Matt Stempeck is a global civic tech researcher, activist and curator of the Civic Tech Field Guide, an open repository of over 10,000 projects that use tech and data to enhance democracy.
He previously served as director of civic technology at Microsoft, where he worked to leverage the company’s platforms and products to achieve civic good in cities. He was also director of digital mobilisation for the Hillary For America campaign, and a Technologist-in-Residence at Cornell University, where he launched public interest tech impact fellowships that embedded PhD students in public interest organisations.
Currently, Matt is the Director of Democracy and Technology at the Brussels-based Evens Foundation.
He holds a bachelor’s degree in government and politics from the University of Maryland and a Master of Science in Civic Technology from MIT.
Matt spoke to indianexpress.com on the state of civic tech, the challenges faced by civic tech initiatives and startups, and the interplay between tech and democracy. Edited excerpts:
Venkatesh Kannaiah: Tell us about your journey to becoming a civic tech researcher and activist.
Matt Stempeck: I call myself a curator of civic tech, though I do have an activist side, and research is a big part of my work.
Story continues below this ad
I’ve always been excited about tech and using that energy for social good.
Early on in the US, I connected with groups exploring how the internet could help bring more people into politics and reduce money and corruption in politics. That led to my understanding of how technology could strengthen democratic engagement.
As a graduate student at MIT, I built Lazy Truth, one of the first tools aimed at fighting disinformation. It was a Gmail plugin to fight fake news. I scraped fact-checking websites, with permission, and created one of the first APIs for fact checks, enabling the system to match forwarded emails with existing fact checks. Users could simply click “Ask Lazy Truth” within Gmail to get an answer.
After MIT, I became interested in giant tech platforms. I felt that if civic tech could be built into those platforms, the scale of impact would be far greater. That civic-tech handshake between institutions and platforms fascinated me. That was one of the main reasons I joined Microsoft — to see whether those large-scale systems could connect more directly with civic participation and public good.
Story continues below this ad
Microsoft had a civic tech team and worked with groups like DataKind on issues such as safe streets, data science for social impact, and partnerships with mayors’ offices around open government data.
Around the same time, the civic tech field itself was becoming more clearly defined. With colleagues, we felt the need to map the field. Other people had built catalogues before, but we wanted to start from what actually existed, not from our assumptions. And that led to the Civic Tech Field Guide.
Venkatesh Kannaiah: Do civic tech projects finally evolve into GovTech solutions?
Matt Stempeck: There’s a big overlap — it’s really like a Venn diagram. Many governments set up civic tech platforms to engage the public, so the line between civic tech and GovTech is often blurred.
Story continues below this ad
For us, with the Civic Tech Field Guide, the distinction was always about power. We weren’t as interested in GovTech that simply helps you renew your driver’s licence or complete an administrative task. That’s useful, and the government should absolutely work well, but it doesn’t really shift power.
So Civic Tech, for us, was about shared, collective decisions rather than things that only improve your individual life. A lot of startups build technology that makes your personal life easier, and that has value. But we were focused on democracy problems — the kinds of issues where your outcome depends on everyone else, not just you.
That was the line we tried to draw. If it’s an urban tech project that helps you personally get through traffic faster, that’s not a collective challenge, it’s just convenience for you. But if it helps reduce traffic deaths and crashes across the city, that becomes a civic issue.
Some civic tech projects do evolve into GovTech vendors — they end up selling their technology to governments. Many of the ones I’ve seen over the years eventually get acquired by larger GovTech companies, and sometimes they disappear altogether.
Story continues below this ad
We track campaign tech and advocacy tech too, even though they aren’t strictly civic engagement tools. But more importantly, you need political power to make civic tech work. When I researched participation platforms, I recognised that the successful ones were the ones backed by real political power.
At least in the US, many people who started in Civic Tech eventually moved inside the government to become reformers. And then something interesting happens — you begin as the outsider pushing for change, but once you’re inside, you become part of the system.
Venkatesh Kannaiah: Tell us about your Civic Tech Field Guide.
Matt Stempeck: It’s been 10 years since we started it. I worked along with leading civic tech researchers like Micah Sifry and Aaron Simpson.
Story continues below this ad
First, we wanted to bring new people into the field, and we were also seeing a lot of duplication of effort and a lack of shared learning. People kept rebuilding the same things or repeating the same mistakes.
That’s why we created sections like the ‘graveyard,’ where we document failed or discontinued projects. We also research on top of the database, and others use our data for research too.
For a regular person, it’s basically a large Wikipedia for tech-for-good projects around the world. At this point, we have more than 11,000 projects in the database, and each one is hand-tagged.
That’s an important point, especially now when AI can generate a directory in five minutes. Our category tagging has always been done by us — the curators — not just by people trying to market their own projects. That makes the catalogue far more useful.
Venkatesh Kannaiah: Tell us about successful civic tech projects.
Story continues below this ad
Matt Stempeck: mySociety in the UK helped set things in motion in the 2000s. They built tools like WhatDoTheyKnow, which became a major freedom of information platform. mySociety showed what was actually possible with these kinds of civic tech products.
In many cases, they were simply the first to create these kinds of tools, which today exist in many versions around the world. People now run versions of WhatDoTheyKnow-style platforms across very different political systems and countries.
Another project I really admire is Decidim in Spain. It’s a participation platform where you can run things like participatory budgeting, citizen assemblies, and other public decision-making processes. It’s modular, so depending on what a city or organisation needs, they can install only the parts they need.
Because the code is open, people are now building bridges between Decidim and platforms like WhatsApp, which creates really interesting possibilities for participation at scale.
Story continues below this ad
There is the Voting Information Project, run by TurboVote in the US. Before they did this work, it was surprisingly hard to find out online where to vote. That data simply didn’t exist in a usable way because elections are managed locally, often at the county level.
So they did the hard work of collecting that information and turning it into an API. Once that API existed, companies like Google, Facebook, and dozens of startups could then help users find where to vote, what was on the ballot, and other critical election information.
That’s a great example of how a relatively small amount of technical work can create enormous public impact. A lot of product teams are willing to include civic features, but you have to make it easy for them, you have to give them the API that already works.
There is Code for America, especially in its early days, where they created local meetup groups in cities across the country for people to work on civic problems. A lot of it looked like hackathons, and of course, some projects were more impactful than others.
In East Africa, people combined citizen science with activism. Residents collected data on water utility bills, tested water quality using sensors, and used that information to advocate for better access to safe water. It brought together science, accountability, and civic action in a practical way.
Venkatesh Kannaiah: Tell us about civic tech projects that did not succeed and why?
Matt Stempeck: With startups, either the founder or a journalist will write about why something failed. We always try to include those in our Civic Tech Field Guide because our goal isn’t to shame projects that didn’t work, we want to help people understand what didn’t work.
One category is niche social networks, especially civic or political social networks. These are platforms built around very specific themes, like civic engagement, where you expect people to leave mainstream platforms and use your dedicated network instead.
That destination strategy almost never works unless you have enormous funding, and even then, it’s difficult.
There are disinformation projects, which became a trend after 2016. There were hundreds of them. I tracked more than 500 with Code for All. But very few actually got to the point of fighting disinformation.
Most of them were just cataloguing disinformation — building databases of false information — without ever reaching the harder part, which is actually intervening with the public and changing behaviour.
Now we’re seeing something similar with AI-built civic projects. Because AI coding tools make it cheap and easy, people outside the field are launching civic tools every day. Everyone is building the same thing: explain legislation, track legislation, wrap an AI layer around congressional data. I’m not sure the public wants a legislation explainer at that scale, and if they do, tools like ChatGPT can already handle some of that.
That also raises a deeper question about impact. Do we really expect that simply creating a website showing political corruption will end political corruption? That would be unrealistic. Civic tech is necessary, but it’s not sufficient.
Formats matter too. Most browser extensions and plugins don’t survive. Part of that is because you’re asking users to do something technical, most people don’t even know how to install a browser extension.
Venkatesh Kannaiah: Why has civic tech as a category not taken off? Or am I wrong?
Matt Stempeck: It depends on how narrowly you define civic tech. If you mean that a website alone will end political corruption, then no, that hasn’t really worked.
And maybe part of the problem is that we were overselling it. Maybe our expectations were too high. At the same time, some things absolutely have worked. Take Vote.org in the US. They’ve helped around 40 million people get information about voting.
Or in Brazil, there’s a local version of Decidim used for the National Participation Programme — Participa + Brasil. Within just months of its launch, around 1.5 million Brazilians were using it to tell the government what they wanted to happen.
A lot of civic tech has been built on the theory that people are simply missing information, and that if we provide the right information platform, the problem will be solved. Since computers are good at organising information, the assumption was that information itself would create change. But time and again, we’ve seen that information alone isn’t enough. That’s rarely the full story. Usually, the barriers are power, incentives, and political structures, not just access to information.
Venkatesh Kannaiah: Tell about your work on how tech platforms are deploying users as lobbyists.
Matt Stempeck: User lobbying is common in the sharing economy or gig economy companies like Uber and Airbnb.
They launch their services in a city and, even if the legal status is unclear, build up a large base of customers and people making money on the platform, and then by the time the government moves to regulate them, they already have a political constituency. At that point, regulation becomes a political fight, because they now have users with a direct stake in the outcome.
I noticed that when regulators pushed back, these companies would directly mobilise users through the product itself to fight for their legality.
On the positive side, there are what I call civic features. Large tech platforms like Google or Facebook help people register to vote, find polling locations, or access election information.
If Facebook can help millions of people register to vote, or Google can tell users where to vote, that scale matters. It would be unrealistic to ignore that. That’s why I call it a civic feature — it’s a civic action pathway built into a platform people are already using every day.
At the same time, we also have to think about the long-term game: building real alternatives. That’s why regulation matters so much, especially what Europe is trying to do around platform regulation. It’s important that those efforts survive and aren’t weakened. And alternatives like the AT Protocol, which powers Bluesky, are exciting because they represent actual usable alternatives. The key is that alternatives have to be real, they have to be easy to use, practical, and attractive enough that people will actually move.
Venkatesh Kannaiah: What does AI add to the mix of tech and democracy?
Matt Stempeck: One simple way I describe it is: civic tech is the technology we build for democracy.
Then there’s a much bigger conversation: the broader tech sector’s effect on democracy. That includes issues about platform power, regulation, AI, and how large technology companies shape political life.
Obviously, I’m excited about using what technology makes possible, including AI, for democratic participation and public good.
As an American, one thing that concerns me is how much US foreign policy increasingly seems aligned with what Silicon Valley wants, like pushing back against European regulation of large tech platforms. That’s a little frightening in terms of tech’s influence on democracy.
The same tension exists with AI. If the same giant tech companies are the ones building our AI systems, I don’t think AI is likely to be our democratic saviour.
We are also building engines that can generate endless personalised versions of reality. AI can take that even further and can reinforce exactly what you want to hear. If everyone ends up with their own private version of truth, that becomes a very serious democratic problem.