I think the Precautionary Principle" is a bogus way of thinking about risk. Eg, reopening schools. The precautionary principle must work in both directions – we (a) mustn't risk another outbreak and (b) mustn't risk children missing more schooling. It's a useless guide to action.

Instead, I like what's called the "error cost" approach: ask what path is likely to be worse if we're wrong, and weigh your decisions that way. What's the likely cost if, eg, it turns out kids do spread Covid and we send them to school? What if they don't and we keep them home?

A key element of this is reversability. If one option turns out to be a mistake, how easily can we change course? Another is identifiability: How easily can we tell that we've made a mistake?

This is a useful way of thinking about some forms of regulation, too. Sometimes a regulation against some kind of activity wipes it out, so we can't learn in future if we were wrong about it, whereas allowing it allows us to learn more – and prohibit later, if necessary.