "text": "@Ivan\n Given the choice between a 100% chance of saving one life and a 70% chance of saving two (30% chance of no one), I'd take the latter. Given a choice between a 100% chance of retiring with $200K and a 70% chance of $400K (30% chance of $0) I'd take the former. With investing for yourself you diversify to reduce the chance that you lose money, but I don't think that's a goal with charity.\n
\n
\nFormally, I don't think we should be risk averse in our donations.\n
\n
\n(If you give to a few organizations that you think are all about the same level of effectiveness, that's not actually much of a problem. If I managed to convince you to just give to the one charity you thought was best, we'd end up with only slightly better outcomes on average.)\n
\n
\n(Having multiple charities running is good, in that they compete with each other and try different things. If you need to give to multiple charities to make that happen, you should (though I think as \n@Julia\n points out you probably don't have to because there are others out there giving too.) But this isn't diversification in the risk reducing sense as much as a kind of research.)", "timestamp": 1330834188}]}