Mason  🏃‍♂️ ✂️

I think the evidence for cloth masks is really ambiguous and we're all making a high stakes call in an environment of intense uncertainty. (The evidence for non-cloth masks is much stronger, but we have shortages of those.)

Trying to have the courage of my convictions, here, but not more courage than I have convictions. I give, like, a 40% chance that cloth helps and a 30% chance it hurts and a 30% chance that we are never able to detect an effect of any kind.

I think "30% hurts" is *really* high. No study is perfect, but the one everyone is passing around does not, in any meaningful sense, have a control group. Second, looking at health workers who'd be at high risk even in full PPE w/ decontamination protocols doesn't tell us much.

yeah, that's fair. Divia's point (that having a reservoir close to your face potentially matters a lot for bacteria, and can't for viruses) made me update a little; I'd probably put it lower now

I'm curious where that 30% is coming from. is it a specific model of how harm could happen? or just a general "medicine is messy we should be less confident" thing?

impression that medical studies that are counterintuitive and not extremely methodologically strong replicate like 30% of the time

I mean, here's how I'm reading this study: 1) throw out the control group; it's genuinely worse than useless as described — for all we know, they always wore a medical mask in close contact with the sick 2) conclude that medical masks work better than cloth masks — makes sense!

Hmmm, I think you can do better than my 30% heuristic here but throwing out the control group, the study still finds that the difference between surgical and cloth is too large to be consistent with cloth being net-positive unless surgical is much much better than anyone thought