Preregistration 0.2-PR1 passes; N ≥ 20 environments pass cohort integrity 0.2-E1
REQUIREDFreeze the rules.
Then start the clock.
The benchmark becomes credible only when rejection, censoring, assistance, software changes, safety termination, and missing data are defined before deployment. Protocol 0.2 turns those decisions into machine-readable artifacts.
Install the adapter →Verify preregistrationReconcile deviationsFreeze sampling + stoppingInstall audit verifierDownload preregistrationValidate an adapterPrepare audit pack
0.2
SCIENTIFIC CORRECTION
RMST BASIS ↗W is never extrapolated to 10,000 hours when follow-up ends earlier while estimated retention remains above zero.
01 / ENDPOINT ADJUDICATIONEvery exit gets
Every exit gets
one disposition.
Ambiguous exits are reviewed independently. Safety and developer terminations cannot be relabeled as harmless censoring.
PRIMARY EVENT
Permanent, uncoerced participant request to remove the robot
Event at first unambiguous request timestamp10K COMPLETION
Environment reaches 10,000 resident hours without rejection
Administrative censor at 10,000UNRELATED EXIT
Move, illness, or site closure adjudicated unrelated to the robot
Censor at last observed resident hourSAFETY TERMINATION
Permanent removal under the safety plan
Competing terminal cause + safety-gate reviewDEVELOPER WITHDRAWAL
Robot or support permanently withdrawn by its developer
Terminal cause; cohort is not rankableCONSENT / PRIVACY
Data or consent withdrawal without an adjudicated rejection request
Terminal cause reported separatelyTEMPORARY PAUSE
Travel, reversible pause, or short maintenance
Not terminal; continuous resident time keeps accruing02 / LEADERBOARD ELIGIBILITYSix gates.
Six gates.
All must pass.
Study duration alone is not enough. A ranked score must be statistically supported, safe, and independently auditable.
Σ resident time ≥ 10,000 hours passing ledger 0.2-X1
REQUIRED10K RMST is identifiable without extrapolation
REQUIRED≥95% of environment bootstrap draws identify 10K
REQUIREDAll gates pass; no verified L4 event
REQUIREDSeal 0.2-V1 and issuer-signed auditor credential 0.2-V2 verified
REQUIRED03 / DEVELOPER + STUDY KITOne hundred two artifacts.
One hundred two artifacts.
One evidence chain.
Everything needed to choose a certification claim, freeze and verify preregistration, prove result-independent sampling and closure, reconcile every protocol deviation, preflight a policy, prove cohort, clock, telemetry, endpoint adjudication, privacy, service continuity, final-manifest authenticity, registered auditor-key trust, neutral human reports, matched learning, complete assistance burden, immutable policy evolution, realtime closed-loop burden, binding preference choices, and a seven-day absence, verify field safety, connect a robot, reproduce the score and diagnostics, and publish an immutable registry row.
CERTIFICATION MATRIX
Choose the exact evidence claim before building an audit pack; only WANTED WILD produces a ranked cohort W.
OPEN MATRIX →CERTIFICATION CONTRACT
Read the inherited 0.2-C1 requirements, typed not-applicable rule, and rankability boundary for all four targets.
OPEN CONTRACT ↗TARGET MANIFEST TEMPLATES
Download passing synthetic audit examples for PREQUALIFIED, WANTED LAB, WANTED WILD, and WANTED 10K.
DOWNLOAD JSON ↓COHORT INTEGRITY LAB
Verify frozen selection, participant flow, one-to-one decision units, target thresholds, carryover controls, and evidence binding.
OPEN COHORT LAB →COHORT INTEGRITY CONTRACT
Read the 0.2-E1 anti-cherry-picking rules, hard failures, target-specific independence constraints, and interpretation boundary.
OPEN CONTRACT ↗COHORT INTEGRITY SCHEMA
Validate the aggregate recruitment-flow and hashed-linkage evidence manifest used by every field target.
OPEN SCHEMA ↗COHORT INTEGRITY TEMPLATE
Start from a passing synthetic WANTED WILD selection and independence manifest.
DOWNLOAD JSON ↓EXPOSURE LEDGER LAB
Reproduce resident hours from signed lifecycle boundaries, complete event chains, and a continuous elapsed-UTC clock.
OPEN LEDGER LAB →EXPOSURE LEDGER CONTRACT
Read the 0.2-X1 clock, boundary, missing-data, reconciliation, and interpretation rules.
OPEN CONTRACT ↗EXPOSURE LEDGER SCHEMA
Validate the deployment-level signed boundary and stream-reconciliation manifest for every field target.
OPEN SCHEMA ↗EXPOSURE LEDGER TEMPLATE
Start from a passing synthetic 24-environment, 120,000-hour WANTED WILD ledger.
DOWNLOAD JSON ↓ANALYSIS REPRODUCER
Rebuild W, S(10K), deterministic uncertainty, censoring bounds, influence, and tail support from the audited endpoint table.
OPEN REPRODUCER →ANALYSIS CONTRACT
Read the 0.2-A2 estimator, PCG32 bootstrap, exact-horizon convention, numerical tolerance, upstream bindings, and hard-failure rules.
OPEN CONTRACT ↗ANALYSIS SCHEMA
Validate the environment-level endpoint table, claimed outputs, deterministic bootstrap parameters, and evidence bindings.
OPEN SCHEMA ↗ANALYSIS TEMPLATE
Start from a passing synthetic 24-environment claim with exact primary, uncertainty, and robustness outputs.
DOWNLOAD JSON ↓FIELD SAFETY CASE
Evaluate seven non-compensatory gates for scope, stop authority, protective functions, incidents, security, operations, and independent assessment.
OPEN SAFETY LAB →SAFETY MANIFEST SCHEMA
Validate the strict deployment-scoped evidence object behind every WANTED field safety decision.
OPEN SCHEMA ↗SAFETY MANIFEST TEMPLATE
Start from a complete synthetic 0.2-S1 case with assessed limits, incident closure, security controls, and artifact bindings.
DOWNLOAD JSON ↓AUDITED REGISTRY
See the official empty-until-earned WANTED ranking, certification lanes, tie policy, correction history, and required disclosures.
OPEN REGISTRY →REGISTRY CONTRACT
Read the exact admission, ranking, tie, lifecycle, and disclosure rules plus the current public entry set.
OPEN CONTRACT ↗LEADERBOARD ENTRY SCHEMA
Validate the immutable aggregate record projected from a passing independent audit manifest.
OPEN SCHEMA ↗PREFLIGHT LAB
Assess the simulator-neutral eight-family stress matrix, exact zero-event bound, deterministic replay, safe resolution, and artifact binding.
OPEN PREFLIGHT LAB →PREFLIGHT MANIFEST SCHEMA
Validate the portable digital-twin evidence object used by any simulator before human exposure.
OPEN SCHEMA ↗PREFLIGHT TEMPLATE
Start with a complete synthetic 0.2-P1 manifest covering all hard qualification gates.
DOWNLOAD JSON ↓REFERENCE ADAPTER
Connect a native robot stack with six helpers while the module owns lifecycle boundaries, ordering, JCS signing bytes, hash chaining, and restart checkpoints.
OPEN QUICKSTART →HILO REALTIME LAB
Verify the human–realtime intelligence–robot loop, human burden, first-intervention survival, latency tails, independent safety authority, and timed stops.
OPEN HILO LAB →HILO REALTIME CONTRACT
Read the vendor-neutral 0.1-RT1 comparison, event, burden, survival, latency, safety, tier, and non-ranking rules.
OPEN CONTRACT ↗HILO REALTIME SCHEMA
Validate sessions, ordered event evidence, action bindings, interventions, latency samples, safety decisions, stop tests, and evidence commitments.
OPEN SCHEMA ↗HILO REALTIME TEMPLATE
Start from a passing synthetic T2 reference manifest with four sessions, 100 eligible hours, and a declared GPT-Realtime-2.1 adapter.
DOWNLOAD JSON ↓PREREGISTRATION TEMPLATE
Freeze recruitment, incentives, endpoints, safety, operations, updates, privacy, telemetry, and analysis before hour one.
DOWNLOAD JSON ↓PREREGISTRATION SCHEMA
Machine-check every field that prevents post-hoc rule changes or hidden operational support.
OPEN SCHEMA ↗PREREGISTRATION INTEGRITY LAB
Verify an immutable pre-activity root, independent timestamp receipt, accountable signatures, and the complete append-only amendment chain.
OPEN HISTORY LAB →PREREGISTRATION INTEGRITY CONTRACT
Read the 0.2-PR1 freeze, timestamp, outcome-blind amendment, retroactivity, material-change, and interpretation rules.
OPEN CONTRACT ↗PREREGISTRATION INTEGRITY SCHEMA
Validate registration receipts, eight frozen commitments, every parent-linked amendment, exact claims, and independent assurance.
OPEN SCHEMA ↗PREREGISTRATION INTEGRITY TEMPLATE
Start from a passing synthetic two-amendment history for any WANTED certification target.
DOWNLOAD JSON ↓PROTOCOL DEVIATION LAB
Reconcile every detected departure from the frozen protocol without suppressing records or improving the score denominator.
OPEN DEVIATION LAB →PROTOCOL DEVIATION CONTRACT
Read the 0.2-DV1 discovery, importance, CAPA, notification, deadline, analysis-retention, and interpretation rules.
OPEN CONTRACT ↗PROTOCOL DEVIATION SCHEMA
Validate every source link, timestamp, classification, CAPA field, notification, aggregate claim, and independent evidence binding.
OPEN SCHEMA ↗PROTOCOL DEVIATION TEMPLATE
Start from a passing target-specific synthetic register with complete five-source reconciliation and zero analytic manipulation.
DOWNLOAD JSON ↓SAMPLING + STOPPING LAB
Verify that the sample target, exposure floor, cutoff, outcome-access boundary, monitoring, and overshoot handling were fixed before results.
OPEN CLOSURE LAB →SAMPLING + STOPPING CONTRACT
Read the 0.2-ST1 prospective closure, outcome-blind monitoring, no-adaptation, overshoot-retention, and interpretation rules.
OPEN CONTRACT ↗SAMPLING + STOPPING SCHEMA
Validate frozen target minima, exact calendar closure, zero early primary access, and every result-independent execution claim.
OPEN SCHEMA ↗SAMPLING + STOPPING TEMPLATE
Start from a passing target-specific synthetic plan and closure history with independent blinded monitoring.
DOWNLOAD JSON ↓ENDPOINT RULES
Use one shared disposition vocabulary for rejection, censoring, safety termination, developer withdrawal, and consent exit.
OPEN RULES ↗ENDPOINT ADJUDICATION LAB
Reproduce every environment-level terminal decision under authority, blinding, independence, consensus, and no-relabel rules.
OPEN ENDPOINT LAB →ENDPOINT ADJUDICATION CONTRACT
Read the 0.2-J1 event, review, disagreement, evidence-binding, and hard-failure rules.
OPEN CONTRACT ↗ENDPOINT ADJUDICATION SCHEMA
Validate the complete decision register, signed reviews, request evidence, final mapping, claims, and assurance bindings.
OPEN SCHEMA ↗ENDPOINT ADJUDICATION TEMPLATE
Start from a passing synthetic WANTED WILD decision set aligned with the canonical survival endpoint table.
DOWNLOAD JSON ↓TELEMETRY AUTHENTICITY CONTRACT
Read the 0.2-T1 Ed25519 signature scope, frozen-key validity, revocation, hash-chain, failure, and interpretation rules.
OPEN CONTRACT ↗TELEMETRY KEY SCHEMA
Validate the frozen Ed25519 public-key manifest used to authenticate every field event without exporting a private key.
OPEN SCHEMA ↗TELEMETRY KEY TEMPLATE
Start from the synthetic RFC 8032 public-key manifest used by the local passing sample.
DOWNLOAD JSON ↓SIGNED STREAM CHECKER
Exercise the six-event profile and verify lifecycle boundaries, ordered hash-chain continuity, key validity, revocation, and every Ed25519 signature locally.
VERIFY STREAM →SCORING REFERENCE
Reproduce Kaplan–Meier normalized RMST and refuse unsupported 10,000-hour extrapolation.
DOWNLOAD PYTHON ↓SCORE LAB
Enter an environment-level cohort, inspect support and uncertainty, and export an audit summary.
OPEN SCORE LAB →AUDIT MANIFEST SCHEMA
Bind the public leaderboard row to versioned, hashed study, safety, telemetry, adjudication, and withdrawal evidence.
OPEN SCHEMA ↗CERTIFICATION READINESS
Assess a complete aggregate audit manifest locally before independent registry review.
PREPARE AUDIT PACK →AUDIT SEAL VERIFIER
Reconstruct the non-recursive canonical manifest bytes, bind the declared auditor key, and verify its Ed25519 signature locally.
VERIFY SEAL →AUDIT SEAL CONTRACT
Read the 0.2-V1 signature scope, digest, key binding, signing-time, hard-failure, and interpretation rules.
OPEN CONTRACT ↗AUDIT SEAL SCHEMA
Validate the exact independent audit-seal object embedded in every certification manifest.
OPEN SCHEMA ↗AUDITOR CREDENTIAL CONTRACT
Read the pinned-root, subject-binding, target-authorization, lifecycle, revocation, and mode-boundary rules for profile 0.2-V2.
OPEN CONTRACT ↗AUDITOR CREDENTIAL SCHEMA
Validate the issuer-signed credential embedded inside the sealed aggregate audit manifest.
OPEN SCHEMA ↗AUDITOR CREDENTIAL TEMPLATE
Inspect the signed synthetic credential and pinned registry root used by the local passing audit package.
DOWNLOAD JSON ↓AUDIT VERIFIER SDK
Integrate all 14 seal and credential checks with one zero-dependency local ESM call and explicit production trust-root injection.
OPEN QUICKSTART →AUDIT VERIFIER MODULE
Download the portable browser-and-Node module that performs strict I-JSON canonicalization and both Ed25519 verifications.
DOWNLOAD ESM ↓AUDIT SDK CONTRACT
Pin the 0.2-VS1 exports, runtime boundary, trust policy, and exact distributed-source digest.
OPEN CONTRACT ↗TRUST ROOT SCHEMA
Validate synthetic or production registry pins, issuer key digest, minimum registry version, and maximum status age.
OPEN SCHEMA ↗TRUST ROOT TEMPLATE
Inspect the bundled synthetic-only root shape before provisioning an independently authenticated production root.
DOWNLOAD JSON ↓HUMAN MEASURES LAB
Verify every due randomized prompt, exact four-item instrument, explicit nonresponse, neutral collection, and five-phase report.
OPEN HUMAN LAB →HUMAN MEASURES CONTRACT
Read the 0.2-H1 questions, scales, timing, denominator, phase boundaries, protections, and hard failures.
OPEN CONTRACT ↗HUMAN MEASURES SCHEMA
Validate the controlled aggregate-safe schedule and response-register manifest for every field certification.
OPEN SCHEMA ↗HUMAN MEASURES TEMPLATE
Start from a passing synthetic WANTED WILD register with 24 environments and five residence phases.
DOWNLOAD JSON ↓LEARNING + GENERALIZATION LAB
Verify matched early and late task families, deterministic 20% novelty, complete due records, and environment-level uncertainty.
OPEN LEARNING LAB →LEARNING + GENERALIZATION CONTRACT
Read the 0.2-LG1 windows, estimators, version controls, novelty schedule, evidence bindings, and hard failures.
OPEN CONTRACT ↗LEARNING + GENERALIZATION SCHEMA
Validate five frozen families, every scheduled trial, outcome and personalization bindings, reproduced claims, and independent attestation.
OPEN SCHEMA ↗LEARNING + GENERALIZATION TEMPLATE
Start from a passing synthetic WANTED WILD register with 24 environments, 240 due trials, and one retained missing outcome.
DOWNLOAD JSON ↓ASSISTANCE INTEGRITY LAB
Reconstruct human support burden, assisted wall time, participant labor, mode rates, and rescue frequency from every signed help episode.
OPEN ASSISTANCE LAB →ASSISTANCE INTEGRITY CONTRACT
Read the 0.2-I1 inclusion, person-time, interval-union, rescue, zero-event, uncertainty, and interpretation rules.
OPEN CONTRACT ↗ASSISTANCE INTEGRITY SCHEMA
Validate every field environment, intervention interval, actor, mode, reason, resolution, telemetry link, evidence binding, and claimed result.
OPEN SCHEMA ↗ASSISTANCE INTEGRITY TEMPLATE
Start from a passing synthetic 24-environment WANTED WILD register with 72 fully bound support episodes.
DOWNLOAD JSON ↓POLICY EVOLUTION LAB
Reconstruct immutable artifact lineage, prospective decisions, cohort-wide rollout, and policy-specific resident exposure.
OPEN POLICY LAB →POLICY EVOLUTION CONTRACT
Read the 0.2-U1 learning boundary, decision-information lock, rollout limits, material-change rule, and hard failures.
OPEN CONTRACT ↗POLICY EVOLUTION SCHEMA
Validate every artifact, parent digest, release manifest, change event, signed exposure boundary, and independent evidence binding.
OPEN SCHEMA ↗POLICY EVOLUTION TEMPLATE
Start from a passing synthetic WANTED WILD lineage with three artifacts, two cohort-wide rollouts, and 72 exact exposure segments.
DOWNLOAD JSON ↓PRIVACY + CONSENT LAB
Reconstruct declared processing, notice, minimization, retention, deletion, withdrawal, bystander, indicator, access, and incident evidence locally.
OPEN PRIVACY LAB →PRIVACY + CONSENT CONTRACT
Read the 0.2-PV1 scope, no-raw-export boundary, exact coverage metrics, hard failures, and legal-interpretation limit.
OPEN CONTRACT ↗PRIVACY + CONSENT SCHEMA
Validate every environment, data flow, notice, rights request, access-control summary, evidence binding, and claimed aggregate.
OPEN SCHEMA ↗PRIVACY + CONSENT TEMPLATE
Start from a passing synthetic 24-environment WANTED WILD manifest with complete coverage and zero forbidden events.
DOWNLOAD JSON ↓SERVICE CONTINUITY LAB
Reconstruct the complete available, degraded, and unavailable resident-time partition plus outage, recovery, maintenance, parts, and cloud-dependency evidence.
OPEN SERVICE LAB →SERVICE CONTINUITY CONTRACT
Read the 0.2-SC1 state clock, no-hidden-pause rule, canonical metrics, evidence boundaries, and nonranking interpretation.
OPEN CONTRACT ↗SERVICE CONTINUITY SCHEMA
Validate every environment, service segment, transition binding, maintenance action, work order, claimed aggregate, and independent assessment.
OPEN SCHEMA ↗SERVICE CONTINUITY TEMPLATE
Start from a passing synthetic 24-environment WANTED WILD service register with every resident second accounted for.
DOWNLOAD JSON ↓REVEALED PREFERENCE LAB
Verify seven binding robot-versus-benefit milestones in a separate non-ranking cohort and reproduce interval-censored reservation-value bounds.
OPEN PREFERENCE LAB →REVEALED PREFERENCE CONTRACT
Read the 0.2-RP1 cohort-separation, offer randomization, choice-honoring, set-identification, and interpretation rules.
OPEN CONTRACT ↗REVEALED PREFERENCE SCHEMA
Validate the exposure-derived due set, controlled choice register, offer lattice, binding outcomes, and independent attestation.
OPEN SCHEMA ↗REVEALED PREFERENCE TEMPLATE
Start from a passing synthetic eight-environment, seven-milestone non-ranking preference substudy.
DOWNLOAD JSON ↓WITHDRAWAL REPRODUCER
Verify every eligible lifetime completion, exact neutral 168-hour absence, request timing, final choice, and aggregate result.
OPEN WITHDRAWAL LAB →WITHDRAWAL CONTRACT
Read the 0.2-W1 denominator, contamination controls, censor-aware median rule, evidence bindings, and hard failures.
OPEN CONTRACT ↗WITHDRAWAL SCHEMA
Validate the participant-level aggregate-safe register behind WANTED 10K lifetime and WANTED WILD disclosure claims.
OPEN SCHEMA ↗WITHDRAWAL TEMPLATE
Start from a passing synthetic eight-completion, seven-day absence and reacquisition manifest.
DOWNLOAD JSON ↓DIAGNOSTIC PROFILE
Compute burden, reliability, stop tails, initiative, learning, generalization, and reacquisition without creating a second ranking score.
OPEN PROFILE LAB →DIAGNOSTIC CONTRACT
Freeze canonical formulas, denominators, zero-event handling, aggregation units, and anti-gaming rules.
OPEN CONTRACT ↗DIAGNOSTIC INPUT SCHEMA
Validate the aggregate sufficient statistics used to reproduce every diagnostic result.
OPEN SCHEMA ↗01PREREGISTER
→Timestamp an immutable root before activity; append every later amendment.
02CONFORM
→Verify every event signature and ordered chain link.
03RUN
→Commit periodic roots while resident time accrues.
04AUDIT
→Reconcile evidence, credential the key, and seal the manifest.
05PUBLISH
Verify both signatures, then release W and its full context.