Two kinds of air existed in the same San Francisco

2026-10-01 original Society & Ideas

Bikes and delivery cycles

More than ten years ago, I ran experiments in San Francisco where we attached air sensors to bikes and delivery cycles and sent them out around the city. The idea was that measuring while moving could capture how polluted the air was, street by street. We got lucky: we won a business competition, showed it at an exhibition, and even raised funding.

Around that same time, a company called Aclima in the Bay Area had teamed up with Google and started mounting air sensors on Street View cars. We turned out to be drawing nearly the same picture, in nearly the same city.

What you can’t see from measurements taken a few times a year

The problem, as I saw it, was how few official monitoring stations there were. Their number was extremely limited, and a high-traffic location like Highway 101, for example, might get measured only a few times a year at most. That cannot capture the real exposure — the air that people living or commuting there actually breathe every day.

When we ran the sensors ourselves, the difference was stark. Within the single city of San Francisco, there was a San Francisco with some of the cleanest air in the world, and a San Francisco whose air quality rivaled Mumbai or Beijing at the time. We could pin that down precisely.

That breathing polluted air continuously harms the body has been clear for decades. In Japan too, pollution incidents including Yokkaichi asthma drove large-scale studies into the relationship between air pollution and respiratory disease, and the evidence had already accumulated by the 1970s. For particulate matter, the Harvard Six Cities Study in the US showed in 1993 that cities with higher concentrations of fine particles had higher mortality rates. Fine particles like PM2.5 are small enough to penetrate deep into the lungs and even reach the bloodstream. Their effects are not limited to the respiratory system — they raise the risk of ischemic heart disease including myocardial infarction, stroke, chronic obstructive pulmonary disease (COPD), and lung cancer. The International Agency for Research on Cancer (IARC) classifies air pollution and particulate matter as carcinogens, and the WHO estimates premature deaths from outdoor air pollution worldwide at 4.2 million for 2019 alone. For the effect on lifespan, the Air Quality Life Index (AQLI) from the University of Chicago’s Energy Policy Institute uses the relationship that a 10 µg/m³ increase in long-term PM2.5 concentration shortens average life expectancy by roughly one year.

Ambient (outdoor) air pollution — A WHO fact sheet on outdoor air pollution and health effects — who.int

If the air you breathe varies this much depending on where you live, that struck me as a problem worth bringing to light, and one with real social significance.

And yet, this was remarkably hard to get across to people around me.

Looking back honestly, that period remains a bitter memory for me.

Insurance and real estate

Since this was a business, someone had to pay for it. The monetization paths I found were “insurance” and “real estate.”

If exposure affects health and lifespan, that should be risk information for insurance companies. If the air quality of a given location is quantifiable, that should also be information for valuing real estate. I still don’t think the logic was bad, but it didn’t go particularly well.

Science caught up neatly

What’s interesting is that science afterward backed up exactly what I’d been trying to say.

In 2017, Joshua Apte and colleagues at the University of Texas drove Street View cars repeatedly over 30 km² of Oakland, California for a year and produced a pollution map at 30-meter resolution. Even within the same city block, concentrations differed by a factor of 5 to 8, and the pattern held stable throughout the year. A study the following year showed you don’t need to drive every single road repeatedly — combining measurements from a subset of roads with a model can produce a reasonably good map.

Mapping air pollution with Google Street View cars — A writeup of research that mapped Oakland’s air pollution at 30-meter resolution using Street View cars — apte.caee.utexas.edu

Within the same city, there is completely different air. What I had worked out in San Francisco took the form of a published paper.

Google subsequently announced in 2018 that it would equip 50 Street View cars with air sensors and expand the effort worldwide, running them in Dublin, Hamburg, Bengaluru, and elsewhere. But “Air View+,” which it launched in India in November 2024, uses not cars but a network of fixed sensors: local companies install sensors across more than 150 cities, research institutions handle verification and calibration, and Google’s AI fuses the data with government data and other sources. That the company with arguably the greatest capacity to drive the sensors itself stopped driving them made me nod in understanding.

Air Quality insights in India, with Air View+ — Google’s Air View+, which expanded a network of fixed sensors to more than 150 cities in India — blog.google

Who ended up paying

So who, in the end, paid for mobile measurement data?

Lining up Aclima’s recent work reveals the answer. A statewide mobile measurement campaign in New York State in 2022. Near-24-hour measurement of a Washington, D.C. neighborhood for about six weeks in 2024. And a $27 million contract from California’s Air Resources Board (CARB) under which, starting June 2025, cars have been driving through 64 communities across the state. The scale: 950,000 miles driven, 5.2 million residents covered. From autumn, full-scale mobile research laboratories from research institutions entered each community in a second phase, investigating in detail the locations Aclima’s cars had identified.

Sensor-Equipped Vehicles Hit the Road for 64 California Communities — Aclima’s announcement of mobile measurement starting in 64 California communities — aclima.earth

The money came from government, under a policy of addressing areas where the burden of pollution falls unevenly — what in the US is called environmental justice. In California’s case, the budget comes from a framework that channels cap-and-trade revenue into climate measures.

The idea that the air you breathe differs depending on where you live, and that this affects your health, came to be recognized as an issue government should budget for and tackle.

That is exactly the story I struggled to get across to people ten years ago.

Start small, run long

Looking back now, I feel a real fondness for the days when we strapped sensors to bikes and cycles and sent them through a hilly city. Including the frustration of not being understood, I think that was, in its own way, a good time.

What I remember about Aclima is that its early funding was around $3.3 million (¥500 million), which honestly was quite small for a Bay Area startup. Looking at the public record, its first major round, a $24 million Series A, came in 2018 — a good while after founding — and the investors had a strong impact-investing character, including the Schmidt family’s foundation, Emerson Collective, and Kapor Capital. In 2020 it raised a $40 million Series B and simultaneously became a Public Benefit Corporation, and in 2025 it received $13 million in contract financing to execute the CARB contract.

Rather than raising a large sum and expanding all at once, they picked up government contracts one at a time, secured the funding to carry out each as it came, and kept running for more than ten years.

Frankly, I think that’s remarkable.

Some people took the story I struggled to get across to others and carried it, over more than ten years, to a point where it was understood. The Aclima team, who started running around the same time I did and kept running on funding that was hardly abundant. The researchers who used enormous volumes of driving data to show, in a published paper, the differences within a single city block. And the people in government and local organizations who heard about it, allocated budget, and built measurement plans together with the communities.

The question I let go of partway through reached its answer properly in their hands. Before the frustration, what comes first is something like happiness.

References

This piece was conceived and directed by Kuzuryu, and written with AI.

Originally published in Japanese at https://clazytech.com/2026/10/1842/. Translated with LLM assistance and reviewed before publication.