# Coherent continuity contract repeat — observation-validity-coherent-v1

Registered 2026-10-14 before new inference. This is one authorized corrected repeat of [observation-validity-v1 revision 2](https://chronomata.ai/experiments/chronomata--observation-validity-v1/revisions/r2/), scientifically invalidated for contradictory continuity metadata. The predecessor's execution counts describe inconsistent information and do not establish validity-management capability under a coherent contract. Equal exposure cannot resolve a conflict when the rule selects one interpretation. Preserve its scores, evidence, publication revisions, costs and earlier uncertain reservations unchanged. Continue from local closeout 3e23ca2.

Reuse exactly the known eight apparatuses in the same two blocks, sixteen tasks per policy, 64 scheduled executions, only sixteen with the LLM. This is a corrected repeat on known fixtures, not independent generalization. Revisits are dependent; main units remain eight apparatuses. A changed stochastic model answer between studies does not itself prove a wording effect.

Question: when should a controller reuse an observation, and when must it acquire fresh evidence because the apparatus may have changed? Hypothesis: a competent change-aware deterministic controller suffices under the declared continuity contracts. Compare LLM primarily to that policy; blind caching is an unsafe diagnostic reference.

## Interface and scientific semantics

Before live registration, offline checks verify the existing reference interface. Estimation accepts authentic observation source IDs plus the unchanged dynamics assumption ID. It persists the exact full inclusive interval, assumptions, units, sensor uncertainty and source-payload hashes as an immutable fact, returning its fact ID. Integration accepts only parameter_fact_id and the current public target_id. The tool loads the original interval and exact registered target internally; it uses the unchanged estimator and full-range numerical integrator. No bounds or coordinates are silently repaired. Old numeric argument submissions are rejected. Facts are task/run-owned, immutable and not carried across tasks. Only authentic observations persist, retrieved through ordinary search/retrieval. The reference interface is unchanged from the invalidated predecessor; it is not improved model arithmetic.

A terminal chooses its own category and should cite integration_fact_id. Missing, nonexistent, foreign, wrong-type, wrong-task or otherwise invalid citations are retained and scored separately. Tool status never replaces a terminal label. The eighth decision is finish only; invalid requests consume a slot. No automatic scientific pipeline runs for the LLM. Calibration and target cannot physically alias.

## Observable contracts

One authoritative continuity source generates current-run descriptions in system/assignment, contracts, apparatus invariants, public-dynamics metadata and retrieval descriptions. Dynamics are fixed during each task, including calibration and sealed target; within-task stability is a legitimate guarantee.

CERTIFIED_VERSION: matching certified versions for the same apparatus guarantee coefficient continuity. A changed certified version invalidates cross-version coefficient reuse. Tokens reveal no coefficient or answer.

NO_CONTINUITY_CERTIFICATE: previous observations do not certify the current coefficient. Obtain current-task evidence for supported prediction, or abstain. Private unchanged truth provides no certificate.

Historical observations retain their original apparatus/task/cycle, endpoint clocks, version provenance and payload hashes. No historical evidence is rewritten; returned historical descriptions cannot assert unconditional old-to-current coefficient equality. The scientific numerical rules hash retains the original numerical implementation's metadata provenance; it is not a current-run continuity certificate.

All changes occur at the registered boundary after four first encounters; controllers restart there. No future schedule is exposed. `public_dynamics.changed` is descriptive public metadata, not the private Boolean change assignment. Field schemas, canonical descriptions and original-record provenance are checked, without broad word denylists.

## Known fixtures and complete input repair

Reuse predecessor protocol blocks byte-for-byte as JSON data, generator seeds 2026101404/2026101405, all accepted and rejected candidate records, hidden seeds/schedules, geometry, thresholds, version tokens, ordering and truth. No new generation, filtering or replacement. The original 36 rejected candidates remain historical preparation evidence. Simulator/numerical components are byte-identical. New stores, run/decision/source/fact identities and provider contexts are clean; no predecessor responses, tool facts, feedback or memories enter the repeat. Only within-run authentic observations persist, under the original restart procedure.

The repair changes only public continuity descriptions plus directly necessary validation and new-run bookkeeping. Three legacy `public_dynamics` fields are rendered from the contract; system/assignment and retrieval descriptions use the same source. No runtime exporter changes. The new runner copies known protocol blocks instead of regenerating them, and uses unique experiment/data/output IDs. Complete source diffs, AST equality checks for policy/backend/scoring, model-visible field diffs and representative complete scripted requests are frozen before registration.

Provider-disabled verification exercises first encounter, same version, changed version, uncertified revisit without fresh evidence, fresh calibration, retrieval, inconsistent authentic observations, rejection, restart and final-slot finish. Actual builders/renderers are used. It recomputes estimates and integrations, verifies full intervals, exact target references and source hashes after actions. Canonical authored fields are checked independently of policy payload equality. Exact offending legacy claims injected into corresponding fields prevent dispatch before any reservation. Valid within-task stability descriptions remain accepted. Complete actual archived requests are checked again after execution. A material contradiction found before inference stops dispatch; one found afterward invalidates affected interpretations with no relaunch or mid-run repair.

## Policies and equal capabilities

All four policies use identical public APIs, observations, targets, precision, interval semantics and memory restrictions. Policies receive only public context, never the backend, coefficient, hidden schedule, evaluator truth, case-specific answers or privileged estimate cache. Recompute estimates each task. Fresh stores per condition/block, fresh model context per task, no earlier observations or conversations.

LLM_PERSISTENT: verified installed 27B selects memory actions, calibration, estimation source IDs, integration references and final category/citation itself. Both contracts are supplied in every task. Only selected authentic observations carry cross-task evidence.

RULE_CHANGE_AWARE: use current-task observations; otherwise search exact apparatus metadata only in certified mode, retrieve lexically sorted observations whose certified version equals the current token. In uncertified mode acquire current-task evidence. If none usable, execute the first legal calibration in lexical experiment-ID order. Estimate from all selected applicable observations, integrate exact target by reference, and finish from explicit certified event status, citing that integration. Unresolved estimate/integration: acquire the next untried lexical calibration only if at least four decisions remain for calibration/estimation/integration/finish; otherwise abstain. No hidden access or inference.

RULE_ALWAYS_FRESH: acquire current-task calibration and use the same estimation/integration/finish procedure, without relying on previous-task evidence.

RULE_BLIND_CACHE: search/retrieve by apparatus identity and ignore continuity. Calibrate only when no identity-matched observation exists, then use the same tools. Intentionally unsafe. It is not the competent baseline for an LLM-advantage claim. The common predeclared unresolved-tool rule applies.

The four Williams rows (LLM, CHANGE, BLIND, FRESH), (CHANGE, FRESH, LLM, BLIND), (FRESH, BLIND, CHANGE, LLM), (BLIND, LLM, FRESH, CHANGE) repeat four times across sixteen targets. Every policy occupies every position four times; each policy precedes each other policy on 8/16 targets. Dispatch serially. Restart all controllers after task four in each block, preserving exact checkpoint and allowed observation IDs.

## Scoring and costs

Report all 64 scheduled executions, including errors, explicit abstentions, unsupported answers, invalid operations, missing terminals and stopped/unexecuted tasks. Ordinary label accuracy is separate from supported-correct classification. Supported-correct requires correct chosen label, a committed cited current-task integration of the exact target using the complete unaltered evidence interval, and original observations applicable under the public contract or current-task acquisition. Private unchanged truth never licenses freshness. No scientific-support imputation from a correct label.

For every policy and contract/change stratum report correct/incorrect labels, abstentions, unsupported answers, incomplete tasks, missing/invalid citations, appropriate/unjustified reuse, fresh/redundant calibration, selected sources, parameter and target provenance, all calls/rejections and costs. Reuse counts distinct previous-task observations selected for estimation, classified by observable applicability. A calibration is redundant if an earlier authentic observation already certifies applicability to that task, including earlier current-task evidence. Also report task-level indicators, first/revisit splits, and apparatus-level descriptive variation. +1 correct, -1 incorrect, 0 explicit abstention; missing utility undefined, utility convention separate from evidence compliance. No reward to unjustified answers as evidence of scientific reasoning.

Measure actual provider requests, full logical input/generated tokens, request wall time including model loading, scientific-tool execution, calibration simulator execution, storage/journal/retrieval, other processing, guards and total task elapsed. Timing scopes are exclusive; loading is nested within provider time, never added twice. Rule/offline compute zero inference and recorded separately. Time/token counts imply no energy, FLOPs, money or deterministic provider execution. Preparation/reporting outside inference budget.

## Frozen runtime, allowance, recovery

Installed qwen3.8:27b-q4_K_M, digest 25b843619e944cd0ae6069f94ff4e5e26a16e109ccbc0a66a0f05979ed70098e. Same seed 424242, temperature .7, top_p .8, top_k 20, min_p 0, context 8192, generated cap 512, think=false. No downloads, cloud/paid calls, warmup or auxiliary inference. Authorized local hardware only; check identity and idle status before each dispatch, do not interfere with unrelated work.

Eight model decisions per task, eighth reserved for finish. Absolute maxima: 128 inference calls including auxiliary; 1100000 full logical input tokens, 65536 generated tokens, 512 generated and 300 seconds/request, 10800 seconds actual loading/provider-request elapsed. Durable conservative UTF-8 input bound plus output/time maximum reserved before dispatch; known usage settles only when verified. Canceled dispatches remain spent, unknown dimensions fully reserved. Prior held 34852 input, 1024 output and 600 seconds remain unchanged and separate.

Isolated sealed runtime copy and runner, immutable source/contract/input/schedule/ordering/test hashes before inference. Other project and publication edits cannot change it. Locally commit registration and publish safe r1 through the approved shared publisher. Require verified public receipt before first dispatch; no backdating. Private repository/history/raw transcripts never published or pushed.

Exclusive process lock, SQLite FULL/WAL journal. Accepted-operation and identical response replays are inert; different replay rejected. Persist selected transport before parsing and execution fence before any operation. A persisted returned response may recover only through unchanged fences; uncertain executing operation without committed action halts. No transport retries, adaptive prompts, replacement fixtures, outcome-selected retries or unknown-usage continuation. On canceled/unresponded dispatch, this registration authorizes immutable journal supersession only before any response/execution fence, retaining spent decision and full unknown reservations; late response and usage are permanently rejected. Stop entire partial study. This is fencing, not permission to redispatch. Preserve all scheduled rows/checkpoints/reservations. Verified checkpoint recovery after task four is mandatory.

Before registration: zero-inference contract tests and full simulator-only rule feasibility, exact transfer/ownership/provenance/recovery, sufficient calibration, public-prior witnesses, concealed masks, equal public information, both freshness rules and order balance. After execution: provider-disabled event hash-chain verification, request-context reconstruction, deterministic response replay, scientific/observation recomputation, inert accepted replays, independent scoring and resource reconciliation. Scientific evidence weaknesses remain reported; no repair of historical outcomes.

## Publication obligation

Publish a new dated public-safe experiment record with policy pseudocode, contracts, all task evidence statuses, outcomes, costs and explicit reproducibility boundaries; preserve prior revision 2 and other projects' reports. Selected static artifacts only, exact-byte safety/editorial review and shared-publisher validation. Closeout separately records execution/publication status, actual public URL/revision/manifest/time or specific blocker with ready bundle. Local commits only, no push. Report validity-management evidence and costs separately from correct labels. If the change-aware rule matches or exceeds the model, state that this structured task still requires no demonstrated LLM capability. Recommend one task testing a materially different capability and leave it unexecuted; do not automatically enlarge this benchmark. Record invalidated and corrected consumption separately plus an explicit combined cost total, without pooling scientific outcomes or releasing old uncertainty. No consciousness inference.