Democracy is a Data Schema
Learning from the EPA about trustworthy tech governance
What does a trustworthy tech governance system look like? Last month I saw the most democratic diagram I’ve ever seen — the schema for who gets notified when a U.S. Environmental Protection Agency air monitor is re-calibrated.
As some of you know, I do environmental pollution research on the side to manage my health and give back to society. To manage my everyday life, I get to analyze official EPA air sensor data, learning how the EPA does science — which is SO COOL.
The US National Ambient Air Quality Standards (NAAQS) is a remarkable public data system from the 90s that collects and publishes data from over 5,000 sensors across the country. Some of these are permanent sensors; others are mobile sensors placed at the fence-line of an industrial facility. These measurements support:
- evaluation of whether a geographic area or industrial facility is in compliance with the Clean Air Act
- emergency response & public knowledge during events like wildfires or dust storms
- evidence in legal cases
Real-time pollution monitoring is hard to do reliably. Last month, for example, I noticed that a little moth had grown a chrysalis inside one of my personal sensors, distorting the pollution measure. I had to decide — do I throw out a month of data or add a “moth adjustment” to my records? Complexities like this one are why the the EPA only has around 5,000 active monitors: precise measurements are expensive and monitors need regular testing/maintenance.
How should democracies handle real-time data? Timely air quality data is a matter of life/death, including mine. It shapes millions of dollars in industry costs. It also directs tax money (via government cleanups). Since we live in a democracy, changes to the Air Quality Index (for example) or re-calibration of a sensor can’t just be done by scientists when they see a problem. Changes need to be to be debated.
For example, last year the EPA adjusted the math to calculate the Air Quality Index (after an extensive public consultation). This was great news for me, because I’ve been puzzled by my own health effects at concentrations of pollution that the EPA used to consider safe. Then in 2024, the EPA adjusted the Air Quality Index to better match the state of public health science. Like everyone else (I hope), I had to spend a weekend changing the code of my sensor network to re-calculate all current, future, and past measurements.
Not everyone thinks that updating the air quality calculations was a good idea, even if it will save lives. Lobbyists for manufacturers said that it was too burdensome. Attorneys general of 24 states agreed. After they lost the debate in the public comment period, they sued the EPA to return to standards that would allow higher levels of pollution. That’s their right. In a working democracy with effective governance, legal attacks on a regulation are positive signs that the regulation is consequential.
For public debate in a democracy to work, people need to know about the debate. And the EPA has built that debate into their air quality monitoring software. This diagram is the schema for defining who needs to be involved and/or notified when EPA scientists propose a corrective action to a sensor measurement. It includes federal and regional authorities, tribal nations, and many other entities who need to know about (or approve) small changes with big potential consequences.
I understand — many people will look at this and see needless bureaucracy. Some industry lobbyists will claim publicly that it’s a needless maze while seeking privately to make it as complex as possible to delay accountability. Advocates will rightly decry how slow it is to wade through endless legal battles for decades while whole generations are facing irreversible damage to our health. And they’re right. At the same time, this messy, complicated democratic system, supported by reliable sensor measurements, prevents 230,000 deaths per year from particulate pollution alone.
Ultimately the Clean Air Act is democracy — built right into science and the software of regulation, and it’s beautiful.
If you want to learn more about the science, politics, and software of environmental data, I encourage you to follow the work of the Open Environmental Data Project. If you want to see new innovations in how advocates are using science to create change, check out the Central California Environmental Justice Network, who I have written about for Global Voices. If you’re curious about inequality and injustices in the sensor network, a good place to start is this excellent ProPublica report on the politics of where sensors get placed (and where they don’t). There’s a new study in JAMA about this as well (Thanks Neil!)
And if you want to meet other community/participatory scientists who are working to broaden access to data and justice, I hope to see you at the annual conference of the Association for Advancing Participatory Sciences in Portland this May!
