We design economic, epistemic, and computational mechanisms for groups to reason and decide together at scale.
The world coordinates through payment rails that don’t price externalities, discourse platforms that reward heat over light, and civic systems with no memory. The mechanisms were built for a smaller, slower world. The problems they leave behind are real and human-scale; so are the returns for solving them.
Markets don’t price what they don’t see. Public goods stay underfunded; harms stay externalized. The opportunity is to make funding them profitable.
Public reasoning rewards conviction over calibration and narrative dominance over accuracy. Better incentives — for being right, and for changing your mind — would reshape how decisions get made.
Coordination systems concentrate power as easily as they distribute it. Designing against capture is part of the work, not an afterthought.
We’re building foundational assets — values-embedded vector money, collective-intelligence platforms, deliberative memory — for problems facing humanity at scale. The stakes are real. So are the returns for solving them.
We draw on mechanisms such as quadratic funding and connection-oriented cluster matching, and we publish the formal models behind our work alongside the code.
Voluntary taxation, wealth equalization, funding for public goods.
A programmable wallet that accepts payment in values-embedded vector currencies. The result is a market mechanism for funding public goods — one in which paying voluntary taxes is in the payer's own interest. We've published a preprint setting out the formal model, and a working demo.
A protocol layer for reasoned disagreement.
A discussion platform that pairs economic incentives with epistemic values. Participants stake credibility on claims; the system rewards those willing to update in light of stronger arguments. It is designed for groups that need to reason together honestly — and for minds willing to change.
AI-powered prediction for FDA drug approvals.
Litmus combines AI research tools, a curated expert network, and a structured forecast elicitation workflow to produce calibrated predictions on biotech outcomes. It is built around a core question in collective-intelligence research: how do you systematically extract what humans know that AI doesn't, through a process that is rigorous, reproducible, and fully transparent? Developed in collaboration with Elanor Davies, with support from NGI.
Civic deliberative memory.
Louie turns the public record of municipal government into something residents and officials can actually use. It ingests transcripts, agendas, and minutes, and makes them searchable in plain language — every answer linked back to the exact moment in the original meeting. Underneath sits a structured argument map: who said what, what was debated, what was decided, and what was left open. A demo of the first deployment is live with transcripts from the City of Mississauga in Ontario, with active outreach to other municipalities ahead of the October 2026 elections.
Scroll
DeltaNGI is a research organization building coordination mechanisms — protocols, incentives, and governance systems that help groups reason and decide together. We work at the intersection of mechanism design, collective intelligence, and economic infrastructure.
Systems that help groups make better decisions collectively. Markets are coordination mechanisms; so are courts, peer review, and elections. We design new ones for problems the existing ones don't solve well.
Public reasoning today rewards conviction over calibration and narrative dominance over accuracy. Our work — the Negation Game, Litmus, Louie — builds infrastructure where being right and being willing to change your mind both pay off, and where the reasoning behind a decision can be surfaced, contested, and recorded.
A protocol for reasoned disagreement. Participants stake credibility on claims, and the system rewards those willing to update in light of stronger arguments. It's designed to make intellectual honesty economically rational.
Index Wallets are programmable wallets that let people and institutions accept payment in values-embedded vector currencies. The result is a market mechanism for funding public goods — voluntary taxation that benefits the payer.
Researchers, civic institutions, funders, protocol teams, and communities trying to coordinate at scale. If you're working on a problem where collective reasoning or public-goods funding is the bottleneck, we'd like to hear from you.
A short letter each month on what we're building, what we're reading, and where we'd welcome help. Sent to investors, collaborators, and friends.