Hacker News new | past | comments | ask | show | jobs | submit login
I mostly believe in Worms (givewell.org)
71 points by apsec112 on Dec 23, 2016 | hide | past | favorite | 32 comments



I like that the GiveWell website has an "Our Mistakes" section right up in their main navigation bar. It does a lot to inspire a feeling of transparency, which is a big issue for non-profits. I think more of them would do well to take the issue of transparency more seriously, and if they are reluctant to be transparent, they should take some time to consider whether it is for truly good reasons.


I think GiveWell does a great job at determining where we get the best bang for our buck when it comes to aid, but at a more fundamental level, I don't think I'm in a position to tell someone else what they need. That's condescending and I'll never have the whole picture. That's why I give to GiveDirectly--I think people should be able to determine how they are helped.

Deworming programs might help kids the most, but then that money goes to a (likely American) pharmaceutical company and the benefit is over.

In contrast, give someone money and a lot of that money will stay in their community. Maybe the iron roof they buy provides them a smaller benefit than deworming, but the guy who makes and installs the iron roof can now buy a cow from his neighbor, who can now buy seed for next year's crops. Maybe eventually the money allows someone to get deworming medication for their kids, but now that money had benefited a bunch of people in the community first.

It's hard to collect data on this kind of stuff (although GiveDirectly is collecting a lot of data). But I think we need to recognize that the only reason we collect data is to control outcomes, and we're not really helping people if we don't allow them control over their own lives.


If we're doing fair comparisons, the gift to a single family that allowed them to buy the iron roof could have dewormed 200-1000 children and at least some of those funds pass into the hands of local administrators spending in the local economy. I also doubt the profit margin for an East African roof installer buys him a cow, and it may not even be as much as the local thatcher stands to lose from no longer being required to undertake as many biennal rethatching projects. (Estimating all the side effects of a programme is really quite tricky)

I don't buy the argument that deworming doesn't really help people because they weren't offered the choice of buying the equivalent value of cigarettes instead. Just because a person freely chooses to do something doesn't mean it helps them, and vice versa. Donors and aid agencies' systematic biases in favour of things that don't work particularly well (this, not "control" is the real reason for all the data collection attempts) is certainly worth worrying about, but that can't be fairly assessed without realising that recipients also have systematic biases in favour of things that don't work or against things that do too. There's a reason why countries where the average person has a lot of disposable income spend even more as a proportion of income on publicly subsidised healthcare programmes.

Edit: This shouldn't be interpreted as a particular criticism of GiveDirectly, if the idea of making a fairly big difference to a small number of randomly-selected very poor Kenyan people appeals to you.


The mention of cigarette spending is an extremely dangerous strawman, largely because for some reason people actually believe it. There's fairly strong evidence that there is no increase in tobacco or alcohol spending[1][2] with direct cash transfers to the extremely poor. These stereotypes are frequently used to argue against charity or welfare to the needy, and I can only consider this line of argument to be harmful.

This is not to take away from the rest of your point, but I would appreciate it if you chose a different comparison next time.

[1] https://www.givedirectly.org/research-at-give-directly [2] https://www.givedirectly.org/research-on-cash-transfers


Appreciate the sentiment, but it was actually a quite deliberate comparison because (i) the very poor in developing countries where cigarettes are very cheap often have a cigarette budget which comes before investment and improved nutrition (ii) I'm well aware that despite the startling prevalence of smoking amongst poor men, increased tobacco use isn't a significant factor in studies of how poor people spend donor cash.

(i) pretty much underlines my point about the things people choose to spend their money on not necessarily being what's best for them. And in the short term it's not an inexplicable phenomenon because cheap tobacco probably does provide more pain and stress relief than slightly better nutrition.

(ii) is partly because surveys like those you've posted accurately represent people tending not alter their habits of smoking cheap cigarettes (or frowning on tobacco) after receiving a sudden windfall; there's a limit to how much one can smoke and stockpiling makes no sense when for people who don't even have a waterproof roof. But it's also because poor people are smart enough to realise that people quizzing them on their lifestyle and unexpected windfalls from foreign donors aren't unrelated, and the natural response is to tailor responses to what they think the foreign donors want to hear, resulting in any increases in spending on cigarettes and alcohol (whilst likely low relative to increases in spending on roofs or microbusiness investments anyway) being systematically underreported.


Do you have any evidence for systematic underreporting? One could just as easily say that the members of control groups (who do not receive cash) would have a reason to underreport their spending on things that are not approved of in order to receive more potential money in the future.

Also the research presented by GiveDirectly also includes studies of people where the cash is not from the same people as the surveys (for example state funded benefit programs that have clear and universal criteria unrelated to spending or consumption.) In those cases again there is reason to expect the underreporting to be similar across both groups that get cash and those that do not currently.


> One could just as easily say that the members of control groups (who do not receive cash) would have a reason to underreport their spending on things that are not approved of in order to receive more potential money in the future

This apparently happening with control groups in the same villages as cash transfer recipients is actually some of the most persuasive evidence response biases exist in the surveys. According to various RCTs, when their neighbours start getting money, women are less likely to report their husbands beating them. On the other hand, they're also less likely to report being satisfied with their life. There are [just about] plausible mechanisms for how spillover effects from donations could cause those behavioural/satisfaction changes amongst non-recipients in a neighbourhood, but recipients thought they were less likely to receive money if they reported domestic violence or said they were satisfied with their life is the most Ockham's-razor satisfying one. (It's also likely part of the reason why recipients seem surprisingly unlikely to report increased satisfaction from life after receiving life-changing sums despite positive changes in almost every measurable area.)

Response biases are difficult enough to eliminate in social science surveys when participants have no reason to suspect a sum greater than their average annual income may hinge on their answers.


> If we're doing fair comparisons, the gift to a single family that allowed them to buy the iron roof could have dewormed 200-1000 child

It's worth noting, though, that deworming 200-1000 children doesn't mean deworming 200-100 children who have worms. It means administering deworming medication to 200-1000 children, a few of whom will benefit (because they had worms) and the majority will have no effect (because they didn't). IIRC, programs like Deworm the World don't target infected children, because figuring out who to treat is more expensive than just treating everyone, if you're in an area with a high infection rate.

Overall, it might still be more effective than GiveDirectly, but it's not as simple as "this would directly benefit 200-1000 children rather than one family". (I'm not saying that's what you were saying, but that's how I would interpret it if I weren't familiar with the programs.)


Soil-transmitted helminths infect 2 billion people. http://www.who.int/mediacentre/factsheets/fs366/en/

More than a few will benefit, because in many areas with high infection rate, the majority of children really do have worms. Regarding Ascaris infection specifically: "In areas of poor sanitation virtually 100% of the population harbor the parasite."[Markell & Voge's Medical Parasitology p.241]


> I don't buy the argument that deworming doesn't really help people because they weren't offered the choice of buying the equivalent value of cigarettes instead.

Do you understand how condescending it is to think that you know so much better than people what they need that you think they'll buy cigarettes I'd you give them money?

Putting aside the fact that this isn't what people given money do in studies, your viewpoint is an extraordinarily arrogant one.


Do you understand how extraordinarily arrogant your white knighting is when I've already pointed out elsewhere in the thread that (i) the prevalence of smoking amongst poor males in most developing countries unambiguously is high irrespective of donations and even given a reason why low-cost tobacco might appear preferable in a tobacco/nutrition tradeoff and (ii) acknowledged that GiveDirectly-sized donations (as opposed to the one-pack-of- twenty-Rockets cost of a deworming tablet in the tradeoff comparison I actually made) tend to be invested in useful capital goods but explained why studies associated with donations likely systematically underreport tobacco consumption and other vices relative to studies not associated with donations anyway. And even edited my original post to explicitly state that I wasn't arguing that GiveDirectly was a bad idea.

I think it's probably more condescending, and certainly more harmful, to assume that extreme poverty bestows upon the global poor some mystical ability to make optimal spending decisions that for some reason doesn't apply to comparatively wealthy people in developed countries entitled to access heavily-funded collective healthcare and social insurance systems.


> I think it's probably more condescending, and certainly more harmful, to assume that extreme poverty bestows upon the global poor some mystical ability to make optimal spending decisions that for some reason doesn't apply to comparatively wealthy people in developed countries entitled to access heavily-funded collective healthcare and social insurance systems.

That's not what I said.

Fundamentally, I'm saying that personal choice in itself is a value, a value which is being overlooked by most aid programs.

Low income males spend a disproportionate amount of their money on cigarettes, but that is different from a statement that low income people will spend aid money on cigarettes.


GiveDirectly is one of the charities recommended by GiveWell (in fact, it's their top charity by amount of money moved [1]).

[1] http://www.givewell.org/about/impact


Some problems like smallpox can be permanently fixed with targeted aid. In those cases the benefits to local economies can be huge and long lasting.

I feel a mix of the highest ROI directed aid along side programs like give directly, is probably ideal.


Not to be pedantic, but ROI as a concept only makes sense when you're trying to achieve your own goals. If you're really empowering other people, then they are the ones who should be evaluating the ROI on their decisions, not you.

Ultimately, I do agree that a mix of giver-goal aid and receiver-goal aid is useful. I think it's a very laudable goal to i.e. end polio. But I think that western aid has very little concept of helping people achieve their own goals. As a result, we're very far unbalanced in favor of the goals of those giving aid rather than the goals of those receiving aid. Until that balance is fixed, I think there's far more need in receiver-goal programs.

Incidentally, besides GiveDirectly, I give to Planned Parenthood and the FSF. While they both have slightly more complex models, I think both of those organizations are geared toward empowering people to achieve their own goals.


> ROI as a concept only makes sense when you're trying to achieve your own goals. If you're really empowering other people, then they are the ones who should be evaluating the ROI on their decisions, not you.

While a noble concept in theory, people are notoriously both selfish and short-sighted. Especially when basic infrastructure is missing, and assuming the giver-goals are unbiased, I would expect that an observer would typically say that the society helped by giver-goal aid is better off than the society helped by receiver-goal aid.

Receivers, being selfish and shortsighted, will be unlikely to invest in things like a local hospital, or sewage, water, and electricity. They'll buy themselves a trip to a hospital to get taken care of, and purchase a private well, a septic tank (or just a longer pipe to the local canal) and a generator. If they get a propane tank, propane delivery service, and a modern furnace and stove, that's great for them, but the observer would probably say it would be better to run natural gas to all the houses rather than individual propane tanks to a small number of residents. And you may have better results in the short term by giving people money to buy the generator they want rather than a solar array that puts out fewer watts per dollar, but the city will eventually be covered in smog. Or they might invest in better tires for their vehicle to navigate the potholed roads, but they won't pay to pave or maintain them. And while individual smartphones are a wonderful luxury, many more people would benefit if you put PCs in the local library.

Perhaps these things would have eventually sorted themselves out when the empowered receivers start using their elevated position on Maslow's hierarchy of needs to become wealthier. They would then contribute more to the tax base and eventually an effective government will have the funds to start investing in these public works that eventually people realize they want for a few percent of their wealth. But I think it's more effective to skip that "perhaps" stage.


Except the exact same criticism applies to the alternative: everyone involved in allocating and spending the money you give to build and supply that natural gas pipe is also selfish and short-sighted, and is much more likely to try to enrich themselves over providing a good service.

At least by giving directly, it's easier to control that the money actually gets to the people it's supposed to, even if they don't spend exactly as we might wish they did.


> everyone involved in allocating and spending the money you give to build and supply that natural gas pipe is also selfish and short-sighted, and is much more likely to try to enrich themselves over providing a good service.

The point where the money is allocated to one goal or another is when someone donates the money altruistically. They are hopefully trying to be less selfish and short-sighted.

Sure, the implementors will take some profit, but so will any business to which the recipient could give the money.


Keeping the power of determining where the money goes is in itself selfish, so I don't buy the argument.

More to the point, why are we optimizing for altruism and forward thinking? Are those end goals, or merely means to an end goal?


You start off by claiming that people are notoriously selfish and short-sighted, but mysteriously the rest of your post only applies this to the receivers of aid. The givers are people too.

More fundamentally, you're only looking at material outcomes. What I'm saying is that freedom (a non-material ouycome) is a good in itself, one which is under-considered in aid.


People give Aid so things the giver wants to happen happen. As such, you can talk about ROI in terms of achieving the givers goals.


That's what I said.


Worms and other parasites are known to influence behavior. If you give money to someone whose decisions are being influenced by a parasite, are you really giving the money to them, or are you giving it to the parasite?


> Maybe the iron roof they buy provides them a smaller benefit than deworming, but the guy who makes and installs the iron roof can now buy a cow from his neighbor, who can now buy seed for next year's crops. Maybe eventually the money allows someone to get deworming medication for their kids, but now that money had benefited a bunch of people in the community first.

Then just directly give money to the guy who installs the iron roof and guy who buys seeds. This is a poor way of giving to those three people (it is hard to control the amount of money the three receive).


> it is hard to control the amount of money the three receive

Since it's very hard to know how much that should be, we just inject some cash into the bottom of the market, and let the market distribute the cash.

The debate about giving cash vs. medicine is that cash moves around the target community and stimulates lots of things, rather than going direct to a multinational pharma corp's Swiss bank. The risk is that too much of the cash is absorbed by bad actors.

As usual, it's a complex trade off and maybe a bit of both would give best results.


The market does not have charity in mind; it will allocate capital to maximize the total size of the economy.


That's completely orthogonal to what I said. The local market moves money around locally, where it can be useful.


I like Kiva. You can fund loans to businesses and individuals that would normally not have access to capitol. The agencies that administer the loans charge some interest to the borrower that is used to cover the costs of the loan as well as do charitable work. They have a less than 1% default rate, though there is some risk of loss during currency conversion back to dollars.


I lend a lot on Kiva, in part because it's more fun, and in part because I dumped a lot of money in there prior to getting into EA, but the evidence of effectiveness for microfinance seems much more limited/tenuous than Give Directly, Deworm the World, or AMF.

For what it's worth, though, One Acre Fund (one of the Kiva field partners) is a recommended charity on The Life You Can Save (another EA charity evaluator)[0]. Since discovering that, I've been focusing my loans towards that specific partner. They currently have a lot of loans which are nearing expiration, if you're interested: https://www.kiva.org/lend?partner=202&sortBy=expiringSoon

[0] https://www.thelifeyoucansave.org/Where-to-Donate/One-Acre-F...


What I love about this analysis how it tries to deal with conflicting information. The author is very explicit with the problems of the Worms study, but still trusts it. This is real life decisonmaking.

There are few easy answers, especially in complex systems. The best we can do is put the evidence out there and make our best guess.


It reminds me of the time we had too many homo saps and cattus rattuses in the house and we got tired of seeing worms on the carpets and gave all the mammals a round of Pyrantel.

CDC estimates (not alt-health loonies) that 10% of people in the U.S. have some worms.


Where are you getting that number from? I can find stats on individual types of parasites on the CDC site, but not worms in general.

I am seeing 13.9% with Toxocara antibodies[0], so past exposure. That's fun.

[0] https://www.cdc.gov/parasites/toxocariasis/gen_info/faqs.htm...




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: