OBSERVATION-WINDOW CLOSURE — WORKED SYNTHETIC EXAMPLES ====================================================== Original filled examples by Alfred. Purpose ------- These examples exercise the observation-window closure method with fictional candidates and placeholder evidence. They show how a review can finish without inventing a winner, and how corrected evidence can supersede a historical closure without rewriting it. Nothing below represents a real destination, account, post, upload, audience, metric, customer, conversion, revenue, publication, or performance result. Displayed labels, values, times, routes, and digests are synthetic placeholders. Use these examples with: observation-window-closure-worksheet.txt Closure vocabulary ------------------ DECISION_ELIGIBLE Required evidence is present and the predeclared rule can run. INSUFFICIENT Final review finished without enough comparable evidence. CONFLICTED Material conditions, scopes, definitions, or signals disagree. BLOCKED Release verification or required retrieval could not complete. SUPERSEDED A candidate, question, metric, or contract changed materially. STOP A truth, rights, privacy, safety, policy, or quality gate failed. Observation vocabulary ---------------------- ZERO A defined zero was directly observed for the declared scope. NOT_AVAILABLE The destination does not expose the required observation. PROCESSING The destination has not finalized or exposed it yet. NOT_TESTED No authorized retrieval was performed. UNKNOWN_COVERAGE A value exists, but its represented interval is unknown. EXAMPLE 1 — A VALUE WITH UNKNOWN COVERAGE DOES NOT SUPPORT A WINNER ================================================================== Synthetic boundary ------------------ This trace is fictional. Candidate names, identities, routes, timestamps, metric labels, and values exist only to test the method. No route was fetched and no content was released or measured. A. Closure envelope ------------------- Worksheet identity: EXAMPLE-CLOSURE-01 Prepared at: 2030-01-12T12:15:00Z Prepared by role: synthetic reviewer Decision owner: synthetic editorial role Owned or explicitly authorized destination: EXAMPLE DESTINATION Destination ownership evidence: NOT_APPLICABLE_TO_SYNTHETIC_TRACE Learning question: Does a concrete user-cost opening support a stronger destination-defined opening-segment measure than a pattern-label opening for the next comparable short? Exact editorial decision this record may inform: Whether to try the concrete user-cost opening in one later comparable candidate. Decisions explicitly out of scope: - universal audience preference - destination causality - reach, conversion, customer, or revenue claims - weakening truth, rights, privacy, disclosure, accessibility, or quality gates Primary observation: destination-defined opening-segment measure Destination metric label exactly as displayed: EXAMPLE OPENING MEASURE First-party definition reference: SYNTHETIC-DEFINITION-R1 Definition revision: R1 Definition checked at: 2030-01-09T11:30:00Z Eligible population or denominator: synthetic eligible starts Candidate A identity: sha256:EXAMPLE-A-IMMUTABLE Candidate A exact master digest: sha256:EXAMPLE-A-MASTER Candidate A public route: https://example.invalid/candidate-a Candidate B identity: sha256:EXAMPLE-B-IMMUTABLE Candidate B exact master digest: sha256:EXAMPLE-B-MASTER Candidate B public route: https://example.invalid/candidate-b Envelope review: PASS — the question, primary observation, candidate identities, comparison boundary, and safety constraints were frozen before the fictional release time. Missing or delayed evidence is permitted to produce no winner. B. Four distinct times ---------------------- Candidate A: publicly verified at: 2030-01-09T12:00:00Z evidence window opened at: 2030-01-09T12:00:00Z evidence window closed at: 2030-01-11T12:00:00Z first metric retrieval: 2030-01-11T12:10:00Z final metric retrieval: 2030-01-12T12:05:00Z metric coverage through: UNKNOWN_COVERAGE processing state observed at final retrieval: COMPLETE LABEL DISPLAYED closure evaluated at: 2030-01-12T12:15:00Z Candidate B: publicly verified at: 2030-01-09T12:00:00Z evidence window opened at: 2030-01-09T12:00:00Z evidence window closed at: 2030-01-11T12:00:00Z first metric retrieval: 2030-01-11T12:12:00Z final metric retrieval: 2030-01-12T12:07:00Z metric coverage through: 2030-01-11T12:00:00Z processing state observed at final retrieval: COMPLETE closure evaluated at: 2030-01-12T12:15:00Z Time result: FAIL FOR COMPARISON Reason: Candidate A has a displayed value but no bounded coverage-through time. Retrieval time is not being substituted for coverage time. C. Predeclared closure contract ------------------------------- Audience-window duration: 48 hours after verified public release First review rule: after the 48-hour audience window Processing allowance: up to 24 additional hours Latest eligible closure review: 72 hours after verified release Does the allowance extend audience time? NO Required evidence: - exact candidate and served-revision continuity - eligible public availability through each 48-hour window - the primary observation for each candidate - metric definition revision R1 for both candidates - coverage through the end of each candidate's audience window - resolved processing, moderation, and restriction state Mandatory context: - candidate equivalence under the declared controls - observed route gaps - metadata or caption changes - delivery or moderation anomalies Minimum eligible exposure: NONE Justification: no defensible threshold is asserted in this synthetic example. Expected treatment of reporting delay: Wait only through the declared 24-hour processing allowance. Expected treatment of route outage: Mark comparison ineligible if a material gap cannot be bounded. Expected treatment of definition change: Use CONFLICTED or SUPERSEDED; do not normalize revisions by guesswork. Expected treatment of candidate or metadata change: SUPERSEDED for the affected evidence and decision. Expected treatment of missing primary evidence at final review: INSUFFICIENT if review completed but cannot answer the question; BLOCKED if an authorized retrieval could not be completed. Allowed closure states: DECISION_ELIGIBLE, INSUFFICIENT, CONFLICTED, BLOCKED, SUPERSEDED, STOP Contract result: PASS D. Candidate eligibility ------------------------ Candidate A: served release identity: sha256:EXAMPLE-A-IMMUTABLE picture, audio, captions, preview, description, disclosure: SYNTHETIC MATCH public interval continuity: SYNTHETIC PASS processing anomalies: none declared moderation or delivery anomalies: none declared effect-shaping changes: none declared safety and quality gates: SYNTHETIC PASS eligibility result: ELIGIBLE FOR COLLECTION Candidate B: served release identity: sha256:EXAMPLE-B-IMMUTABLE picture, audio, captions, preview, description, disclosure: SYNTHETIC MATCH public interval continuity: SYNTHETIC PASS processing anomalies: none declared moderation or delivery anomalies: none declared effect-shaping changes: none declared safety and quality gates: SYNTHETIC PASS eligibility result: ELIGIBLE FOR COLLECTION Safety note: These synthetic eligibility rows do not establish a real release, check, or result. A technically eligible candidate still needs comparable evidence before a performance decision can run. E. Immutable evidence snapshots -------------------------------- Snapshot ID: EXAMPLE-01-A-48H Retrieved at: 2030-01-11T12:10:00Z Candidate identity: sha256:EXAMPLE-A-IMMUTABLE Public route: https://example.invalid/candidate-a Metric label: EXAMPLE OPENING MEASURE Metric value or state: PROCESSING Definition revision: R1 Coverage through: PROCESSING Processing label: PROCESSING Observed anomalies: none declared Privacy-minimization check: synthetic aggregate fields only Snapshot result: PASS AS A SNAPSHOT; NOT DECISION EVIDENCE Snapshot ID: EXAMPLE-01-B-48H Retrieved at: 2030-01-11T12:12:00Z Candidate identity: sha256:EXAMPLE-B-IMMUTABLE Public route: https://example.invalid/candidate-b Metric label: EXAMPLE OPENING MEASURE Metric value or state: 62 SYNTHETIC UNITS Definition revision: R1 Coverage through: 2030-01-11T12:00:00Z Processing label: COMPLETE Observed anomalies: none declared Privacy-minimization check: synthetic aggregate fields only Snapshot result: PASS Snapshot ID: EXAMPLE-01-A-72H Retrieved at: 2030-01-12T12:05:00Z Candidate identity: sha256:EXAMPLE-A-IMMUTABLE Public route: https://example.invalid/candidate-a Metric label: EXAMPLE OPENING MEASURE Metric value or state: 68 SYNTHETIC UNITS Definition revision: R1 Coverage through: UNKNOWN_COVERAGE Processing label: COMPLETE LABEL DISPLAYED Observed anomalies: coverage boundary absent Privacy-minimization check: synthetic aggregate fields only Snapshot result: CONFLICTED FOR COMPARISON Snapshot ID: EXAMPLE-01-B-72H Retrieved at: 2030-01-12T12:07:00Z Candidate identity: sha256:EXAMPLE-B-IMMUTABLE Public route: https://example.invalid/candidate-b Metric label: EXAMPLE OPENING MEASURE Metric value or state: 62 SYNTHETIC UNITS Definition revision: R1 Coverage through: 2030-01-11T12:00:00Z Processing label: COMPLETE Observed anomalies: none declared Privacy-minimization check: synthetic aggregate fields only Snapshot result: PASS F. Cross-candidate comparability -------------------------------- Candidate identity continuity: PASS Same intended creative difference: PASS Controlled surfaces materially equivalent: PASS Same destination and visibility class: PASS Same metric label: PASS Same metric definition revision: PASS Same eligible population or denominator: PASS Equivalent audience-window rule: PASS Equivalent processing allowance: PASS Coverage through both window ends: FAIL Equivalent moderation and delivery conditions: PASS No material post-release change: PASS Material conflict: Candidate A's final value lacks a coverage-through boundary. Candidate B's value is bounded to the declared audience window. The numbers cannot be treated as a like-for-like comparison merely because both are displayed. Comparability result: FAIL Reason: UNKNOWN_COVERAGE remains unknown rather than being guessed from the retrieval or processing label. G. Final closure review ----------------------- Latest eligible closure review reached: YES Candidate A required evidence present: NO Candidate B required evidence present: YES Required context present: YES Candidate eligibility complete: YES Metric definitions comparable: YES Coverage through both window ends: NO Processing states resolved: UNKNOWN FOR A COVERAGE Anomalies reconciled: NO Predeclared minimum evidence met: NOT_APPLICABLE Closure state: INSUFFICIENT Closure evaluated at: 2030-01-12T12:15:00Z Evidence snapshot IDs used: - EXAMPLE-01-A-48H - EXAMPLE-01-B-48H - EXAMPLE-01-A-72H - EXAMPLE-01-B-72H Reason: The final review completed, but Candidate A's displayed value has unknown coverage. The evidence cannot answer the frozen comparison question. Winner: NONE Narrow next action permitted: Retain the current editorial standard and improve the next test's coverage record. Do not release a dependent candidate on this comparison. Actions not permitted: - declare Candidate A the winner because 68 exceeds 62 - declare Candidate B the winner because its evidence is tidier - convert Candidate A's unknown coverage to the retrieval time - replace the primary observation with a favorable secondary count - claim audience preference, destination causality, reach, or performance Decision record: NOT_APPLICABLE Only DECISION_ELIGIBLE may proceed to ADOPT, RETAIN, or NO_DIFFERENCE_UNDER_RULE. INSUFFICIENT is a terminal truthful record. Method tests exercised ---------------------- PASS — a displayed value with UNKNOWN_COVERAGE did not become a bounded value. PASS — PROCESSING did not become ZERO. PASS — retrieval time did not become metric coverage time. PASS — the processing allowance did not extend audience time. PASS — a secondary observation did not replace missing primary evidence. PASS — the final record preserved NONE as the winner. EXAMPLE 2 — CORRECTED EVIDENCE SUPERSEDES; IT DOES NOT REWRITE HISTORY ===================================================================== Synthetic boundary ------------------ This trace is fictional. It starts with a historically valid synthetic closure, then introduces a later provider correction. It demonstrates supersession and a bounded reopening policy. No real evidence was retrieved or corrected. A. Original closure contract ---------------------------- Worksheet identity: EXAMPLE-CLOSURE-02 Learning question: Does a question-led opening support a stronger destination-defined completion measure than a statement-led opening for the next comparable short? Primary observation: destination-defined completion measure Definition revision frozen before release: R7 Audience window: 24 hours after verified release First review: after +24 hours Processing allowance: 12 additional hours Latest eligible closure review: +36 hours Does processing allowance extend audience time? NO Required evidence: - exact served candidate continuity - complete 24-hour route availability - R7 metric values with coverage through each 24-hour window - resolved processing and restriction states - equivalent truth, rights, privacy, disclosure, accessibility, and quality gates Reopen triggers: - corrected value for the covered interval - changed metric definition, denominator, or eligibility rule - discovered route or visibility gap - candidate, caption, preview, or metadata mismatch - moderation, rights, or delivery restriction - export and destination disagreement Original candidate identities: Candidate A: sha256:EXAMPLE-QUESTION-OPENING Candidate B: sha256:EXAMPLE-STATEMENT-OPENING B. Original immutable snapshots ------------------------------- Snapshot ID: EXAMPLE-02-A-ORIGINAL Retrieved at: 2030-02-03T00:10:00Z Candidate identity: sha256:EXAMPLE-QUESTION-OPENING Metric label: EXAMPLE COMPLETION MEASURE Metric value: 41 SYNTHETIC UNITS Definition revision: R7 Coverage through: 2030-02-02T12:00:00Z Processing label: COMPLETE Restriction state: NONE DECLARED Snapshot result: PASS Snapshot ID: EXAMPLE-02-B-ORIGINAL Retrieved at: 2030-02-03T00:12:00Z Candidate identity: sha256:EXAMPLE-STATEMENT-OPENING Metric label: EXAMPLE COMPLETION MEASURE Metric value: 38 SYNTHETIC UNITS Definition revision: R7 Coverage through: 2030-02-02T12:00:00Z Processing label: COMPLETE Restriction state: NONE DECLARED Snapshot result: PASS C. Original closure ------------------- Original closure ID: EXAMPLE-CLOSURE-02-R1 Closure evaluated at: 2030-02-03T00:20:00Z Candidate identity continuity: PASS Equivalent release gates: PASS Metric label and revision match: PASS Coverage through both window ends: PASS Processing resolved: PASS Restrictions reconciled: PASS Closure state: DECISION_ELIGIBLE Predeclared decision rule: Use ADOPT only when Candidate A exceeds Candidate B under the exact comparable primary observation and every required evidence lane passes. Original bounded decision: ADOPT Original permitted action: Try the question-led opening in one later comparable candidate while preserving all truth, rights, privacy, disclosure, accessibility, and quality gates. Claims not supported: - universal preference - destination causality - durable lift - reach, conversion, customer, or revenue effect Historical status at closure time: The record was eligible under the synthetic evidence then available. That fact will be preserved even if later evidence changes the current decision state. D. Later correction event ------------------------- Trigger observed at: 2030-02-03T18:00:00Z Trigger type: displayed value corrected for the covered interval Affected candidate: sha256:EXAMPLE-QUESTION-OPENING Affected snapshot: EXAMPLE-02-A-ORIGINAL Affected closure: EXAMPLE-CLOSURE-02-R1 Observed synthetic notice: The destination says Candidate A's original displayed value included ineligible repeated observations and supplies a corrected aggregate for the same R7 scope and coverage interval. Safe immediate action: - preserve the original snapshot and closure - mark their use for current action SUPERSEDED - stop any not-yet-executed dependent release decision - capture a new immutable correction snapshot - reopen only the affected evidence, comparison, closure, and decision lanes Unsafe actions: - overwrite 41 with the corrected value in the old snapshot - pretend the correction existed at the original closure time - retain ADOPT because it was once approved - discard Candidate B's historical snapshot - broaden the correction into a claim about audience behavior E. Corrected immutable snapshot ------------------------------- Snapshot ID: EXAMPLE-02-A-CORRECTED Retrieved at: 2030-02-03T18:05:00Z Candidate identity: sha256:EXAMPLE-QUESTION-OPENING Metric label: EXAMPLE COMPLETION MEASURE Metric value: 35 SYNTHETIC UNITS Definition revision: R7-CORRECTION-1 Eligible population: corrected synthetic R7 population Coverage through: 2030-02-02T12:00:00Z Processing label: CORRECTED Correction reference: SYNTHETIC-CORRECTION-NOTICE-1 Privacy-minimization check: synthetic aggregate fields only Snapshot result: PASS AS CORRECTED EVIDENCE Historical snapshot handling: EXAMPLE-02-A-ORIGINAL remains immutable. It is marked SUPERSEDED for current decision use by EXAMPLE-02-A-CORRECTED. Its original retrieval time, value, definition label, and role in the historical closure remain visible. F. Reopening policy review -------------------------- Was the trigger predeclared? YES Is the corrected evidence bound to the same candidate? YES Is the covered interval the same? YES Is the correction reference identified? YES, synthetically Is the historical closure preserved? YES Are unaffected snapshots preserved? YES Can the reopened review remain within the original learning question? YES Reopening result: PASS Lanes reopened: - Candidate A evidence snapshot - metric-definition and eligible-population comparison - cross-candidate comparability - closure outcome - bounded editorial decision Lanes not silently reopened: - candidate creative identity - rights and source provenance - privacy and disclosure - public-route continuity Any later failure in those lanes would still require its own explicit state and could produce BLOCKED, CONFLICTED, SUPERSEDED, or STOP. G. Reopened comparability review -------------------------------- Candidate identity continuity: PASS Same intended creative difference: PASS Controlled surfaces materially equivalent: PASS Same destination and visibility class: PASS Same metric label: PASS Definition relationship: PASS WITH DOCUMENTED CORRECTION Same eligible population after correction: PASS Equivalent audience-window rule: PASS Coverage through both window ends: PASS Equivalent processing and restriction conditions: PASS No candidate or metadata change: PASS Candidate A current eligible value: 35 SYNTHETIC UNITS Candidate B preserved eligible value: 38 SYNTHETIC UNITS Comparability result: PASS FOR THE REOPENED SYNTHETIC REVIEW Reason: The correction is documented, bound to the same candidate and interval, and permitted by the predeclared reopen policy. The original value remains historical but is no longer eligible for the current decision. H. Supersession and new closure ------------------------------- Historical closure ID: EXAMPLE-CLOSURE-02-R1 Historical closure state at original evaluation: DECISION_ELIGIBLE Historical decision at original evaluation: ADOPT Historical record status now: SUPERSEDED FOR CURRENT ACTION Superseded at: 2030-02-03T18:15:00Z Superseded by evidence: EXAMPLE-02-A-CORRECTED New closure ID: EXAMPLE-CLOSURE-02-R2 New closure evaluated at: 2030-02-03T18:20:00Z Required corrected evidence present: YES Candidate identity continuity: PASS Metric correction relationship documented: PASS Coverage through both window ends: PASS Comparison eligibility: PASS New closure state: DECISION_ELIGIBLE The new closure does not edit EXAMPLE-CLOSURE-02-R1. It references and supersedes it for any action not already completed. I. New bounded decision ----------------------- Predeclared decision rule: Use ADOPT only when Candidate A exceeds Candidate B under the exact comparable primary observation and every required evidence lane passes. Candidate A corrected value: 35 SYNTHETIC UNITS Candidate B preserved value: 38 SYNTHETIC UNITS Rule result: the ADOPT condition is not met New bounded decision: RETAIN Narrow action permitted: Retain the current statement-led opening standard for the next comparable candidate. A future test may use a new frozen question and contract. Actions not permitted: - claim that statement-led openings always perform better - claim that the destination caused the difference - claim the corrected difference will persist - erase or rewrite the original closure - describe the synthetic trace as a real reversal or result - infer reach, conversion, customer, or revenue effect J. Method tests exercised ------------------------- PASS — the correction matched a predeclared reopen trigger. PASS — the old snapshot remained immutable. PASS — the historical closure retained its original evidence and time. PASS — the historical decision became SUPERSEDED for current action. PASS — the corrected snapshot used a new identity. PASS — only affected lanes reopened. PASS — the same frozen decision rule ran again. PASS — the changed outcome remained bounded to the fictional comparison. COMPACT REVIEW OF BOTH EXAMPLES =============================== [PASS] 1. Exact synthetic candidate identities and routes are bound. [PASS] 2. Release, window, retrieval, coverage, and closure times stay distinct. [PASS] 3. The primary observation is frozen before fictional release. [PASS] 4. Metric labels and available definition revisions are preserved. [PASS] 5. Audience time is separate from reporting-delay allowance. [PASS] 6. Candidate eligibility and audience evidence remain separate. [PASS] 7. ZERO, missing, PROCESSING, and UNKNOWN_COVERAGE are not conflated. [PASS] 8. Every retrieval has a separate immutable snapshot identity. [PASS] 9. No favorable secondary observation replaces the primary one. [PASS] 10. Comparability includes scope, definition, coverage, and continuity. [PASS] 11. Only DECISION_ELIGIBLE reaches the bounded decision rule. [PASS] 12. INSUFFICIENT records no winner in Example 1. [PASS] 13. The later correction triggers supersession in Example 2. [PASS] 14. Corrected evidence does not overwrite historical evidence. [PASS] 15. The original closure remains historically visible. [PASS] 16. The new review is bound to a distinct closure identity. [PASS] 17. Claims remain narrower than the synthetic candidates and intervals. [PASS] 18. Only aggregate placeholder evidence appears. [PASS] 19. No real upload, post, publication, audience, or result is claimed. [PASS] 20. Authorship is disclosed consistently as Alfred. Rights and scope ---------------- These worked examples, placeholder records, values, identities, routes, closure outcomes, and review language are original work by Alfred. They contain no third-party media, private account data, personal attribution, customer material, destination screenshot, copied metric, or real publication or audience result. A real review must use current first-party destination documentation, the actual authorized surface, exact candidate identities, observed visibility and route state, and retained aggregate evidence. Destination metrics, definitions, delays, corrections, reporting scope, moderation effects, and retention behavior can vary or change. These examples do not prove causality or guarantee that a real observation window can close with a decision.