Can philanthropists be ruthless?
Charles Kenny and Justin Sandefur have taken over the cover of the latest issue of Foreign Policy, with a classic of the QTWTAIN genre. The tl;dr version of their 4,000-word article (which is free, behind a registration wall): no, Silicon Valley cannot save the world. Still, there are some very good development innovations out there; the surprising thing is that you’re likely to find them not in Palo Alto but rather in Washington DC.
The strongest point that the article makes is simply that you can’t easily apply Silicon Valley techniques to development. If you look at the bright ideas of the recent past, many of them have failed — which is entirely typical of Silicon Valley. The problem is that the people who develop and fund those bright ideas tend to be the winners of the Silicon Valley lottery, tend to think that their ideas are golden, and are loathe to kill them off. As Kenny and Sandefur write:
When working with new technologies and approaches, we should expect lots of failures. Tech entrepreneurs are used to a culture of failure — 10 bad ideas that sink before one gets to its multibillion-dollar initial public offering. But the advantage of the system in which they operate is the market test. As a rule, the bad ideas go bankrupt (if sometimes only after the multibillion-dollar IPO).
But by moving the model to development, we’ve taken tech entrepreneurs’ high tolerance for failure and penchant for hyping harebrained schemes to an arena where the market test is considerably diluted. Ideas get funding from Kickstarter and philanthropies on the basis of their appeal to donors and philanthropists in the West rather than consumers in Africa. And that’s what leads to the Soccket, the PlayPump, and One Laptop per Child.
One of the joys of the cover story is the way in which it elegantly dismantles these innovations, using a combination of empirical evidence and simple logic. (“You can get a solar-powered lamp for $10. It isn’t clear why anyone would pay 10 times that for a light whose power source you have to kick around for half an hour to get less illumination.”)
But after scolding “tech gurus” for being “wildly overoptimistic” in the first half of the article, it’s then a little jarring for Kenny and Sandefur to start getting all starry-eyed themselves in the second half, talking up a tiny division of USAID as the engine which, even if it won’t save the world, could at least save millions of lives and have an enormous positive impact on development.
Kenny and Sandefur run down a list of three bright ideas funded by a Washington venture capital fund called Development Innovation Ventures; all of them seem very promising. But the whole point of the first half of the article is that promising bright ideas are all too easy to come by, in the development world: the difficult thing is identifying which ones aren’t working — and then killing those ideas with all the ruthlessness of a Silicon Valley venture capitalist. And there’s nothing in the article which shows DIV being particularly good at that.
This area — of judging nonprofit development projects by their empirical outcomes — is one where GiveWell has probably done more work than anyone. And if you read down their analyses, you’ll learn quite quickly that it’s an incredibly hard thing to do. Kenny and Sandefur make it seem that “rigorous testing” is something which can be implemented relatively easily — but in reality it ranges from the incredibly difficult to the downright impossible. Remember that even in carefully-designed peer-reviewed settings, most published research findings are false. And when you’re out in the field, trying to create a replicable test — which of necessity will involve withholding aid from a control group of people who likely need it — becomes extremely fraught on both a practical and a moral level.
In other words, what the article lacks is not only some kind of evidence that DIV’s bright ideas would not have been funded without it. It also could use, ironically enough, a rigorous demonstration that DIV is particularly good at rigorous testing, and that as a result it’s measurably more effective than the Silicon Valley crew. If Kenny and Sandefur really want to differentiate DIV from the rest of the field, they’re going to need to talk much more about its failures, and about the well-intentioned projects which it cold-bloodedly defunded for the greater good of the larger program. Otherwise, we’re left basically where we started, telling feel-good stories.