The bet
The OpenGov Foundation was born from a victory. During the 2011 fight over SOPA and PIPA, an alternative bill, the OPEN Act, was drafted with public input online, and its backers called it the first crowdsourced legislation in history. Seamus Kraft, who had worked in Congressman Darrell Issa's office, co-founded the foundation with Issa in 2012 to turn that one-off into infrastructure. Its tool, Madison, let citizens read a bill, comment on specific provisions, and suggest edits, so that improving a law could become a public act rather than a private one behind committee doors. The bet was that if you made legislation legible and markable by ordinary people, lawmakers would let the public in.
Peak
The idea traveled remarkably far. Madison was adopted by Washington, DC, by the federal government of Mexico, and by the United Nations, and in 2015 the White House used it to crowdsource its Public Participation Playbook. Bloomberg Philanthropies and the Sunlight Foundation used it to develop municipal open-data policies. For a stretch, Madison was the marquee example of crowd-law, cited everywhere as proof that citizens could help write the rules that govern them.
What worked
The tool was excellent and the mission mattered. Madison helped produce a real crowdsourced bill, and it made the daunting act of reading and editing legislation genuinely approachable. It was open-sourced, so others could build on it, and it was taken up by institutions as serious as the White House and the UN. And Kraft's later honesty about the toll the work took, the burnout he wrote about openly, gave the field something it rarely gets: a truthful account of what sustaining a civic-tech nonprofit actually costs a person.
What failed — the mechanism
Madison solved the citizen's half of crowd-law and could not solve the legislator's. It made public markup easy, and impressive institutions adopted it, but adoption mostly meant a pilot or a single showcase exercise, and the officials who could turn citizen edits into changed law seldom did. The participation, elegant as it was, tended to end in a comment thread rather than a redline in the statute, and a tool whose whole value depends on the responsive side showing up found that side present for the announcement and absent for the follow-through. Beneath the product, the foundation was a grant-funded nonprofit with no recurring revenue. It open-sourced Madison and stopped developing it in 2017, pivoted to a constituent-communication tool in search of a sustainable model, and wound down around 2019, its founder describing the burnout candidly. The idea was right. The counterpart with power, and the money to keep waiting for it, both ran out.
What Agora takes from it
OpenGov Foundation is DemocracyOS's American cousin: a genuinely good way for citizens to engage real legislation, undone by the official side's non-adoption. Agora refuses to build a tool and then wait for officials to notice it. It seeds the official side by auto-aggregating officials' public records, and it anchors participation to their actual work, so the platform is not an empty room on day one and has a reason to exist before any official chooses to take part. The foundation's arc, grant-dependent and ending in burnout, is the business-model void with a human face. It is why the question of who pays has to be answered early rather than deferred, and why Agora is built so that participating stays free instead of running on grants and the willpower of the people who believe in it.
- Americans ElectA $35 million bet that a hidden centrist majority would flock online to nominate a third presidential ticket, undone when neither the voters nor the candidates it imagined ever showed up.
- DemocracyOSArgentina's open-source direct-democracy platform that tried to make legislators vote the way citizens clicked, and when it could not get the legislature to listen, ran for the seats itself and lost.
- VotizenA verified-voter social network that made sure every voice was a real constituent, and learned that the verification which made the signal trustworthy also kept the network too small to matter.