Rules the public can rely on.
The hardest lessons of a build — what never to collect, how to prove a number, how to keep a system out of a regime it should never touch — are too valuable to live in one team's heads. We write them down as standards, policy, and governance.
What we do
Governance design
Community review boards, data-use workflows, consent and retention regimes — modeled as real, logged, enforceable systems.
Standards & frameworks
Reproducibility, representativeness, suppression, and audit requirements, written so others can adopt them.
Policy & regulatory strategy
Mapping a system to the right posture — and deliberately keeping it out of regimes it shouldn't enter.
Research & briefings
Field-shaping writing for agencies, funders, and the wider public-interest-technology community.
AI governance
Where AI is appropriate, where it isn't, and the constraints that keep it from being turned against people.
A rule you can watch work.
Move the threshold. When a group is too small to report safely, the system suppresses it — automatically, every time. Illustrative, but it's how a real suppression rule behaves: enforced at the data layer, not left to good intentions.
"The strongest protection you can give people is a system that was never built to harm them — and rules that make sure it stays that way."