Data Economy · Mechanism Design · Civic AI · 2026-09-20

Mechanisms, Institutions, and Data Soil

The games people can genuinely play and share matter more than the technology being built.

I attended a meeting in New York about data. There was a great deal worth carrying away; because of the nature of the meeting, I will share only part of it — and only my own reading of it.

Cooperative games

Silicon Valley has long played winner-take-all. But many open-source advocates — including the mechanisms in Plurality — play the opposite, cooperative game: Bell Labs and UNIX, an operating system built from many small, cooperating parts. You could call it the "architecture of participation". The web and many open-source projects, whatever the licence, succeeded where that quality was present.

Mechanisms and institutions

The world moves forward through the design of two things: mechanisms and institutions.

Which comes first? There is no single answer. The Royal Society formed in response to the rise of experimental science, and peer review took about 150 years to emerge — often the institution forms first and feels its way toward the mechanism. Open source ran the other way: from patch to git, the mechanism changed and the institution followed. The two carry each other forward. One honest sentence is worth keeping: we can invent mechanisms, but if the institution has no power to make them stick, they do not work.

"It" is not one thing

Nor do I treat AI entirely as an "it" — not a separate entity, and not a single use. AI is a choice, not a necessity; RSI is not the only path, and "replacing people" is not the only way to imagine it; it is Silicon Valley that has flattened it into one. Science from Kepler and Galileo forward meant setting aside prejudice and following the evidence; the Turing test pulled the standard back to whether you can fool a person — precisely what science was designed to refuse. The industry's dogma is often that the illusion of the model being alive must be maintained; call a model an amalgamation of people's efforts — like a large Wikipedia — and you breach a taboo, while saying "it might destroy humanity" stays within orthodoxy. I think what has stalled may be a dislocation we are living in. I like how Plurality describes science — from its emergence to its compatibility with different disciplines — and I think AI will be the same.

Piercing the illusion

I have never liked communicating through doom. On this US trip, from DC to New York, I could feel that the current resentment toward AI and data centres comes from the public, not from manipulation. Put another way, this may itself be a kind of piercing the illusion — and seeing what is real. Put simply: even without AI, it is still precarious, still heading toward destruction — climate, environment, justice. These problems were always here. What I keep turning over are moral interiority and linguistic justice; and on both, we can see technology loosening what had been stuck for a long time.

Data is soil, not oil

On data and creation: they should not be merely extracted — if creators lose the economic incentive, capital only concentrates further. Roughly, YouTube's near-halving feels closer to a workable fairness than "everything to whoever owns the compute". But data is not oil; it is soil — not a stock bought once, but a flow that yields only with tending. So the real question is not only how to divide, but — starting from more empathy — who tends, and who can check the harvest; and at that step we do not need perfect attribution — measurement the contributor can check is enough to make reuse observable, and only then to speak of revenue. How to divide and how to attribute is the mechanism still to be designed together.

Fifty-five years

And then the Wright brothers: fifty-odd years from the first flight to flying being something anyone could do (first flight December 1903; the Federal Aviation Act creating the FAA, August 1958 — 54 years and 8 months). Once everyone could fly, we stopped remembering who built the plane and started caring about safety, operations and takeoffs. As I read it, about the environment too.

So there is a great deal to think about and to do right now — let us not point it the wrong way (a reminder to myself, too :).