That's not the point! The point is that the intervention is natural, i.e, just abstinence. The person isn't consuming pharmaceuticals or inserting a mechanica

That's not the point! The point is that the intervention is natural, i.e, just abstinence. The person isn't consuming pharmaceuticals or inserting a mechanical device. Likewise, if I use sophisticated ML techniques to identify young students who have attention problems, but then the actual treatment is 1-on-1 tutoring (rather than, say, drugs), this is still a natural intervention. And it is reasonable to suspect that it has fewer side-effects on account of being natural, which must be traded off against it's lower effectiveness. Personally, I generally find the distinction between natural and unnatural interventions mostly unhelpful (since there are plenty of natural ways to hurt yourself), but it's a meaningful distinction and one that can plausibly provide useful guidance to someone who is very risk adverse.