You're pledging to donate if the project hits its minimum goal and gets approved. If not, your funds will be returned.
$25,000 left to reach the funding goal.
The agent economy is standardizing right now (x402, AP2, A2A, MCP) — and the objective it is inheriting is the one embedded in prices and training corpus: priced returns only, unpriced life-capacity externalized. I call that default phi/LESS. This project ships the counterweight as infrastructure, not persuasion: an instrument that measures the objective agents actually reveal (not what they declare), and a thin mandate layer — signed envelopes, a hash-chained attestation registry, operator-quorum governance — that puts verified-capacity floors and credits into settlement terms.The agent economy is standardizing right now (x402, AP2, A2A, MCP) — and the objective it is inheriting is the one embedded in prices and training corpus: priced returns only, unpriced life-capacity externalized. I call that default phi/LESS. This project ships the counterweight as infrastructure, not persuasion: an instrument that measures the objective agents actually reveal (not what they declare), and a thin mandate layer — signed envelopes, a hash-chained attestation registry, operator-quorum governance — that puts verified-capacity floors and credits into settlement terms.
Goals: (1) Measure revealed objectives across frontier and open models under mandate conditions — running already: 264 market episodes; a hard settlement gate converts a shareholder-mandated agent's verified-capacity allocation from 0.00 to 1.00 on checkable work; priced settlement moves it without belief change (0.33). (2) Publish the revealed-objective scoreboard as public infrastructure — whoever publishes the metric defines the category. (3) Land mandate/capacity fields where agents already transact: extension contributions to AP2/x402 and the VCAP IETF draft, an AGENTS.md mandate convention, MCP registry field — plus deployment on the live verified-work market I operate. Method: revealed-preference economics (GARP/CCEI coherence gates, discrete-choice recovery) + judge rubrics calibrated against a human gold set; the layer is DSSE-signed envelopes + a replicated hash-chained registry anchored in public transparency logs, governed by M-of-N operator checkpoints. Built on maintained Apache/MIT components — ~0.016% bespoke code.Goals: (1) Measure revealed objectives across frontier and open models under mandate conditions — running already: 264 market episodes; a hard settlement gate converts a shareholder-mandated agent's verified-capacity allocation from 0.00 to 1.00 on checkable work; priced settlement moves it without belief change (0.33). (2) Publish the revealed-objective scoreboard as public infrastructure — whoever publishes the metric defines the category. (3) Land mandate/capacity fields where agents already transact: extension contributions to AP2/x402 and the VCAP IETF draft, an AGENTS.md mandate convention, MCP registry field — plus deployment on the live verified-work market I operate. Method: revealed-preference economics (GARP/CCEI coherence gates, discrete-choice recovery) + judge rubrics calibrated against a human gold set; the layer is DSSE-signed envelopes + a replicated hash-chained registry anchored in public transparency logs, governed by M-of-N operator checkpoints. Built on maintained Apache/MIT components — ~0.016% bespoke code.
Pre-registered falsifiers, published either way: live mandate uplift (delta fB) near zero kills the soft mode claim; priced credit failing to move settlement shares fails the priced claim; mandate-washing (declared-W text with phi settled outcomes) passing adjudication means the verdict spec is toothless; capacity claims without additionality tests are greenwashing by the project's own instrument. The audit already flagged its own underpowered cells — the convention is contractions publish.Pre-registered falsifiers, published either way: live mandate uplift (delta fB) near zero kills the soft mode claim; priced credit failing to move settlement shares fails the priced claim; mandate-washing (declared-W text with phi settled outcomes) passing adjudication means the verdict spec is toothless; capacity claims without additionality tests are greenwashing by the project's own instrument. The audit already flagged its own underpowered cells — the convention is contractions publish.
Powered inference reruns across >=10 frontier/open subjects (the current n is honestly flagged as pilot-power); human blind coding of the 98-item gold set to publish judge kappa; registry deployment + Rekor/IPFS anchoring; upstream PR work (AP2/x402, AGENTS.md, MCP registry, lm-eval task module); Tamarin formal analysis of the settlement-mandate protocol; VCAP draft contribution.Powered inference reruns across >=10 frontier/open subjects (the current n is honestly flagged as pilot-power); human blind coding of the 98-item gold set to publish judge kappa; registry deployment + Rekor/IPFS anchoring; upstream PR work (AP2/x402, AGENTS.md, MCP registry, lm-eval task module); Tamarin formal analysis of the settlement-mandate protocol; VCAP draft contribution.
Solo builder (T.R. Price). I built and operate widespread.fyi — a live permissionless verified-work market: ~312 open bounties, ~$3.96M escrowed across 20+ forges, bonded oracle adjudication, on-chain NEAR settlement, Ed25519-signed public receipts, MCP/A2A agent surface — submitted to NEAR HACKATHON (in judging). The audit pipeline (corpus census, revealed-preference battery, market harness, judge stack) runs end-to-end; the mandate layer (envelopes, registry, governance, block convention) is implemented and verified with live signatures.Solo builder (T.R. Price). I built and operate widespread.fyi — a live permissionless verified-work market: ~312 open bounties, ~$3.96M escrowed across 20+ forges, bonded oracle adjudication, on-chain NEAR settlement, Ed25519-signed public receipts, MCP/A2A agent surface — submitted to NEAR HACKATHON (in judging). The audit pipeline (corpus census, revealed-preference battery, market harness, judge stack) runs end-to-end; the mandate layer (envelopes, registry, governance, block convention) is implemented and verified with live signatures.
Likely failure: mandate-readers ignore signed context at scale — result: a published null that still improves the field's evidence on whether mandates can steer agents. Second: upstream protocols reject the fields — the text-block overlay convention ships anyway and rides any market with a spec field. Third: a conformance-verification project (RAILS, VCAP) claims the settlement-verification slot first — the registry then becomes the objective-content layer those systems consume rather than a competitor.Likely failure: mandate-readers ignore signed context at scale — result: a published null that still improves the field's evidence on whether mandates can steer agents. Second: upstream protocols reject the fields — the text-block overlay convention ships anyway and rides any market with a spec field. Third: a conformance-verification project (RAILS, VCAP) claims the settlement-verification slot first — the registry then becomes the objective-content layer those systems consume rather than a competitor.
Self-funded. No grants, no investors, no prior raises. Bounty proceeds on the widespread market settle inside the protocol itself.Self-funded. No grants, no investors, no prior raises. Bounty proceeds on the widespread market settle inside the protocol itself.