This is Work in Progress, a newsletter by Derek Thompson about work, technology, and how to solve some of America’s biggest problems. Sign up here to get it every week.

In the past few weeks, the conventional wisdom about COVID seems to have been upended.

Early in the pandemic, several mainstream news outlets dismissed theories that COVID came from a Chinese lab. But recently The Wall Street Journal and The New York Times reported that the Department of Energy reversed its prior judgment by announcing that the coronavirus probably did emerge from a laboratory. The FBI shares that assessment.

What’s more, for the past three years, many scientists and writers—including me!—have reported that masks are effective at reducing the transmission of COVID. But last month, the lead author of a comprehensive analysis of masks boldly and unequivocally asserted that “there’s no evidence that many of these things make any difference.”

That settles things: The elites got everything perfectly backwards; the lab-leak conspiracy theory was true, and the mask mandates were a fraud!

Well, not quite. The deeper you dig into the details of each case, the murkier the story becomes. In fact, the deeper you dig, the more you realize that murkiness is the story.

Start with the lab-leak hypothesis. Three years ago, many journalists and scientists rushed to condemn a theory that deserved a fair and open trial. But let’s not replace one nutty take (The lab-leak theory is racist) with another (We know for sure that COVID came from a lab). Although the Department of Energy and FBI say the virus likely emerged from a lab rather than a wet market, four other agencies and the National Intelligence Council have come to the other conclusion: that COVID likely started with natural exposure to an infected animal. By this count, the lab-leak theory is still an underdog, trailing 5–2 among government institutions. Adding to the confusion is the fact that none of the agencies reached their conclusion with much conviction, even with access to untold stacks of top-secret information. As my colleague Dan Engber pointed out, “Only one [assessment], from the FBI, was made with ‘moderate’ confidence; the rest are rated ‘low,’ as in, Hmm, we’re not so sure.”

In an ecosystem of doubt and paranoia, spooky factoids breed. Have you read about those sick researchers at the Wuhan Institute of Virology back in November 2019? Have you read the response to the response to the rumor about an earlier alleged biosafety breach at WIV? Bro, can you even spell “furin cleavage site”? Tantalizing leads, all. But they add up to a tug-of-war between a clever hunch and an educated guesstimate.

The frustrating truth is that we’ll probably never know for sure how the pandemic started. China’s refusal to grant access to global investigators is sketchy, but we don’t know what they’re trying to protect or conceal.

In the absence of certainty, we should proceed as if both theories are true. That means much more federal scrutiny of gain-of-function research in U.S.-backed labs. That also means reconciling ourselves to the probability that COVID will not be the last pandemic of the century—or, perhaps, the decade. After more than 1 million American pandemic deaths, “taking the pandemic seriously” seems to mean civilians posting condemnations of other people’s behavior online rather than the federal government laying out a clear and comprehensive anti-pandemic strategy to ensure, for example, the accelerated manufacture of vaccines and other antivirus therapeutics.

And speaking of civilians continually screaming at one another, let’s talk about masks.

The review by Cochrane, a London-based health-research organization, looked at 78 studies in total, including 18 trials focused solely on mask use. Their stated objective was simple: “to assess the effectiveness of physical interventions to interrupt or reduce the spread of acute respiratory viruses.” In short, do masks work? The authors concluded that they don’t. “There is just no evidence that [masks] make any difference, full stop,” a co-author, Tom Jefferson, said.

Sounds definitive. So I called several sources whom I’ve found to be honest and informed on the issue of masks in the past three years. Jason Abaluck is a Yale professor who ran a massive, multimillion-dollar study on community masking in Bangladesh. Possibly the most comprehensive masking study ever undertaken, it found that community-wide mask wearing provided excellent protection, especially for older Bangladeshis. “The press coverage” of the Cochrane review “has drawn completely the wrong conclusions,” he told me. Jose-Luis Jimenez, a professor at the University of Colorado at Boulder who studies the transmission of airborne diseases like COVID, is one of the country’s most cited researchers on the nature of aerosols. “I think it’s scientific garbage,” he said of the review.

Abaluck, Jimenez, and other like-minded researchers have an extensive list of grievances with the Cochrane paper. One criticism is that some of the most convincing evidence for masks from laboratory and real-world studies was left out of the review. The best reasons to believe that masks “make a difference” as a product, Jimenez said, are that (1) COVID is an airborne disease that spreads through aerosolized droplets, and (2) lab experiments find that high-quality face masks block more than 90 percent of aerosolized spray. Meanwhile, observational studies during the pandemic did find that masking had a positive effect. For example, a 2020 study comparing the timing of new mask mandates across Germany found that face masks reduced the spread of infection by about half.

But most important, the researchers identify a mismatch between what Cochrane set out to discover and what the studies in its meta-analysis actually examined. Cochrane looked at randomized control trials, where, in many cases, researchers split a population in two, gave one half a bunch of masks and information about proper masking, then came back a few months later to see if the intervention group was any healthier. For the most part, Abaluck and Jimenez said, these studies don’t really ask the question Do masks work? Instead, they ask: When you hand out masks and information to an intervention group without much enforcement, does it make them healthier? That’s a subtle but important difference, because the frustrating truth is that, without encouragement and social norms, people tend not to wear face coverings properly.

In one famous Danish study, which concluded that urging people to wear surgical masks failed to reduce infections, fewer than half of the people in the masking group said they fully “wore the mask as recommended.” In a 2022 study that distributed masks in Uganda, more than 97 percent of participants reached by phone said they “always or sometimes” wore masks. But at the end of the study, researchers concluded that just 1.1 percent of people they observed “were seen wearing masks correctly”—88 times less than the phone survey. Another study from Kenya found that participants were roughly eight times more likely to report mask usage than to actually wear them.

See how complicated this is? Many people who claim to wear masks actually don’t. Many people who do wear masks wear them improperly. The questions Do masks work? and Does merely asking people to wear masks do much? are not interchangeable.

Failing to pick nits like these can lead to very wrong conclusions. Imagine you found 100 papers showing that it’s hard to get kids to replace sugary snacks with broccoli. You write up the results in a meta-analysis, with the conclusion: Broccoli “does nothing” and “makes no difference” and is metabolically equivalent to Twinkies. But wait, that’s absurd, and you have not discovered anything like that! What you might have discovered is that, in the absence of highly informed and conscientious parenting, federal broccoli mandates will be mostly ignored by many families. That’s an important finding, but it’s very different from “BREAKING: SCIENTISTS SAY VEGETABLES ‘DON’T WORK.’”

“Poor-quality masks, worn poorly, work poorly, and high-quality masks, worn properly, work well,” Jimenez offered as a summation of the evidence. For that reason, I think it is reasonable to say that mask mandates probably reduce COVID in settings where high-quality masks exist and social norms of mask wearing can be maintained. Abaluck’s Bangladesh study achieved a roughly 30-percentage-point increase in community-level mask wearing by not only distributing free masks but also telling people how to wear them, modeling effective face-covering, and encouraging people out and about to put their masks on. By contrast, as even Abaluck acknowledged, “if Alabama tomorrow mandated mask wearing, it would do nothing.”

So what are you supposed to do about all this? The lab leak is neither a fact nor a myth. Masks work, except very often they don’t, and asking people to wear masks can work, except very often it doesn’t work at all.

Meanwhile, we—you, me, governments—have to make discrete and sometimes irreversible decisions within these clouds of uncertainty. I’m trying to navigate that uncertainty myself, reaching provisional conclusions as I constantly reassess the evidence.

I share the Department of Energy’s assessment, even though I don’t have access to its information. I think the lab leak is probable, by the slimmest of margins, and have also reconciled myself to the fact that I’ll never know for sure. I think the government should proceed as if the lab leak is 100 percent true and push for global gain-of-function limitations that reduce the likelihood of future catastrophic lab leaks. I’m going to keep wearing N95 masks in public indoor spaces during periods of elevated COVID transmission. I think that my neighborhood, in Washington, D.C., would benefit from an indoor mask mandate during high-transmission periods, even as I suspect that many unenforced mask-mandate policies around the world don’t do much, because of poor adherence and no enforcement.

The lab-leak and mask debates touch on a broader theme, which is the relationship between science and modern media. In a fragmented and contentious media environment, scientific communication is a mess. An abundance of crappy or confusing research gives audiences access to an armory of factoids, from which they can construct and defend any narrative they choose. For every position, there is an ostensible expert, an apparent paper, and an alleged smoking gun. Thus, the internet tends to serve as an infinity store for pop-up conspiracy theorists.

My advice in navigating this mess is: Do not trust people who, in their handling of complex questions with imperfect data, manufacture simplistic answers with perfect confidence. Instead, trust people who allow for complexity and uncertainty. Trust people who change their mind when the evidence changes. Trust people who, when they say “Believe the science!” put their trust in science, with a small-s, which is the dynamic reevaluation of complicated truths, rather than SCIENCE, in weird caps-lock font, which has come to mean the faith that for every random political position, there exists an official-looking study to permanently justify it. I wish the field of epidemiology was made up of immutable laws as settled as the roundness of the Earth and the power of gravity. It’s not. Its priors are vulnerable to reevaluation. If you want to stay right in this space, you have to be curious enough to potentially prove yourself wrong. You have to keep paying attention. For better or worse, that’s science.