The problem with principles is they don’t tell you much about consequences. We might all agree that we should help very poor people in region X, but what will a

The problem with principles is they don’t tell you much about consequences. We might all agree that we should help very poor people in region X, but what will actually help most? Micro loans, micro loans with coaching, unconditional lump sum cash at random, unconditional lump sum cash to targeted promising candidates, free schools, schools with small fees? We can’t know how the different effects of these things compare from a principled argument. That’s where evidence is useful.