There are a few very good reasons why the BAAQMD’s devices sacrifice speed for accuracy, and for using a system that calibrates with what environmental agencies across the nation are using. “When the EPA, the federal government, makes decisions about air quality on a national level, they can say with some level of confidence that the network in New York is giving you the same type of information as a network in the Bay Area,” says Michael Flagg, principal air quality specialist at the BAAQMD.

This data has to hold up in court when, say, the government needs to prove a company is polluting a given area. Accordingly, the feds have strict policies in place for these AQI-testing machines. “They have to meet certain EPA siting requirements: They have to be greater than 10 meters away from trees. They have to have unobstructed airflow,” says Flagg. “And also the regulatory data undergoes rigorous quality assurance and quality control to ensure the data is accurate.”

PurpleAir’s sensors don’t have to meet these strict rules. People can put them anywhere, including places an air quality expert would know to avoid. Owners might be placing them near chimneys, for instance, throwing off the readings for wildfire smoke. But what PurpleAir might lack in accuracy, it makes up in sheer numbers: AirNow.gov’s map shows one monitor in San Francisco, while PurpleAir’s map shows dozens of monitors within a square mile of my apartment. If one monitor is showing a wildly aberrant AQI reading, and all the others nearby are in general agreement, you get a kind of accuracy by way of averages—and you’re getting it in real time.

“This network is designed to know what the quality is right now,” says Dybwad, of PurpleAir. “And also by virtue of how many there are, you can then say, ‘Look, this one over here is reading, let’s say, green, and I don’t believe that because all of these others are reading orange.’ So just by sheer numbers, it becomes very persuasive in terms of the fact that they all agree.”

And just because PurpleAir’s monitors aren’t as accurate as BAAQMD’s, doesn’t mean the agency’s staffers scoff at the data. It’s quite the opposite, in fact. “The regulatory monitoring network is kind of the backbone of our decision-making, and we do that because we can trust the data are accurate,” says Flagg. “And with PurpleAir, we use that data in a qualitative sense. It can be really good at understanding if concentrations are increasing rapidly or decreasing, or if one area is experiencing poor air quality compared to a different area, and things like that. What PurpleAir can be good for is looking at the spatial distribution of smoke during a wildfire, like we’re experiencing now.”

All that data may also be useful in another way, says Adrienne Heinz, a research psychologist at the National Center for PTSD: It’s oddly compelling. Like for many others sitting in San Francisco’s orangey gloom, relentlessly updating my PurpleAir and AirNow.gov maps offers me a way to grasp at some kind of certainty—any kind of certainty—as the Bay Area suffers through this historic collision of disasters. “The more that you can put data into the hands of users, it can be comforting,” says Heinz, who studies the effects of disasters like wildfires and the Covid-19 pandemic. “Obviously, there’s a threshold, right? Like checking PurpleAir 20 times a day, that’s not helpful. But anything that can put it in the hands of consumers and citizens, helps us all come together to make more informed decisions.” So, for instance, timing forays into the outdoors when air quality improves.