I think that this sort of thing actually validates the fine grained, very numerical and technocratic approach of things like EA. Handing out malaria nets does not need a narrative, while boilerplate critiques of EA usually heavily rely on very unreasonable assumptions typical of left wing academia. Whenever people invoke “politics” as an unalloyed good, they mean a very particular type of it, and that type has done the developing world no favours.
In other words, EA should it keep it simple and recognize its limits, because getting anthropologists involved is a pretty risky bet.
Nice book review! As a recovering Development Economist, I have my own thoughts on the problem with the discipline. My main gripe is the over-reliance on game theory. In grad school it was made painfully clear that to be taken seriously, your paper had to have a model. I just never understood why applied mathematicians noodling around in their offices in Europe and America would have all that much of value about what was actually happening in any given development context. I would much rather have Anthropologists and Sociologists coming up with theories to test via qualitative research. My gripe with some qualitative researchers is the belief that qualitative results are evidence, when in fact they're theory. Why can't we marry the two? Qualitative research for theory, statistics for evidence.
This nods to my prejudices so much that I'm not sure I can be fair in my assessment of it. Having said that, it does seem very well-written. It is also extremely restrained, which I particularly admire as if I'd written something similar I would be seething with anger.
My general reaction to seeing the same things that Ferguson writes about is also the same as his - antipathy to developmental NGO's, charities and EA's who mostly seem to be motivated by self-calming and the amelioration of guilt.
But... a really good review with a great deal of substance.
Should I be seeing footnotes in the text, or did substack eat them?
This is the best review I've read this year.
Very cool, thanks.
I'm interested in your claim that "unconditional cash transfers can hurt neighbors who didn’t receive the cash." You give a link to a rather dense seventy page academic paper. Where in this paper is there support for your claim?
> I think one of the highest impact investments an Effective Altruist fund could make right now would be to hire a handful of trained anthropologists [...] to really study how effective altruism works as a system. [...] What more are we missing?
EA is not actually interested in criticism. They say they are but I've repeatedly been attacked for offering it. I've proposed (not here) various projects to look at things from a structural perspective and I've never gotten funding. Nor have at least two other people I know. I suppose it might partly be my fault because I expect to be paid to do so. But the reality is that if you're not going to pay researchers you're going to end up with outside researchers producing product for whoever pays them (see the NYT article) or from their own point of view.
People are telling you what you're missing. They're even offering to write thorough reports on it. They're just not getting funding. And the truth is the community values where it's willing to put its dollars. I've seen a bunch of EA stuff say they want criticism. I've never seen them fund it though.
I get it's hard, as a movement, to seriously take criticism onboard or actually accept challenges to your assumptions. And I get that most movements have myths they tell about themselves. But if you don't have a clear view of yourself you can't actually be effective. But as I've said a few times the EA movement is not actually about being effective. That's just its mythology, what the movement tells itself.
> For any sort of context-specific intervention to work, an intimate knowledge of the specific history, needs, and geography of individual villages and regions is necessary.
I wouldn't say this. I'd say the level of confidence needs to match the level of force. If you want to go to a country and offer voluntary transactions your level of knowledge can be pretty low because the other side's consent acts as a safeguard. They might say no. But that's a relatively low cost way to take rejection and you can then move on.
If you're using government force or if you're doing something against the villager's will you need to be absolutely, positively sure that it's in their interest. And you can imagine a world where it is. For example, if a village doesn't understand modern medicine but suffers from guinea worm. But you have to be that level of sure. (And even there you can meet things like local doctors who make their income curing guinea worm and don't want it cured permanently.) But most of the time people know their default interests.
By the way, there's been some successful work in Lesotho in setting up manufacturing. It works because they already have a culture of leaving home for employment and they prefer factories to more dangerous mines. But of course force wasn't used and the companies are turning a profit. Still, I'd say they did more good for Lesotho than this project. (Which gets back to my EA vs SE synthesis.)
> Ferguson’s work exposes a possibility I hadn’t thought of before, in which “technical” and “apolitical” projects can expand the power of the state in unforeseen and potentially dangerous ways.
This is a bog standard anarchist critique. I really want to be kinder than this but it seems like Rationalism/EA devalues things like philosophy and political theory and then discovers, oops, there's actual value in these fields! This is why I've warned multiple people off doing exactly this kind of work. I've said something like, "Yeah, they say they want that, but they don't. You won't get funded." And the ones who ignore my advice and applied anyway haven't gotten funded. EA/Rationalism just don't value that kind of thing.
So it goes. It's your movement. I'm not even sure it's a good idea to be open to criticism. The most successful movements often aren't. And you'd definitely need some kind of immune system to keep out the hacks. (And perhaps, even in the best case, I'd be kept out by it.) But regardless it's just not where things are now.
“I think one of the highest impact investments an Effective Altruist fund could make right now would be to hire a handful of trained anthropologists (or other outside experts in qualitative research / ethnography) to hang out in places like GiveWell or the Machine Intelligence Research Institute for a few years and really study how effective altruism works as a system.”
I’d suggest a neo-classical micro economist. Not because I agree wholeheartedly with all the neo-classical approach, but because it would be well-suited to anticipating where the problems might lie.
This is an extremely outdated, anecdotal take on development economics. Which, come to think of it, describes anthropology quite well, in many different ways. Not the highest quality theory or evidence.
Current models of development economics, (actually even these models are by now at least 20 years old) which have evolved by learning from failed attempts at development (and successful ones), recognise very clearly the fundamental importance of political and legal institutions, as also the importance of markets.
"There’s been a little bit of research on the quantitative side --Recent research has found, for instance, that GiveDirectly’s 2014 unconditional cash transfer trial increased community participation but did not change voting patterns, so at least in 2014 the Kenyan government wasn’t using the program to stay in power."
This on its own is an idiotic line of reasoning, that I see far too often. Observing that after intervention x, variable y did not change and blithely stating that therefore x had no effect towards y. In this case the Kenyan government could have used the program to buy itself goodwill, which hadn't they done that, would have led to an even greater loss of their power than the unchanged voting patterns would have already led to. What's the offical term for this fallacy? Birdbrain monocausal-constancy assumption mania?
When you included "Many families had large flocks of underfed cattle. Even when money was tight, the team rarely observed cattle sales.", my instinctive reaction was "Well, yeah, cattle are wealth!" but that's because I'm from a rural background with a strong tradition going back centuries that cattle represent wealth:
So my initial takeaway from this is that the lessons EA can learn are:
(1) Fewer technocrats, more rednecks on your staff, and by "rednecks" I mean "people who know what land supports cows and what land supports goats and why you might raise one rather than the other"
(2) Work with the locals - and this does *not* mean the Minister for Embezzlement of Foreign Do-Gooder Funds sent to meet you and have a nice chat about developing that project, it means the guy in the village who has lived there all his life
(3) The problems you think are the big issues may not be the problems the locals think are big issues: why are you trying to breed ponies in cattle-raising country?
(4) The real problems may be something different from what you or the locals think
I appreciated this review, thanks!
The beginning of the final section left me with a criticism -
"So what do we do as effective altruists........"
I'm not an EA in any shape or form. For reasons similar to those explored in Ferguson's book, I have a negative attitude towards most effective altruism. So it grated a bit for the reviewer to talk as if "We" are all EA's.
I know Scott is more than a little sympathetic and yes, it's his blog. But he's also a Dem voter generally, and it would be really weird to have a reviewer of a book to say "So what do we do as progressives..."
I would rather the author had made an effort to put aside his or her EA prior beliefs and also not to assume that all of his/her readers shared them.
I still really liked the review.
The problem is that the trained anthropologists would bring their own host of biases to the project, in particular that they tend to have an instinctual aversion to Western NGOs getting involved in the "Global South".
I was re-reading Samzdat's 'Banish Plump Mouse Deer and Banish All the World' just a few days ago.
1. Funny how James C. Scott keeps popping up.
2. One of the handful of Samzdat pieces that references SSC directly.
3. Extremely apropos to this (which I really enjoyed and will probably vote for).
Fantastic book review/essay.
There's a "sunnier essay" about the quant/qual combo platter approach. About schools in Kenya. "Beyond Moneyball" in the journal Education Next.
>> ”…really study how effective altruism works as a system. How are decisions being made, and how is evidence being used to make them? What does “EA discourse” help make visible and which problems and concerns does it hide from our view?”<<
This is truth-seeking activity that in practice today would likely receive major push back from NGOs and their stakeholders, answering such questions emphasizes the fundamental problems of charitable aid approaches (perhaps why Ferguson gave up the idea altogether?).
Effectively the same criticism can be made of other elements of the liberal state approach to Africa, such as governance reform, defence and police reform, justice reform, elections and so forth. After thirty years of involvement with these sorts of issues on the ground, I have decided that the best thing the West could do is to go away. On balance, and for all that there are individual successes, I think we do a lot more harm than good. It's not that countries can't benefit from foreign expertise (Japan and Korea both did, for example) it's that the recipient state has to be an expert customer, able to pick and choose what makes sense and is useful. (Ha-Jun Chang has written about this at length). But development aid in the widest sense ultimately comes to dominate the country. Western-funded NGOs pay much better and so attract able people out of government. And local governments eventually wind up having policies on, say, police and justice which are just the sum total of projects that various donors are prepared to fund. And as the review points out, horizons are very short: the average EU project is expected to show results in two years, because the average length of tour in a job is three years, so the incumbent wants to start something and see it through to the end. Thus, African governments tend to be presented with much the same projects every five years or so, because donors have no institutional memory of what was offered before. And of course donors themselves are in competition with each other.
Maybe we should just go away and let them sort it out.
So if development programs don't work, what does? Not a single African country has risen from developing to developed status. In fact, I think the only non-European of any size that went from very poor to developed in the 20th century was South Korea. Maybe economic success is just a fiendishly hard problem that's beyond our current capabilities to understand. It might be true that the World Bank missed many important aspects of the problem in Lesotho, but society is so complex and multifaceted, and the motivations of its millions of actors so opaque (often even to themselves), that *any* investigation is bound to be superficial.
You could explain some of the lack of development by saying that governments are run by thugs who care about power and status above economic development, but is it plausible that every government in every African country is like that? Especially because economic success is a good way to purchase legitimacy, status, *and* state power, as China demonstrates? And what about the citizens--do they want to live in squalor forever, or are they also exhausting their mental and physical energies to improve their financial situation? When billions of people working day and night on a problem haven't managed to solve it, we should admit that it's a hard problem, and avoid blaming the World Bank, Effective Altruists, or whoever else for also failing to solve it.
It ain't just economics, and it ain't new, yo. "Body Rituals Among the Nacirema" is a hoot and a classic, going back to the 1950s.
In addition, I have seen Ghanians, Kenyans, etc. re-write western MSM headlines about western countries to parody the condescending tone of western MSM articles about their own countries. These are also quite funny, talking about the hereditary rulers of the British Isles and the quaint customs of the flavor-deprived natives there, for instance.
> Are 401(k)s merely a “product” that young Americans hold onto because they’re too hard to sell?
This *definitely is* the self-concept of the 401(k) program; that's why there are so many legal barriers to liquidating your 401(k).
>As funding gaps for “low hanging fruit” like malaria disappear, EA is going to have to focus on more complicated interventions
A few years ago I read a story claiming that the biggest problem with mosquito nets is getting people to actually use them. Has this problem been solved? If so, how was it solved?
Very good review, I greatly enjoyed it.
I very much like this essay, though I think it is more successful as an essay than as a book review.
As a big fan of webcomics, I was delighted to see a webcomic link in there. Until I clicked it!
That is one of the worst made comics I have ever seen in my life. It is not hyperbole in the slightest to say that I have seen 10 year olds make better comics than this. Why? Why?!
I mean comics are hard work to make, even bad ones! Why go to all the effort to turn a thesis into a comic if you can't draw (at all!) and don't even know how to make a word balloon? There is no added benefit to this being a comic. It would make a perfectly fine prose piece. As a comic it is offensive to the eyes and makes the content more difficult to comprehend. And it's so long! It must have taken so long to make, and for a result worse than doing nothing at all!
I can't even respect it as a passion project, because for all the extra work the creator must have done they still did it in as lazy and slipshod a fashion as possible! He obviously would draw one (awful) picture, and then crop it in multiple different ways to make different panels so he didn't have to bother drawing the same people in the same scene multiple times. But since its all from the same still image there is no sense of motion, movement, or the passage of time. So why make it a comic at all?! The whole benefit of the comic medium is to be a static and visual artform that creates the illusion of movement and sound. If you're not going to bother creating an illusion of movement or the passage of time then just write an essay and throw in your terrible illustrations every other page!
It reminds me of when I took a business class and the professor had advertised it as being creative and fun because the textbook was a novel that she had written herself, about a woman who inherits a bookstore. That sounded good until I started reading and it turned out to be the most unartful 12-year-old-fanfic piece of dreck that I'd ever read. By the second chapter I was longing for a real textbook: at least they're not pretending to be interesting!
Intervention requires legibility to decide how to intervene. But that legibility is created in the mind of the interveners; it is not a reality in the lives of the targets of "aid," unless – as all too often happens – it is imposed.
Did the country of Lesotho every say, "Hey, we're all screwed up. Can you smart guys come over here, figure out what's going on, and help us?"
When England began to undergo the industrial revolution in the late 18th century, was that a result of some foreign country coming from outside and giving it development aid?
The root problem is that of altruism itself: the idea that those unfortunate people there should be "helped" by us smart people here. If you start with that premise, measures of "effectiveness" will never measure the right thing.
If you want qualitative problems with MIRI here's a start: The out of hand dismissal of deep learning is based on no data whatsoever. The pursuing of a completely unproven alternative plan of research on the premise that it may overtake deep learning is also based on no evidence whatsoever. Their acting important isn't supported by any past accomplishments.
Time for a spot of relevant poetry....
The Development Set
by Ross Coggins
Excuse me, friends, I must catch my jet
I’m off to join the Development Set;
My bags are packed, and I’ve had all my shots
I have traveller’s checks and pills for the trots!
The Development Set is bright and noble
Our thoughts are deep and our vision global;
Although we move with the better classes
Our thoughts are always with the masses.
In Sheraton Hotels in scattered nations
We damn multi-national corporations;
injustice seems easy to protest
In such seething hotbeds of social rest.
We discuss malnutrition over steaks
And plan hunger talks during coffee breaks.
Whether Asian floods or African drought,
We face each issue with open mouth.
We bring in consultants whose circumlocution
Raises difficulties for every solution –
Thus guaranteeing continued good eating
By showing the need for another meeting.
The language of the Development Set
Stretches the English alphabet;
We use swell words like “epigenetic”
“Micro”, “macro”, and “logarithmetic”
It pleasures us to be esoteric –
It’s so intellectually atmospheric!
And although establishments may be unmoved,
Our vocabularies are much improved.
When the talk gets deep and you’re feeling numb,
You can keep your shame to a minimum:
To show that you, too, are intelligent
Smugly ask, “Is it really development?”
Or say, “That’s fine in practice, but don’t you see:
It doesn’t work out in theory!”
A few may find this incomprehensible,
But most will admire you as deep and sensible.
Development set homes are extremely chic,
Full of carvings, curios, and draped with batik.
Eye-level photographs subtly assure
That your host is at home with the great and the poor.
Enough of these verses – on with the mission!
Our task is as broad as the human condition!
Just pray god the biblical promise is true:
The poor ye shall always have with you.
Shouldn't the main takeaway to all this be, "stay out of developing countries, and especially Africa" ? The prospect of boldly going into this benighted land to bring the light of civilization sure *sounds* good, but in practice the sociopolitical landscape is impossible to navigate (for a non-native, at least); and therefore any action you can effect at all is likely to cause harm.
Does Ferguson portray himself as the first to appreciate these political-contextual issues? Or did the review just want to save space by not digging into the history?
Ferguson's thesis is a variant on a long-established critique of development aid. As an example, here is a summary of parts of my notes from the first relevant book-length exposition I know of: Hanan S. Aynor's Notes on Africa, published all the way back in 1969.
Aynor was an Israeli diplomat. His book mixes memoir, anthropology, and policy recommendation. He based it on his observations in Congo-Kinshasa, Senegal, and Gambia; he was joint ambassador to the latter two.
His memoir part is lightly fictionalized, because diplomats must be courteous, so we can't verify what happened where. (The Foreign Ministry later made him ambassador to a few more important African countries.) But he doesn't seem like a Ryszard Kapuściński. In contrast with the stories in Kapuściński's The Shadow of the Sun, for instance, I think it's likely that these interactions happened largely as told.
Aynor describes development specialists, sent to Africa to improve something, quickly growing deeply impatient with diplomacy and political context.
Sometimes the African government formally approves a project, then refuses to implement it. Without explanation. Frustrating, right?
Aynor interprets this as a senior politician blocking the project because it somehow intrudes on his turf. As a diplomat, he knows he is not allowed to press for action, or even ask for details.
But it's hard for development specialists to accept that, for undisclosed reasons, sometimes they just can't implement what they were sent to implement.
On other occasions, implementation goes forward, but something unexpected happens.
Just one example, from a description of how a village irrigation project stalled:
“The irrigation system consisted of a pumping station and an intricate network of earth canals channeling the water into the fields. If not constantly watched and repaired, water breaks out easily, flooding fields and destroying large sections of the canal system - which has to be rebuilt patiently and quickly after each irrigation. During the hot season, this system demands night shifts, and it generally imposes a radical change in working attitudes. Tensions developed among individuals and family groups in which the African government technicians assigned to the project took an active part. Accusations and counteraccusations reached the danger point of physical violence. During these weeks, irrigation ceased entirely. Then the quarrelsome technicians departed, and the settlers distributed the collective fields among themselves and each started to exploit his own plot. Meanwhile, the pumping station had stopped working and needed to be repaired. The farmers seemed relieved by the disappearance of the feud and the return to the old, well-known order. It was a situation they understood and could handle themselves. They were cheerful and optimistic.” (114)
Aynor draws these lessons:
First, motivate villagers as a whole. Do not just appeal to individual farmers' profit motive.
Second, let villagers critique the project before it is implemented. Get confirmation that each aspect of it has utility to them. Ensure that they think the demands (i.e. upkeep) are worth it.
I do recall a whole chapter of specific development advice at the end, but I didn't take many notes on it, due to how long ago it was written.
I thought it was an interesting review, but I’m not sure it actually succeeded at one of the things I expect a review to do, namely inform me on whether I should read the book.
There are footnotes at the bottom of the piece, but they seem to have been removed from the body. How do we know which sentences correspond to which footnotes?
This case study happened in 1975, so I have to wonder how much things have changed since then. Like, there were a lot of fields of study doing things we'd consider irresponsible in the 70s - that's the same decade as the Stanford Prison Experiment.
I read Paul Collier as a contrast with William Easterly, finding the latter more persuasive (checking a decade+ old blog post, I see that the former didn't even use footnotes). But it should be noted that Collier himself has displayed some awareness of how governments act in a way that is self-interested rather on behalf of their constituents. In "The Bottom Billion" he references the "political business cycle" in which governments try to stimulate aggregate demand in the run-up to an election.
Have any of you read <a href="https://www.amazon.com/Shock-Old-Technology-Global-History/dp/0199832617">The Shock of the Old: Technology and Global History since 1900</a>?
The author discusses some innovations that have had tremendous influence long past their heyday. One of the “low tech” innovations he discusses is corrugated iron.
If you are a poor African/Brazilian living in materially poor conditions, a corrugated iron roof is probably better than anything else you could build with.
There’s another book, it might be called “AK-47” or “Kalashnikov.” It is a horrible fact that the AK-47 is an extremely popular Western (by a loose-ish definition) technology in the third world. I couldn’t find the book in a quick perusal of Amazon.
If you look at all the things we’ve tried to “give” people in the third world that never “took” despite considerable effort, like solar cooking, maybe we can draw some lessons from tech that’s been adopted wholeheartedly? If you compare the AK-47 to any other weapon, the AK was
1) Reliable 2) Simple to use 3) filled a need that the Africans (or Pakistanis or whomever) felt, as opposed to fixing a “problem” that a few 22-year-old Americans wanted to “fix.”
I read somewhere that there was an African village where the peace core dug a well so the women did not have to walk long distances to the river. Labor-saving innovation adopted, right? Nope! The women (and children?)-only walks to the river were part of their culture. They could talk freely without men around.
One thing we don’t really “get” is how atomized Westerners are compared to people who live more traditional lives. People who know everyone they see daily. Things become part of the sea in which they swim. Ok, this is dated, but consider the smoke break of yesteryear. Do you think workers would have welcomed nicotine patches so they could get their fix and work at the same time?
I assume some of you have read <a href="https://www.amazon.com/Design-Other-90-Cynthia-Smith/dp/0910503974/ref=sr_1_1?keywords=Design+for+the+other+90%25&qid=1653701107&s=books&sr=1-1">Design for the Other 90%</a> I think the problems designers have creating products for the third world are the physical, material equivalent of more abstract developmental economics.
Lastly, I’ll end on an unhappy note. There are some oil-soaked exceptions, but only European, European-derived, and East Asian countries have developed in the whole history of the world. There was a book, IQ and the Wealth of Nations, that got half-assed IQ averages for a bunch of countries. IQ and GDP correlate really well. There are rich nations in the Middle East that got rich from resources and guest workers exploiting them, but the natives did not end up smarter, so we know that intelligence caused wealth, not the other way around.
People have criticized Lynn (the author) for not engaging in the developmental economics literature, but did he really need to? Knowing that intelligence is a huge input into development is like knowing that heat is related to molecular motion. If you discovered that, you don’t need to deeply engage in the literature on phlogiston. You just need to show that your model works better with fewer inputs.
Given the phenotypic (not getting into nature and nurture) IQ differences, is there much room for Western charities to bend the curve? I’m not arguing that development is impossible because African Americans have higher material standards of living than Africans, but I think it’d be really hard for anyone from the outside to predict what will work.
Even the most obvious, vaccination, because everyone with kids wants them to be healthy, and illness is fairly objective, but it did not fit with the cultural desires of actual Africans. Now, you have 8 kids, and only two die instead of six, and you are a high-status, good mother/father. But everyone had more kids survive, and now Africa’s population is booming, but their economies mostly are not. Unstoppable force, meet immovable object.
Maybe Star Trek had it right? Non-interference is the only ethical path?
"Most households grow fruits or vegetables in their yards"
That doesn't fit my recollection of the suburbs.
“The Anti-Politics Machine” is quite the poorly chosen name, almost seeming to mean its opposite. Should have been “The Political Laundromat” or something.
There are more relevant comments to make than this one, but I think it had to be said!
I wonder if apart of this problem, imo, is that some of ppl most associated with advocating local knowledge, qualatative understanding and playing up risk of unintended consequences are ideologs. It's really easy to ignore criticism when it sounds exactly like the things that ppl who are ideologically hostile to the idea of quantatively informed methods, using equations to weigh impact and want to just tell a story about the value of respecting everyone's culture and the harms of western imposition.
Hell, I can feel that emotional response and desire to just reject Ferguson's arguments as just more hippie-dippy cultural bullshit. That persists even tho I'm sure that even if the claims are exaggerated or blame improperly assigned this kind of failure is still an important issue.
Unfortunately, if not these attitudes, some of the way of speaking that's associated with them is widespread in areas like cultural anthropology. Maybe the right anthropologist in MIRI would be helpful but if you just picked a random culturally anthropologist my prediction is that it would make things worse as the MIRI ppl would end up feeling they were being condescended to in much the way some mainstream media articles on rationality make rationalists feel and they might then be more resistant to those ideas.
I'd add that this review feels a bit too credulous (especially given the lack of non-Fergusun cites) but, having said that, I do feel that another issue is the incentives in academia.
If it's not the kind of argument that the economist in your seminar can evaluate/disagree with it's not the kind of thing that will get you a job.
I feel like this critique of development economics is pretty outdated with the rise of RCTs. A really beneficial side effect of the RCT revolution is that it has made Western academics spend extensive amounts of time in the countries and cities/villages they study, because they have to meet stakeholders/do fieldwork in order to actually _implement their studies_. It used to be possible to just download official statistics, run some regressions and claim you have understood everything. Now, it's very common for development economists to spend months in the field because correctly understanding the field context is _literally the only way_ to run a successful project (i.e. one that documents a real phenomenon). No researcher drops $500,000 of their research budget and years of their life into an RCT without doing some qualitative work to make sure they are actually onto something real (i.e. won't get a bunch of null results). The incentives just don't add up to support academics doing something that stupid today.
Totally off-topic, but the portrayal of Wakanda in the hit movie "Black Panther" as being quite snowy was influenced by director Ryan Coogler's visit to Lesotho, which is so high altitude and far enough south that it has a ski hill.
The main takeaway from this is that should be that the west should stay the hell out this stuff and leave these people be. It's extraordinarily difficult to have a positive impact, there's plenty of potential to make things worse, and there's no debt that we owe to these people (i.e. colonisation neither made Africa poor nor Europe wealthy - if anything the opposite is true).
If EA really want to help people in developing countries, it seems to me that efforts aimed at reducing climate change would be vastly more effective than e.g. trying to ensure that there's as many people in Africa as possible.
"Ferguson’s work exposes a possibility I hadn’t thought of before, in which “technical” and “apolitical” projects can expand the power of the state in unforeseen and potentially dangerous ways."
If you hadn't thought of this before, perhaps you're not the correct person to be writing this book review.
Is the author even familiar with EA?
"GiveWell has now released a survey, conducted by the group IDInsight, of extremely poor residents in Ghana and Kenya who are likely to benefit from programs like the ones that GiveWell’s top charities conduct. The survey was meant to provide input for how to weigh saving lives versus reducing poverty in ways that reflect the values of the charity’s recipients, not just its staff."
Ferguson's critique is a little unfair. What Ferguson proposed in its place in TAPM was to.... encourage revolution in South Africa and Lesotho through the National Union of Mineworkers, the then-largest union in COSATU. Certainly an answer that can't be mistaken for being anti-politics, but would that have passed any EA test? Why would a sitting government in Lesotho have allowed it? Would apartheid Pretoria have?
As we know, apartheid *did* actually then end in the following years anyway, and the COSATU-allied ANC took power. Did that lead to a transformation of the Lesotho economy via improved conditions for Basotho mineworkers and a greater or more reliable flow of remittances? No, in fact Lesotho's growth has emerged from a transition *from* a remittance economy to an export-oriented garments industry created through Taiwanese FDI principally oriented around the US market, as a strategy consciously pursued bureaucratically by the Lesotho government via the Lesotho National Development Corporation. Would any anthropologist have forecasted *that*?
Even a commitment to invite "locals" is fraught, because I would guess that there is always a selection effect as to who knows Western languages best; who is most active and ambitious and incentivized to bringing outside investment; etc.
For example, in the situation described above: if they worked in very close good faith with the government.... But the government didn't have the best.onterwstw of those specific villagers at heart. Or if you speak to someone from the village who is familiar with development workers and knows how to talk to them... Isn't necessarily the best spokesman for the village's best interest as a whole. And there will always be winners and losers in these interventions, who can then point to their loss and rightfully blame the interventionists.
That should be an important consideration too, and signs point to being careful and conservative about complex interventions, erring on the side of not, lest you incur the unpredictable consequences.
This book sounds so similar to Shumacher's "Small Is Beautiful," published a shade under half a century ago, and which I had to read in college, 40 years ago, that I had to laugh. As a species, our ability to learn nothing from experience, if that would conflict with our shibboleths and desires, is unsurpassed. I suspect very similar books will be published in 2072.
My first thought was, 'Hadn't Lesotho changed its name recently?' But I was confused: it was the other small country next to South Africa, which changed its name: now Eswatini, formerly Swaziland. (Lesotho was formerly called Basutoland, but that wasn't recently.)
I haven't read the book, just this review, so forgive me if this is addressed more carefully there. But some of these critiques sound a little too simplistic to me.
The fact that trees were uprooted and ponies killed is evidence that some individuals within the society opposed these things. They are not necessarily evidence of broad based opposition. For instance, lots of people in the US recently loudly proclaimed that all cops are bastards, and we should defund them. Whether or not you happen to agree with that, it is certainly not the majority view. But that small group is more than large enough to go around uprooting trees if they wanted to.
It's true that most interventions end up altering the balance of power in a society - even Malaria nets and deworming do this (the local witch doctor has got to eat too). Whenever power balances change, someone is hurt, and those people may lash out in some way. But that is not prima facie evidence that the society as a whole opposes them, or even that the development economists didn't consider these issues.
So, as someone very unfamiliar with development economics, and with no real preconceptions on the field, I really don't like this review. As I understand it from the review, the book claims that a field is essentially all systematically wrong-headed, and that it ignores naively obvious concerns in its haste to squeeze everything into its inadequate framework. That's not *implausible*, but we've heard effectively identical claims about many fields which are nonsense. See Scott's manual for reading popular media about psychology, for instance.
The first thing I want to see in a review of a book with claims like this is skeptical pushback on three fronts:
1.) Is this case he describes actually an absolute failure?
2.) Is it representative of development interventions, or a cherry-pick?
3.) What is a steelman, or better yet a strongman response from the mainstream? Is it fair, or wrong?
Without that critique, I think this review/book is kind of epistemically hostile to the uninformed on the topic. It's persuasive in the same way listing thousands of crimes commited by cardiologists is; honest, but not actually providing any more Bayesian evidence for the position than we would expect to see if the position was wrong.
While doing internal EA ethnography would be helpful, the more direct parallel would be to do ethnography of the charities (e.g. Against Malaria Foundation, GiveDirectly, etc.) into which EA constitutes an intervention. I think there are interesting stories to be told about how staffers and organizations are affected by positive and null results and how the layer of EA funding interacts with the existing cultures of measurement and evaluation.
Great review - both in writing style and in the choice of topic.
I particularly liked the analogy to an American suburb, I'm not sure if it's from the book or the reviewer's idea but it works well either way.
I'm not sure how much the potential criticism of Just Giving applies to GiveWell - as far as I know they settled on "give everyone in the village some cash" rather than means-testing.
My favourite example of metis in the comments is Deiseach and others going "why are you trying to breed ponies in cattle country?" as if that's the most obvious difference in the world - which it probably is to anyone who raises cattle. The locals would have told you this if you'd asked!
Thanks for breaking this issue down for beginners. I’m now curious about an anecdote from a friend of a friend who went volunteering in Africa. It was a local water supply project which the locals never seemed to appreciate. They preferred for the village women to continue fetching water from some distance. She decided the men were feckless & lazy. Perhaps so. Or maybe she missed something.
This was an interesting digression (as in, the point being made is interesting) but a pretty poor book review, and written a bit awkwardly. Scott's biases are showing a bit, I think.
I'm on mobile so this might not be true for everyone, but none of the footnote numbers are showing up in the body of the review. It's possible to figure out where each one goes, but since they're numbered I'm betting the references were supposed to be present in the text.
Before responding to the content, I want to thank the anonymous author for the thoughtful criticism of EA, and for helpfully summarizing a book that others have recommended. On net, this review has helped clarify some thoughts and potentially saved me a lot of reading time. Much appreciated.
Overall, I end up disagreeing with a few of the interpretations of Ferguson's book and its implications for EA. Some more detail:
- I'm confused about what a Pro-Politics Machine would look like. I agree that "development economics" (and EA) should avoid pouring money into projects that are net-negative due to factors like political corruption. This might even suggest a meta cause area of anti-corruption advocacy?
- But like a lot of highly-politicized areas, the scale (high) is often outweighed by the tractability (low) and neglectedness (very low). Also, wading into local/national politics risks the same sort of cultural bias and/or infantilizing that the authors warn against. We don't even know how to do this in Oregon, let alone Lesotho.
- There's a lot of talk about the ways in which the World Bank's project was suboptimal, but as an EA, I'm mostly concerned about whether it was net-positive/cost-effective. While it may not be the case here, it's not hard to imagine a project with similar downsides nevertheless being a worthwhile intervention, if the upsides were large enough.
- The review uses malaria nets as an example, but mostly to inject some FUD into existing EA analysis. I'm definitely not an expert, but my impression (e.g. from the recent 80K interview with James Tibenderana on SMC) is that lots of effort goes into understanding local cultural factors.
- And since saving children from premature death due to malaria seems like a really good bet, it would take a fair amount of "Anti-Politics" to even the scales.
- While it's impossible to know how the Lesotho project would have been rated by a contemporaneous EA movement, things like improving livestock quality/liquidity are not currently being promoted for obvious animal welfare reasons. And as a side note, the pony thing is quite sad too.
- While I'd be curious to see the results, hiring a bunch of anthropologist types to follow EAs around seems like the kind of thing that could lead to extremely bad publicity for the movement. A safer and probably better idea is to just directly fund "red team" projects (the FTX Future Fund lists this among its project ideas).
- In general, I agree that "low hanging fruit" will slowly be disappearing, and future EAs will have to consider more complicated interventions. And this is will be a really good thing, since it implies a much better world than our current one. But it seems a ways off for now, and the best opportunities for impact probably still avoid the types of political interventions necessary to solve the really sticky problems.
RE: RCTs, there was an interesting discussion of the limitations of the EA / GiveDirectly approach in this episode of Rationally Speaking with Angus Deaton: http://rationallyspeakingpodcast.org/246-deaths-of-despair-effective-altruism-angus-deaton/. Specifically, starting in the second half around 37 minutes in.
Deaton is skeptical about the value of RCTs in measuring the impact of development initiatives, and more generally is skeptical that you can inject money into a country with politically oppressive government and have that actually help the people you're targeting rather than the aid just ending up in the pockets of the government -- this seems to fit well with the narrative of the BNP using the development program's resources and reputation as a cover for attacking their political opponents.
I don't think I fully bite the bullet and completely dismiss the value of RCTs in the development context, but I think it's hard to be sure that you're correctly capturing the downstream consequences of an intervention with an RCT, and I think this bears further discussion/investigation.
I think you could perhaps be confident about a position like "very large mortality reductions from very targeted interventions are unlikely to have substantial negative side-effects that make the whole intervention net-negative" (e.g. mosquito nets), but it's harder to infer the downstream effects of a (by design) less targeted intervention like GiveDirectly's unconditional cash transfers. But to be clear, I do think it's still worth considering downstream effects even from mosquito net programs.
This is a well-written summary, but I thought the generalizability and "suggestions for EA" elements were weak.
I know a few staff at GiveDirectly well and have spoken to them in detail about their work. They are much, much more clued-in than the development workers portrayed here (who may or may not be typical of "development workers in the 1970s", let alone "development workers today"). They are constantly thinking about local cultural effects and potential downsides of transfers, and they have access to tons of qualitative information on how their programs are going from the local staff they work with and the thousands of locals they + their staff speak with.
GiveDirectly's blog has a bunch of interesting posts on the on-the-ground experience of managing cash transfers. I like this one for the "what we got wrong" examples: https://forum.effectivealtruism.org/posts/qjD9LjdJj4xzuCQoW/the-last-decade-10-things-we-got-right-and-wrong
I don't have as much experience talking to other people running top development charities in EA, but I'd be surprised if they didn't have the same focus on the qualitative aspects of their work + respect for what can go wrong. These programs are often their life's work; they care a lot about getting things right, and they know how easy it is to screw up (see Scott's "Yes, We Have Noticed the Skulls" for more on this mindset). And there's at least one prominent example of an EA-linked organization realizing a program wasn't working and shutting it down: https://www.evidenceaction.org/were-shutting-down-no-lean-season-our-seasonal-migration-program-heres-why/
Top development charities make mistakes, like any organization, and there's always more to learn. And some of the overall paradigms in development may be weak — I'm sympathetic to arguments that growth should get more focus (example: https://forum.effectivealtruism.org/posts/bsE5t6qhGC65fEpzN/growth-and-the-case-against-randomista-development). But viewing the people who work on evidence-driven development as naive armchair theorists seems totally off-base. (This last point isn't about this review, which is more open/curious, but is about many of the comments on the review.)
One more post that I think is a good intro to the things EA development people think about (note that it's from 2014 and GiveWell may be quite different now): https://forum.effectivealtruism.org/posts/NpuxBQTrHb9QuWHDA/what-i-learned-from-working-at-givewell
I'm a little hesitant to draw a lot of conclusions from the example provided. Obviously there was a screwup in looking at Lesotho; but was that emblematic of the way things usually go, or was it an exception? Lesotho is a really small country on a very large continent; it strikes me as very possible that the deveconomists on the project phoned it in, or maybe just weren't very good. It's a low sample size, and it might be a biased sample.
Great review! I don't know who the writer is, but I'd be happy to talk when the identities are revealed, I think we have much in common.
1. I've actually spent a few days in rural Thaba-Tseka speaking to villagers (though not in the 90s), and did not get the impression that their farming was just a pastime or the force of tradition - it's hard work! Though I did know many people worked in mines in South Africa. I'm curious if this theory of the sources of income and reasons for farming of the villagers has been quantitatively empirically tested on a reasonable sample. It could still be true, and I missed it, which goes to show that more in-depth work is required.
2. As others have said, development economics has improved a lot in this respect. Researchers and NGOs do in fact go and spend time speaking to villagers (I've done this in multiple countries, but also read "Poor Economics" or listen to the Conversations With Tyler interviews with Chris Blattman for examples). Part of the revolution of development economics becoming more micro and empricial has led to this, as researchers want to know which interventions to test to find an effect. The field does still have quite a bit of academic signalling, which is why I prefer more simply-empirical papers than sophisticated models and conjectures.
3. It's difficult to get good evidence from just talking to people. People often say contradictory things, they often don't even themselves know real reasons they do or don't do certain things. It is a very good tool to generate hypotheses, though.
4. I'd be wary of "having more EAs from the Global South" as a solution to all our problems. Most developing country EAs (and techies, and high-skilled immigrants) come from (relatively) very rich backgrounds. Mumbai or Bangalore are much farther from rural Bihar than you'd imagine, looking from the outside at India. It's true that they are closer, and would have more cultural context and speak the language, so they can learn faster and better. But I wouldn't automatically consider their views informed just because they are from the same country/continent. Getting representative beneficiaries as employees/advisers is harder than it sounds. They are often not very literate and skilled, don't speak English, etc..
5. Do you think Anthropologists studying EAs would have valuable input without contrasting it with some blind spot? I'm imagining someone studing EAs and saying "they eat, they sit next to computers, they talk." You need to know what they're missing in order to then investigate how it is that they're missing it, no? I actually had an anthropologist studying a development organization I was working in, I read her report and it was not at all informative (and barely legible, because of how sociologists write).
This is an interesting, focused, clear review with a mostly correct take on an important topic. Even so, I believe it doesn't help the discussion of that topic much; 8/10 is an intentionally generous rating compensating for my strong and somewhat eccentric disagreements with its approach.
Here's my problem: EAs are already plenty familiar, at least on a surface level, with the "High Modernist" accusation and its application to development economics in particular. Repeating it won't do much good. They'll just say "yes, we have noticed the skulls" and flatten the disagreement into a request for pointing out quantifiable shortcomings. (Cf. https://thezvi.wordpress.com/2022/06/06/transcript-of-a-twitter-discussion-on-ea-from-june-2022/ for another example of a rationalist-adjacent writer becoming frustrated with this flattening.) There's a hard-to-bridge gap between those who see the failures of "nerdy" approaches as being basically deficiencies in rationalist thought that can be driven toward zero with better data and better models, vs. those who see rationalist thought as a system that by nature generates and overlooks such deficiencies to a fatal extent no matter the quality of data and modeling available. In other words, they can't agree on whether more or less number-crunching is the fix here. (Also cf. in this respect the Scott Alexander vs Gary Marcus disagreement on AI trajectory.)
I sympathize with the rationalists in this sense: the anti-High-Modernist accusation is _cheap_. No matter how many specific flaws the rationalist addresses, the accuser can stay hard to pin down, can insist that there must be more unspecified problems out there, that the rationalist hasn't understood the gist of the accusation. If one accepts the objection in principle it's hard to say why it doesn't constitute an objection to attempting *any* meaningful change at scale. On the other hand it's hard to deny that there exist problems with rationalist approaches, not just object-level gaps but an operational Problem Factory.
This review is helpful insofar as it gives a detailed Mistake Theory description of how one particular instance of High Modernism went wrong. I also thought the suggestion of an "anthropology of EA" was clever, and aligned in its own way with a different thread of rationalist tradition (awareness of bias). But it doesn't go into enough depth on either of those avenues of investigation to really make them stick. As it stands EAs can just shrug it off. Either a much more fundamental critique of EA, or a much more detail-intensive one, is required.