I should probably rephrase that as overwhelming evidence that they are effective in a clinical setting---ie, in a lab, and in medical settings. The point is, if

I should probably rephrase that as overwhelming evidence that they are effective in a clinical setting---ie, in a lab, and in medical settings. The point is, if you're sitting in a room with an infected person, there have been many studies that show that you will get a lower exposure if you're masked, and a much lower exposure if the infected person is masked as well. I posted a more detailed explanation of my theory above. You don't need to account for imperfections in masks, or ventilation, or any other details. Even if you make the assumption that masks are perfectly effective, the observations seen by the researchers might be explained by exponentially increasing spread among the unmasked. To use an analogy, a 4' child can drown just as easily in a 5' deep pool as in 100' deep lake. The percentage of mask usage referenced in the paper would only dictate the depth of the pool, so it's not surprising it would have little effect once the virus takes hold in the unmasked population. Barring very high mask usage (making the pool only 2' deep, to continue the analogy), exponential infection spikes in the unmasked population are unavoidable. It doesn't mean masks don't work, it only indicates just how high usage needs to be to stop these spikes.