Mechanisms, with the maths attached

Each supporter has a conviction that grows toward a ceiling while support is held and decays once it is withdrawn. Total credit for a proposal aggregates supporters by square root, so a hundred small supporters count for more than one large one. The threshold a proposal must clear scales with the size of what it asks for, which is what keeps a small ask cheap and a treasury-draining ask expensive.

The parameter that matters is the half-life. It sets how quickly the system can change its mind, and choosing it is a governance decision rather than a technical one: fast enough to correct a mistake, slow enough that the treasury is not steered by a bad week.

Delegated weight is treated as a solute crossing membranes rather than a proxy handed over whole. Weight flows along edges of the community graph, transitively, and only within the domain it was given for, so trusting someone on infrastructure does not hand them your say on grants. Streams are continuous: weight accrues per millisecond rather than in discrete grants.

Viscosity is the same conviction half-life used for funding, which is what makes the two mechanisms one system instead of two. A concentration cap limits any member's share of total power, and the model reports concentration each tick as a health signal — if power is pooling, you can see it happening rather than discovering it at a vote.

Tokens are minted and burned against a reserve along a published price function, so supply absorbs part of any demand shock instead of price taking all of it. A tribute on each purchase is routed to a common pool the community governs, and an exit tax on sales makes short-term round-tripping unprofitable.

The design question is the split. Too small a tribute and the commons is not funded; too large and nobody joins. The model exists so that argument can be had over a curve rather than over opinions.

Funding is drawn as sources, funnels and outcomes on a canvas. Each tick, the simulator moves value along the graph and checks that everything in equals everything out — a funnel that fills past its overflow threshold spills to the next one down rather than quietly absorbing the difference.

Drain rates and overflow thresholds are the policy, and changing them is itself a proposal gated by conviction, so nobody adjusts the shape of the flow unilaterally. Outcomes settle against attestations rather than self-reporting. It is flow funding in the old sense — money passed onward by the people closest to the need — with the accounting made explicit.

Instead of treating each financial instrument as its own product, this models them as coordinates in a basis-axis space: eight curves — par, residual, contingent-binary, mutual credit, bonding, LMSR, parimutuel, scalar — composed with a set of morphs, and settled by attestation engines ranging from a peer signature to a zero-knowledge ring proof.

A route planner finds multi-hop paths across the space, so a claim denominated in one instrument can be settled through another, and a conservation auditor checks the books afterwards. The point is not financial novelty. It is that a community should be able to issue the instrument its situation calls for without adopting a different platform each time.

Every item is a small content-addressed document — a holon — that can carry a reference to its own definition. When a peer receives a kind it does not recognise, it resolves that definition and installs what it needs to interpret the thing, rather than falling back to a generic placeholder. Ceptr stated this as a requirement years ago; closing the loop is mostly a matter of refusing to special-case any one document type.

The structural payoff is that dozens of unlike collections reduce to one store plus an axis, links and a lens: order is a position on an axis, relations are edges rather than tables, and a view is a projection. Time, geography, capability and identity are all just axes. That is what makes exit credible — your data is not shaped by the application that happened to create it.

Four strategies drawn from how a forest handles drought — local collection, subsurface storage, living filtration and a network that can flow in either direction — sized against the demand the Georgian Bay pipeline was scoped to meet. Each strategy has a working precedent and a named pilot site, and the permitting path stays inside Ontario's existing framework.

The model is a supply-and-buffer balance rather than a hydraulic simulation: what each strategy contributes, what it costs, and how much of the shortfall remains. It won the 2026 Biomimicry Commons B-Prize, which funds the next stage of it.