Speculation on Current Opportunities for Unusually High Impact in Global Health

post by johnswentworth · 2022-11-11T20:47:03.367Z · LW · GW · 31 comments

Contents

  What’s The “Sahel” Region?
  Bad How?
  Die How?
  The Next Year In Particular?
  So What’s To Be Done?
  How Much Impact?
  How Would You Measure Impact?
  Why Expect This Fruit To Be Unpicked?
  Why Are You Writing This?
None
31 comments

Epistemic Status: armchair speculation from a non-expert.

Short version: I expect things to get pretty bad in the Sahel region over the next year in particular. The area is an obvious target for global health interventions even in good times, and impact is presumably higher in bad times. A simple baseline intervention: fill a backpack with antibiotics, fly to the region, and travel around distributing the antibiotics.

What’s The “Sahel” Region?

The Sahel is a semi-arid region along the southern edge of the Sahara desert. Think roughly Mali, Niger, Chad and Sudan.

Bad How?

Based on statistics on the Sahel, it’s one of the few remaining regions on Earth where the population is near Malthusian equilibrium. Fertility is high, contraception is rare; about half the population is under age 16. Infant mortality is around 6-8%, and ~a quarter of children are underweight. (Source: CIA World Factbook entries on Mali, Niger, Chad and Sudan.)

Being near Malthusian equilibrium means that, when there’s an economic downturn, a substantial chunk of the population dies.

Die How?

Traditional wisdom says: war, famine, disease. In this case, I’d expect famine to be the main instigator. Empty bellies then induce both violence and weak immune systems. On priors, I’d expect infectious disease to be the main proximate killer.

The Next Year In Particular?

The global economy has been looking rough, between the war in Ukraine shocking oil and food markets, and continuing post-Covid stagflation. Based on pulling a number out of my ass without looking at any statistics, I’d guess deaths from violence, starvation, and disease in the Sahel region will each be up an order of magnitude this year/next year compared to a good year (e.g. the first-quartile best year in the past decade).

That said, the intervention we’ll talk about is probably decently impactful even in a good year.

So What’s To Be Done?

Just off the top of my head, one obvious baseline plan is:

• Fill a hiking backpack with antibiotics (buy them somewhere cheap!)
• Fly to N'Djamena or take a ferry to Timbuktu
• Obtain a motorbike or boat
• Travel around giving away antibiotics until you run out
• Repeat

Note that you could, of course, substitute something else for "antibiotics" - maybe vitamins or antifungals or water purification tablets or iron supplements or some mix of those is higher marginal value.

There are some possibly-nonobvious considerations here. First, we can safely assume that governments in the area are thoroughly corrupt at every level, and presumably the same goes for non-government bureaucracies; trying to route through a local bureaucratic machine is a recipe for failure. Thus, the importance of being physically present and physically distributing things oneself. On the other hand, physical safety is an issue, even more so if local food insecurity induces local violence or civil war. (That said, lots of Westerners these days act like they’ll be immediately assaulted the moment they step into a “bad neighborhood” at night. Remember, folks, the vast majority of the locals are friendly the vast majority of the time, especially if you’re going around obviously helping people. You don’t need to be completely terrified of foreign territory. But, like, don’t be completely naive about it either.)

Also, it is important to explain what antibiotics are for and how to use them, and there will probably be language barriers. Literacy in these regions tends to be below 50%, and presumably the rural regions which most need the antibiotics also have the lowest literacy rates.

How Much Impact?

I’m not going to go all the way to estimating QALYs/$ here, but… according to this source, the antibiotic imports of the entire country of Mali in 2020 amounted to $53k. That’s for a country of 18 million people and change. Now, I certainly wouldn’t take that statistic at face value, but I think we can safely conclude that Mali does not have anywhere near the amount of antibiotics the population could use.

Even if each course of antibiotics bought and distributed has only a 0.1% chance of saving someone’s life, if you can get the antibiotics for $1/course someplace cheap (and not spend too much money travelling around distributing them) that’s still $1000/life - pretty respectable impact/$. And that’s a pretty conservative estimate -  I would guess that you could get closer to $0.10/course with a little shopping around and a large bulk buy, and I’d guess that each course saves more like 1/100 life rather than 1/1000 (assuming the antibiotics aren't completely wasted - starving people are known for eating things, which is another potential issue). So impact could easily be higher by one or two orders of magnitude.

How Would You Measure Impact?

The simple first-pass answer is to not measure impact, and just operate entirely on priors. Not great, but hey, the lack of good feedback loops is often a major cause of low-hanging fruit.

The next better answer would be to do another round a year later, revisit some of the same places, and ask people what they did with the antibiotics.

Why Expect This Fruit To Be Unpicked?

Obvious factors here:

• Risk tolerance, specifically risk of physical violence. I’d guess that most individuals would be fine with a moderate amount of physical risk, but for a large organization it’s a PR disaster when one person out of a hundred or a thousand gets kidnapped once. So, I expect orgs to systematically underinvest in this sort of thing (and/or partner with local governments, which have massive overhead at best and usually massive corruption too).
• Just directly physically doing things seems to be high-impact in general

Why Are You Writing This?

That last point is actually the main reason I’m writing this post: just directly physically doing things seems to be high-impact in general. It is not a coincidence that most of GiveWell’s top charities directly distribute physical things. Just ignore the many opportunities to “partner with <bureaucracy>”, or network with Important People, or whatever other social games are fashionable, and go directly solve (some part of) the object-level problem.

It seems to me like a majority of newcomers to EA get lost in social reality [? · GW], and lose track of physical reality. That manifests, for instance, as a tendency to generate ideas like “try to Gain Influence and use it for X”, and to not generate ideas like “fill a backpack with antibiotics and travel around the Sahel distributing them”. My current best guess is that marginal focus on social influence, rather than thinking about physical reality, is orders of magnitude less valuable across cause areas; this post is meant to be an example. (And yes, this also applies to AI.)

… but hey, maybe I’m wrong about that. I’m a non-expert engaging in armchair reasoning here, after all. So if you think the “backpack full of antibiotics” plan is a dumb idea, feel free to tell me I’m wrong.

31 comments

Comments sorted by top scores.