In my mind any evaluation of charities based on 'expected value' has to take into account not only the error bars on how good the charity is, but also the error bars on how good that evaluation is. In other words, how well the evidence stands up to an evaluation by someone who's sceptical about the whole thing. After all, the whole point of effective altruism is being rigorous in evaluating impact.

For an example, imagine a nuclear disarmament charity. Nobody doubts that mutually assured destruction would be very bad! The question is what's the chance of that happening - and estimates of the chance can vary by two orders of magnitude, putting the charity either at the top or the bottom of the impact league tables, depending on who does the evaluation.

In my mind that's a charity whose impact and effectiveness can't be measured, so it doesn't count as EA.

That's why I give my donations to the Schistosomiasis Control Initiative and Against Malaria Foundation.