Versioned. v4 (2026-05-08), v3 (2026-03), v2 (2024-04-20), v1 (2024-04-04), with v0 (Lagrange) reference.
Living document. Pick a version below — latest is selected by default. Earlier drafts are preserved verbatim.
"Don't trust, verify."
The world is interconnected by data. Complex systems form intricate networks. Money moves through networks. Finance moves through networks. Banks, exchanges, OTC desks, correspondent chains, stablecoin issuers, central banks, payment-system operators, P2P markets — all of it. They share the same structure: nodes connected by rates, by trust, by regulation. But the pipes don't see each other. Each one operates blind to the others. ƒxyz holds the union.
The original whitepaper (v1 and v2 in this archive) framed this as a multiplex network the modern financial system needs to be computed as. The intuition is unchanged. What is different is two years of building.
Lagrange's USDL became Florin — a multi-currency unit of account, collateral-backed, on Solana mainnet. The LAG governance token became House of Wisdom (HoW, ℏ) — allocated and earned, not sold. The Loyalty Reward formula became the Joule emission curve, on the same intuition: contribution × time.
Serum, FTX, and TerraUSD all collapsed in 2022. The team lost capital and lost the venue the entire architecture rested on. What survived was the thesis: the graph. The trading layer was a dependency. The graph was the architecture.
The legal vehicle moved from Cayman to Poland. VASP registration is active. MiCA CASP Level 3 is in process. Members onboarded. Three contribution units deployed. The DAO + Holacracy hybrid is shipped, not theoretical.
The financial system is already a multiplex network. The same actors meet across many relationship types — settlement, contribution, governance, knowledge, identity — and the structure of one layer constrains the others. Treating these as separate problems is convenient. It is also wrong.
Padgett & Ansell (1993) showed it for fifteenth-century Florence: the Medici won by strategically segregating their multiplex ties — marriage, patronage, partnership, debt — not by maximising any single network. Mitchell (1969) formalized the same intuition for African urban kinship a generation earlier. Kivelä et al. (2014) provide the modern mathematics. ƒxyz did not invent multiplex networks. ƒxyz is applying a fifty-year body of work to financial infrastructure.
The graph is not the product. The product is what the substrate makes possible: a route that didn't exist, a price that was hidden, a contribution that became measurable, a vote that had standing.
We map all the closed networks of finance as one substrate. We run functions over that substrate — finding routes, detecting where value is out of sync, scoring contribution, weighting governance. We grow the substrate by introduction, not by acquisition. Each member opens networks the substrate alone could not reach.
We hold no member value. Members keep their own custody. The system computes, proves, and routes; it never possesses.
These are architectural exclusions, not legal hedges. Each constrains what the system can be used for and what it cannot fail at.
The identity model nests:
HumanBeing → LegalPerson → OnlineID → Persona → Member
A human can hold multiple legal personhoods. A legal person can hold multiple online identifiers. An online identity can host multiple personas. A persona is what acts in the network. Members are personas with credentials.
Privacy is a property of the identity, not the data:
We never store real names or personal emails. KYC happens through Bridge.xyz; the network never sees the documents, only the approved/rejected status. Each member receives a generated mailbox scoped to the network — not a personal email, no identifier leakage.
Three Solana mainnet mints. Token-2022 standard.
Florin (ƒ) is the unit of account. A multi-currency, collateral-backed basket. It is how value is denominated across the graph.
Joule is the contribution unit. Emitted by verified work × time, with a diversity adjustment that penalises concentration and decay on pending contributions. Joule is not a return product. It is the unit we use to measure collective effort.
House of Wisdom (HoW, ℏ) is the governance unit. Proposals consume it. Votes weigh it. Parameter changes burn it. It is allocated and earned. It is never bought.
The three combine via a CES production function with substitution elasticity near zero — concentration in one type is penalized, balanced contribution outscores narrow contribution. The synthesis is in the operational code; members can recompute their own score.
| Allocation | Share | Vesting | |---|---|---| | Anchor Circle | 10% | 10-year, 1%/year | | Operations Circle | 10% | none specified | | Early Contributors | 10% | none specified | | Partners and Advisors | 10% | none specified | | Seed Circle | 10% | none specified | | DAO | 50% | distributed via DAO governance |
Holacracy gives us roles, accountabilities, and structured grievance handling without a hierarchy. The DAO gives us on-chain proposal execution.
We run them together. Roles are defined in the graph and queryable. Proposals begin in Holacracy circles, get ratified on-chain, and execute as code where the proposal is executable. The full constitutional text lives on the docs site.
Where conventional finance asks for trust between counterparties, ƒxyz uses cryptographic primitives both parties accept: signatures for intent, on-chain anchoring for finality, zero-knowledge attestations for compliance proofs without identity leakage, threshold signatures for collective custody. Trust is not eliminated — it is moved from "trust this counterparty" to "trust this math."
The whole system is a graph and we treat it as one. Every concept — currency, counterparty, instrument, route, regulation, member — is a node. Every relationship between them is a typed edge. The same vocabulary, FIBO, that the EDM Council maintains for major banks and regulators is the spine; we extend it where finance is moving (programmable money, contribution layers, multiplex relations) without forking the parts that are stable.
Patterns are everywhere; the work is identifying them. A minimum node count is needed for patterns to emerge. The goal is to move from mazes to structured trees in pattern analysis. Routing, arbitrage, AML detection, governance traversal — every algorithm is a function over this graph, with graph-typed inputs and outputs.
The graph is the version-tracked substrate of canonical claims. When a claim is superseded, both the old and the new remain queryable.
We follow Kivelä's parallel-axis formalism. Containment first, then orthogonal axes.
Containment (ontic): realm ⊃ domain ⊃ territory.
Orthogonal axes within a territory (analytical):
Cardinality is a design choice. The 5 + 2 layer count is a working configuration, not a theorem, and it's the first DAO-vote candidate after emission.
The 21-grid claim from older drafts is retired in canon. Layers can be many. They may even be realm-dependent.
The full 48-entry bibliography from v1 is preserved at /resources/bibliography.