In March I discussed what I described as a positive trend in the discussion of the Covid-19 pandemic policy: the New York Times quoted an infectious disease expert about the six-foot social distancing rule who admitted “There’s risk at six feet, there’s risk at three feet, there’s risk at nine feet. There’s risk always.” He added: “The question is just how much of a risk? And what do you give up in exchange?” His characterization of the decision problem is exactly correct but does not seem to recognize that calculation of the tradeoffs is not a scientific or medical question.

The latest policy scrum is about the wearing of masks outdoors. David Leonhardt has recently challenged the C.D.C. for its continued support of outdoor mask-wearing. The basis for the recommendation is a characterization of the literature as concluding that “less than 10 percent” of Covid-19 transmission was occurring outdoors.

Leonhardt plays the role of literature reviewer and concludes that many of the outdoor transmission cases in the literature come from Singapore and are probably misclassified. The author of one study classified “workplace, health care, education, social events, travel, catering, leisure and shopping” cases as outdoors. And yet even with this classification outdoor transmission cases were less than 1 percent. But one Singapore study suggested the share of Covid transmission occurring outdoors was close to 10 percent even though many of those cases involved Singapore construction workers who probably transmitted it in enclosed spaces.

The director of the C.D.C appeared before Senate to defend the 10 percent number, which came from a study published in The Journal of Infectious Diseases. In her testimony, the director used the term “meta-analysis,” as if the 10 percent figure was the point estimate of the average from a statistical review of all existing studies. But the study’s author, in an interview with Leonhardt, said that the study was not a meta-analysis. It was a “systematic review,” a narrative rather than statistical summary of the literature. Most studies in the review found the share to be below 1 percent. But there was one study that suggested the share of Covid transmission occurring outdoors was close to 10 percent. That is why the authors used the phrase “less than 10 percent” in the paper.

Leonhardt provides a more complete and thorough description of the literature than the C.D.C. director. And he criticizes the C.D.C. for not understanding that risk aversion is separate from science. “Telling people to wear a mask outdoors all summer is an example of extreme caution — like staying out of the ocean to avoid sharks — that seems to have a greater cost than benefit.” “C.D.C. officials have acted as if extreme caution has no downsides. Everything has downsides. And it is the job of scientific experts and public-health officials to help the rest of us think clearly about the benefits and costs of our choices.”

A year ago the narrative was just ‘follow the science’. Now ‘follow the science’ as long as it is not too risk-averse. But that is not quite right either. Very risk-averse people might want to take the 10 percent outdoor transmission risk estimate seriously. The C.D.C should disseminate knowledge, but not tell us what to do.