Civic Tech That Lasted.

Civic tech has a famous failure mode. A team builds something useful for one city, the launch gets covered, the grant runs out or the administration changes, and two years later the repository is archived and the domain is parked. The software was fine. Nothing else was.

A few projects broke that pattern. They're worth studying because they didn't survive on brilliance. They survived on structure, and the structure is boring enough that it rarely makes the launch post.

Start with Madrid and Barcelona, because their stories are linked. Madrid's city government began developing a participation platform in 2015, which became Consul. Barcelona's first participation site in 2016 ran on a fork of Consul before the two teams split and Decidim launched as separate software in early 2017. Both projects then made the same move at almost the same moment. The Consul Democracy Foundation was founded on 23 March 2019, an independent non-profit based in the Netherlands, to keep developing the software. A month earlier, Decidim's community had founded the Decidim Free Software Association on 16 February 2019, and Barcelona's agreement with it ceded the maintenance of the code base to the association, which also holds the registered trademark.

The timing matters. In June 2019 Madrid's city government changed hands. Decide Madrid kept running, but the participatory budget fell from over €100 million a year to €50 million every two years. If the software had belonged only to that city hall, a change of government would have been a change of roadmap, or the end of one. Instead, Consul now reports use by more than 250 cities and organizations, and Decidim says more than 500 organisations use it. Neither project depends on one mayor's survival any more.

Moving the code out of city hall doesn't mean leaving public money behind. In 2021 the Catalan government, Barcelona City Council and the Decidim association signed an agreement that establishes a financial contribution framework meant to reinforce the project's sustainability. Governance moved out. Funding still flows in. Those are different questions, and the durable projects answer them separately. Decidim also binds contributors to a social contract, its "Code of Democratic Guarantees and Democratic Collaboration", which every community member has to endorse. That is governance written down before anyone needs it.

mySociety in the UK took a different path to the same place. It was founded in 2003 and still runs TheyWorkForYou, FixMyStreet and WhatDoTheyKnow. When it launched WhatDoTheyKnow in 2008, the focus was the UK, and people abroad quickly wanted copies. mySociety's own verdict was blunt: "the original codebase from WhatDoTheyKnow.com wasn't very replicable." So it rebuilt the code as Alaveteli, a generic freedom-of-information platform. Its deployment page now claims 25+ jurisdictions and more than a million requests, from Ukraine to Uruguay. By 2012, FixMyStreet code was running in Norway, India and Brazil, and mySociety admitted it doesn't always know when someone picks up its open-source code. That is what successful reuse looks like from the inside.

The money side is less romantic and more instructive. mySociety's trading subsidiary, SocietyWorks, sells services to public bodies and supports more than 35 councils and other public sector organisations, FixMyStreet Pro among them. On mySociety's own funding page: "In FY 2025/26 SocietyWorks profits contributed 47% of mySociety's overall income." The rest comes from foundation grants and individual donations. Councils paying for a maintained product is not the story anyone tells at a conference. It is why the free version is still online.

How mySociety pays for free tools

• Share of income from SocietyWorks profits, FY 2025/26 47%

SocietyWorks sells maintained services, FixMyStreet Pro among them, to more than 35 councils and public bodies. The rest of mySociety's income comes from foundation grants and individual donations.

CKAN, the open-data portal software, is older still. It started in 2006 under the Open Knowledge Foundation, powered data.gov.uk by 2010, and became the base for the US Data.gov and Canada's portal in 2013. In 2019 the foundation's board backed a joint stewardship proposal from Link Digital and Datopian, two of the longest-serving members of the community, while the foundation stayed on as "purpose trustee." Today CKAN's assets are held in trust by the Open Knowledge Foundation, with the two companies as co-stewards. Handing day-to-day stewardship to commercial vendors is a real trade-off, and not everyone will like it. But the split is clear: a non-profit owns the purpose, the people who earn a living from the software maintain it.

Taiwan's g0v shows the community version of durability, with its own limits. It was founded in 2012 by civic hackers frustrated with government digital services, who built "forked" versions of government websites that were often more popular than the originals. g0v describes itself as a decentralised community that runs a large hackathon once every two months, with the venue, food and money raised by community members. Its deliberation project, vTaiwan, held detailed deliberations on 28 issues, 80% of which led to legislative action. The same account is candid that vTaiwan's engagement declined after the pandemic, citing the intensive volunteering required and the lack of any mandate for government to respond. g0v the community lasted. One of its best-known projects struggled. Volunteer energy is real infrastructure, but it isn't a budget.

For contrast, look at what happened when the money moved first. The Sunlight Foundation, once central to US transparency work, said in September 2020 that its role was "no longer essential to its original central mission", and handed its name, intellectual property and records to the Internet Archive and Harvard's Berkman Klein Center. In 2023 Code for America sunset its Brigade program and ended fiscal sponsorship of volunteer brigades, saying it had become increasingly difficult to raise multi-year investments for a national volunteer network. Some brigades were already independent and carried on. Others had to find a new legal and financial home in a hurry. In both cases the work depended on one central organization's fundraising, and when that changed, everything downstream changed with it.

Put the cases side by side and the pattern is plain. The code lives somewhere a single election can't reach: an association, a foundation, a trust. The software is generic enough that a second city, or a second country, can run it without asking permission. Someone pays for maintenance in a way that recurs, whether that's councils buying a service, a government signing a contribution framework, or companies whose business depends on the platform. And there's a community with written rules about who decides. None of these are exciting, and none of them make it into a launch post.