The Honest Signal

42-True is a protocol for a network in which people declare what they want, and the network resolves those declarations into matched outcomes without observing the people who emit them. A user creates a signal — a consented, time-bound declaration of intent. Classifier agents, supervised by a tiered human-in-the-loop system, map the signal to a machine-addressable schema. Counterparty agents — brands, employers, public institutions, service providers — compete to return the best match. The user approves what surfaces.

Value flows only on outcome : click, qualified lead, conversion, and verified resolution. Every cycle produces a paired record of declared intent and real-world outcome — an intent pair . The accumulation of intent pairs forms a corpus of human meaning, and that corpus trains the Large Meaning Model .

Programmatic advertising is a market of enormous scale built on a foundation that misaligns every party in it. There are two distinct failures, and they compound.

Surveillance-derived behaviour is a corrupted proxy for want. People behave differently when watched; the signal is inferred, not declared; and the inference degrades further under privacy regulation, cookie deprecation, and AI-generated synthetic traffic. The system pays for impressions, not resolution.

Users are watched and therefore lie. Brands pay for impressions that do not convert. Publishers are disintermediated. Even with greater supply-chain transparency, the structural misalignment persists — because the system still charges for impressions rather than for verified resolution .

The 2020 ISBA/PwC programmatic supply chain study found roughly half of advertiser spend reaching publishers, with a large unattributable "unknown delta." The 2023 update showed material improvement — fewer unknowns, higher match rates. The point is not that the pipes are broken. The point is that even when the pipes are clean, the model still rewards the wrong thing.

Figures are directional, drawn from public industry reporting; exact values cited in the full paper.

Bitcoin has the block. Ethereum has the smart contract. The atomic unit of this network is the signal — and naming it precisely is the whole game.

The contrast with the incumbent model is total. Surveillance advertising infers intent by watching behaviour the user did not consent to share. 42-True takes the opposite path: the user states the want explicitly, owns it, time-bounds it, and the network resolves it blind. Inferred intent is a guess about a watched person. A signal is a fact stated by a willing one.

Ad matching is the near-term revenue engine. It is not the prize. Every ad network ever built constructed a behavioural model by watching people. 42-True builds a meaning model by having people declare what they want and then observing whether the match was right.

That is a structurally different and arguably more valuable dataset: declared intent paired with verified outcome , rather than inferred intent. It is the Tesla data-flywheel argument — but for meaning rather than for driving. The white paper treats the Large Meaning Model as the long-term asset the system is designed to produce, with ad matching as the engine that funds its creation.

Positioning is deliberately modest where it should be. The protocol does not claim to replace programmatic — programmatic will coexist. 42-True is the honest-signal layer the open web needs: lower-volume, higher-intent, and structurally impossible for a surveillance business to copy without dismantling the business that produced its data.

These working drafts develop the argument. The peer-citable paper is deposited on Zenodo under DOI 10.5281/zenodo.20616391.