Say you want to help improve the lives of the poorest people in the world. There are lots of organizations at work on that problem — building schools, digging water wells, vaccinating kids, providing medical care. How do you figure out which ones can help the most people effectively with your donation?

This is the question that the charity evaluator GiveWell set out to answer when it was launched in 2007. It’s a complicated question that gets to the heart of what makes effective altruism challenging, and today I want to focus on one key aspect of it.

Imagine you’re comparing two potential interventions. One of them has lots of evidence for how well it works, and in this case we can be pretty sure that the effects are real and substantial. A good example would be distributing bed nets that protect against malaria transmission. It’s well studied, the effects and side effects are well known, and it saves a lot of lives.

The other potential intervention is much more uncertain. There’s some high-quality research suggesting that it could have massive, transformative effects. That research has been debated endlessly and it’s largely held up — but other studies on other populations haven’t found effects nearly as large. This intervention is very cheap, so if it works, it would be an incredible deal. But it might do almost nothing at all.

Which of those is the “better charity”?

The return of the worm wars

The World Health Organization estimates that 1.5 billion people worldwide, mostly in the poorest places on Earth, are battling helminth worm infections, while more than 200 million are affected by another kind of parasitic worm called blood flukes. These worms generally feed off the tissue of the host, and can cause anemia, malnutrition, and potentially significant health complications.

In 2003, Nobel Prize-winning economist Michael Kremer published the results of a large randomized controlled trial (RCT) finding that deworming schoolchildren in Kenya, where intestinal parasites are very common, had large long-term effects on lifetime income, likely thanks to the students being healthier and missing less school time due to illness. As a result, many affected countries launched or scaled up intestinal parasite treatment programs.

But Kremer’s study wasn’t the final word on deworming, with many international development experts questioning whether deworming actually works as well as his paper implies. Some researchers purported to “debunk” Kremer’s study, followed by other experts debunking the debunking.

Other studies of deworming different populations have struggled to find similarly large long-term effects, though they do find that children gain more weight once their parasites are treated. That’s a good sign but not enough to make deworming a world-leading health charity by itself.

Throughout that controversy, GiveWell has continued to recommend some charities working to eliminate intestinal parasites. But that’s not because they’re decisively on the pro-deworming side of a debate that has become known as the “worm wars.”

“We believe that there is strong evidence that administration of deworming drugs reduces worm loads but weaker evidence on the causal relationship between reducing worm loads and improved life outcomes,” the very first sentence of their review of a top deworming charity reads. “We consider deworming a priority program given the possibility of strong benefits at low cost.”

In other words, the logic for deworming currently goes: whether combating intestinal parasites has a large effect on long-term health and therefore on life outcomes isn’t yet clear. But since it’s so cheap and might have those huge benefits, it’s a good use of money in the crowded field of ways to improve the lives of those in extreme poverty even if it’s reasonably likely to not work as well as projected.

Like any long-running conflict, the worm wars flare up from time to time — and they’ve done so in the last few weeks. On a recent episode of the nutrition podcast Maintenance Phase, former HuffPost reporter Michael Hobbes made the case that deworming is probably useless. I think he overstates the argument, but he’s right that it might not do much (beyond directly removing the intestinal parasites, which everyone agrees that it does).

But our big disagreement isn’t about whether or not deworming has major long-term effects. It’s about what to do with that uncertainty.

You can’t dodge acting under uncertainty

I talk to people who are excited about work on global poverty and health interventions but skeptical about projects like trying to make the development of artificial intelligence go well or prevent the next pandemic. This position often gets phrased in terms of wanting to do things with certain benefits instead of things with uncertain benefits. In other words, wanting to be sure that you’re actually doing good in the world.

Some interventions are much more of a sure thing than others. The evidence around malaria treatments is much less muddled than the evidence around deworming, and if someone tells me that certainty is very important to them, I recommend they donate to anti-malaria efforts rather than to deworming.

But fundamentally, I think doing as much good as we can — and tackling the world’s hardest and most important problems — requires coming to terms with uncertainty and being willing to act even when our evidence is mixed. Those interventions where all our research points in the same direction are rare.

Despite our best efforts, uncertainty remains. But it’s not responsible or effective to refuse to act on your best current understanding until there’s sufficient evidence to dispel that uncertainty. Instead, you should do what you currently think is best, taking into account all your uncertainties, and be ready to change course if you’re wrong.

That might mean writing a recommendation like the one GiveWell offers for deworming, which reads quite weirdly if you’re used to more typical charity recommendations. It more or less directly says “we’re very unsure if this works, but if it does, the benefits are sufficiently large that it’s worth doing.”

GiveWell recommending deworming while acknowledging that it is unclear whether the intervention has the hoped-for long-term effects has confounded some critics. “GiveWell has doubled down on its advice” to back deworming programs, while saying openly that it’s unclear if they have long-term effects, philosopher Kate Manne wrote in a recent piece arguing that the uncertainty about deworming should undermine the entire premise of effective altruism. “This is difficult to reconcile with their advice to keep on giving.”

But the reconciliation is quite simple: to do as much good as possible, sometimes you have to act on mixed and limited evidence. That can mean supporting deworming while openly saying it’s unclear how much good it does. It also might mean working on a longer-term problem like developing a malaria vaccine or discouraging nuclear proliferation or figuring out safe AI development, where it’s very hard to tell in advance if your work will make any difference at all.

This shouldn’t be taken as an excuse to just do whatever you feel like without reviewing all the evidence. (That’s a recipe for ineffective altruism.) But it will often mean that when you’ve summed up all that evidence, it won’t offer certainty, just a best guess — and you still have to move forward.

The worm wars show that you can’t escape that challenge, even if you work on interventions recommended by RCT-backed Nobel-prize-winning economists.

Correction: Michael Hobbes was formerly a reporter at HuffPost, not BuzzFeed.