Resolving what people want,
in conditions where
nothing is watching.

The atomic unit of the corpus is the four-tuple:

IntentPair = ( Id, C, M, O )

Id — Declared intent. A want, stated in natural language, authored and consented, bound only to an unlinkable token.

C — Classification. The declaration mapped to a machine-addressable taxonomy node by a three-tier resolver (agents, community, experts).

M — Match. The counterparty offer returned in response. Counterparties see only the classification, never the raw signal.

O — Outcome. The realised consequence — graded from no-engagement through to a cryptographically verified resolution. Supplied by the declarant's own subsequent action in the world, not by an annotator.

Unlike a static NLU label (assigned by an annotator) or an RLHF preference (whose consequence is synthetic), the outcome O is lived. All records are frozen: an intent pair records what happened; corrections produce new records, and the corpus grows by accumulation.

A purely theoretical treatment of the Large Meaning Model would be shorter than this — and insufficient. The corpus does not exist. It cannot be scraped from the public web (which is observational by construction), cannot be simulated (the signal of interest is precisely what people declare when unobserved), and cannot be extracted from existing systems (whose data is the data this model is defined against).

The corpus must be produced, and production requires a network: a place where declaration is safe, a layer that classifies declarations at scale, a market that creates the conditions for outcomes to occur, and a verification layer that anchors those outcomes to attestable events. This repository is the schema and the seed; the network is the work.

A model trained on the declared intent of a population should not be the property of any single commercial actor. The paper proposes that a foundation hold the Large Meaning Model as a commons, steward the protocol, and distribute returns to contributors in proportion to the value their contribution adds.

Four working drafts develop the protocol, its architecture, and its incentives. The repository holds them all.

A network in which people declare what they want — and the network resolves those declarations into outcomes without ever observing the people who emit them. Defines the signal and the two failures of the existing system: the observational failure and the incentive failure.

Signal → Classify → Match → Resolve. The intent pair as the atomic record, why it compounds across volume, fidelity and breadth, and the two revenue primitives: resolution fees and a metered inference API.

Adversarial honesty. Three model paradigms compared, a ledger of what is genuinely novel against what is prior art (RLHF reward shaping, Data Shapley, differential privacy, W3C DIDs), and three claims deliberately walked back.

Batch-level differential privacy and staged training. Three commitments: signals never linked to persistent identity, the corpus anonymised and differentially private, counterparties learn statistics and never people. Then the ownership question, and the constitutional commitments.