People Have the Power — How Volunteers Pushed Government into the 21st Century

On Tuesday nights, in a room we didn’t pay for, I watched a dozen people give away the most valuable thing they had. They were software engineers, data scientists and one patient designer. Most of them had day jobs that paid them well to build software for companies that already had everything. Then they came here, after work, to build software for their community.

The idea that put us in that room belonged to a tech publisher named Tim O’Reilly, who in 2009 had a thought that sounded simple and was therefore radical: what if government were a platform?¹ Not a vending machine that took your taxes and dispensed services, but data, tools, and APIs that anyone could build on, the way a generation of companies had been built on top of the web. Open the plumbing, he argued, and a thousand things nobody at City Hall had thought of would bloom on top of it. The idea spawned a book, a circuit of conferences, and eventually a nonprofit called Code for America.

Code for America started with a fellowship: take talented technologists, embed them in a city for a year, see what they build. It worked well enough that a question followed it around: what about everyone who wanted to do this but did not want to quit their job to do it? So in 2012, on the back of a $1.5 million grant from Google, Code for America launched the Brigades — local volunteer chapters for people who would give a few evenings a month to their community but were not going to leave their careers to work in government.² One of the first chapters was in Boston.

At first we called it civic hacking. A TSA agent once asked me, suspicious, about my National Day of Civic Hacking sticker, “are you a hacker?” I had to explain it was not about breaking into corporate systems to steal data or demand ransom. It meant almost the opposite: finding a clever, usually unsanctioned way to solve a public problem without waiting for anyone’s permission.

The best example I know happened in February 2015, in the middle of one of the worst winters Boston has ever had. The MBTA had effectively stopped working. Code for America had an event that weekend called CodeAcross; in other cities it got cancelled for weather. Boston ran it anyway. Three coworkers from a company called Panorama Education had spent that whole miserable week texting each other about which trains were actually moving, because the official alerts were useless, and somewhere in there one of them — Geoffrey Litt — said the thing that turns a private annoyance into a public tool: what if everyone in Boston could do this together? They built it that weekend and called it MBTA Ninja. You opened the site, picked your line, and saw what other riders were reporting in real time: delays, a disabled train, a car so crowded you should just wait for the next one. You could confirm someone else’s report or wave the all-clear. A Waze for a transit system that had stopped telling people the truth. Within three days more than ten thousand people were using it, and it was routinely surfacing problems five and ten minutes before the T did. Nobody put a single new sensor on a single train. The sensors were the riders. The network was already there; MBTA Ninja just gave it somewhere to connect.

That’s the part of the story I love. Here’s the part the conference talks skipped: many things we built never fully landed. EnerSave was going to help Cambridge’s hundred biggest apartment buildings stop bleeding heat, chasing a five-million-dollar energy prize across hundreds of Tuesday nights; its repo was archived, read-only, in 2021. Cliff Effects shipped something real and good — a calculator so a Project Hope caseworker could show a client whether a raise would quietly cost her more in lost benefits than it paid — and it still runs today, known to almost no one. MyCity launched with the City of Boston, a press release, and trash days you could ask Alexa about; the city’s own digital lead later conceded that hardly anyone used it.

And MBTA Ninja? It worked. Then the MBTA learned to track its own trains — track circuits, a real-time API, countdown clocks on every platform — and the Ninja wasn’t needed anymore. That’s the rarest fate of all: not failure, but a success so complete the institution finally builds the thing itself, and the need you were filling quietly closes over you.

We changed the mantra to “build with, not for.” Instead of engineers parachuting in with an app nobody had asked for, the good projects started to come the other way — from a caseworker, a city department, a nonprofit that actually knew where the problem was. And once you’re working with someone instead of for them, the deliverable stops being the software. It becomes the relationship. A city agency gets to pilot a risky idea cheaply; its staff gets to watch modern software get made up close; and the next time, they call you before they write the RFP. You don’t ship trust. You earn it.

Which is how, one winter five years later, two strangers handed us a website.

In early 2021, Massachusetts had vaccines and no usable way to find them; the state’s own site would make you fill out a long form just to tell you, at the end, that the location was out. Two people, Zane Stiles and Kunal Shah, built the opposite of that on their own — a site called VaccinateMA that showed you where the doses actually were, up front. And then it worked, which turned out to be the problem. It got big. Big meant real people depending on it, and real people depending on it meant liability, and liability is a heavy thing to carry on a side project you started out of conscience. So they did the responsible thing: rather than carry it alone, they brought it to Code for Boston — and stayed on as founders of a team that could carry what two people no longer could. We kept it alive — eighteen volunteers phoning clinics every day to verify availability across four hundred and fifty locations, feeding what they found into a database powering our tool — straight through the months when an open appointment was the single most valuable piece of information in the Commonwealth. At its peak the site was taking six thousand hits an hour. There was a morning when the commonwealth’s own vaccine finder buckled and ours didn’t, and the reason was unglamorous: someone had spent two weeks moving the thing onto AWS Elastic Beanstalk. Sometimes someone with experience making good technical choices is the difference between success and failure.

Software for government was never new — agencies bought it by the truckload, mostly through fixed-spec contracts that shipped years late and wrong. What Code for America’s brigades carried in was a creed, codified by 18F and USDS: start with what people need, ship small and iterate in the open, put practitioners inside instead of outsourcing the thinking. So the scoreboard was never an app. It was people — volunteers who showed up to a Tuesday hack night and walked out building government. Code for DC’s Matt Bailey went from the brigade to the White House budget office. And it stuck where it was hardest to reverse: the forging of a new kind of vendor. Nava and Ad Hoc, founded by HealthCare.gov alumni, won prime contracts on the strength of having actually shipped, so when DOGE gutted 18F and USDS, the firms were still standing, still winning, still building in the open. You can watch the whole arc in one career — Harlan Weber, founder of Code for Boston, founding member of the Massachusetts Digital Service, now Director of Design, Strategy & Growth at Nava. The weeknight hobby became a career, then a state institution, then an industry.

So has the movement earned a victory lap?

Here’s the thing nobody tells you about victory laps: you can spend the whole race waiting for a finish line that was never there. There’s no app that fixes government. There’s no headline, no trophy, no afternoon when somebody with a title walks into the room and tells you that you were right. I waited for that day for years. It isn’t coming, and the reason it isn’t coming is the work does not end. But that does not mean we did not win.

Tim O’Reilly said government would become a platform. He was right. He just pointed one layer too low. It was never the data. It was never the APIs.

It was us.

So take the lap. We earned it.