Full mechanism reclassification results

Date: 2026-07-29

Status: complete; replacement evidence ledger passes its pre-registered stability rule; maintainer adjudication still required before vocabulary adoption or migration.

Protocol: full mechanism reclassification protocol

Frozen surface: manifest TSV

All boundary records: 387-vote TSV

Result

The replacement run is usable: 127/129 rows (98.4%) have a stable 2/3-or-better majority, above the pre-registered 90% floor. 101 rows (78.3%) are unanimous, 26 (20.2%) have a 2/3 contested majority, and 2 (1.6%) are UNSTABLE.

The stable replacement ledger contains 65 EX, 40 OP, 18 OTHER, and 4 EN rows. The EX and OP cohorts are both large, cross-file, and highly reproducible: EX spans 58 source artifacts with 56 unanimous rows; OP spans 33 source artifacts with 31 unanimous rows. Both classes occur in active mechanism and deferred-grounds origins. The full run therefore restores a strong evidential case for an explanatory/operational distinction while replacing the prior row assignments.

This result does not adopt explained-by or operates-through. Under the protocol, the 26 contested rows require explicit maintainer adjudication before migration, the two unstable rows have no successor candidate, and EN/OTHER rows remain outside either proposed successor.

Acceptance and distribution

measure	count	share	rule/readout
Stable majority	127/129	98.4%	pass: ≥90%
Unanimous	101/129	78.3%	reproducible candidate dispositions
Contested 2/3	26/129	20.2%	maintainer adjudication required
UNSTABLE	2/129	1.6%	no successor candidate

Majority by origin

origin	EX	OP	EN	OTHER	UNSTABLE	total
mechanism	37	33	2	8	2	82
grounds-deferred	28	7	2	10	0	47
total	65	40	4	18	2	129

Reproducibility by majority class

majority class	stable rows	source artifacts	unanimous	contested
EX	65	58	56	9
OP	40	33	31	9
EN	4	3	2	2
OTHER	18	18	12	6

Rebaseline and attrition

The frozen current surface is 129 tuples: 82 active mechanism edges and 47 surviving members of the exact deferred-grounds cohort. All 129 source and target artifacts resolved at dispatch, all received three votes, and a post-run tuple check found no runtime deletion or movement.

Relative to the prior 128-row review, all 79 old active-mechanism tuples remain and three new active tuples arrived. Two of the 49 deferred-grounds tuples disappeared before dispatch, leaving 47. These two are pre-dispatch attrition and were excluded rather than silently retained:

• prior EX: kb/notes/a-consumption-channel-delivers-force-without-the-history-that.md → kb/notes/definitions/reflective-system.md;
• prior OP: kb/notes/a-consumption-channel-delivers-force-without-the-history-that.md → kb/notes/a-derived-copy-of-recomputable-truth-must-be-checked-or-absent.md.

The three additions and all line movements are recorded in the protocol and manifest.

Comparison with the prior review

prior disposition	EX	OP	EN	OTHER	UNSTABLE	total
EX	29	5	0	6	0	40
OP	25	33	0	11	2	71
EN	6	1	3	0	0	10
X	2	1	0	0	0	3
P	1	0	0	0	0	1
E	0	0	0	1	0	1
NEW	2	0	1	0	0	3

On the 111 surviving rows previously assigned EX or OP, only 62 (55.9%) retain the same majority. Direct EX↔OP reversal affects 30 (27.0%), boundary movement to EN/OTHER affects 17 (15.3%), and 2 (1.8%) are unstable. Prior EX reproduces on 29/40 rows; prior OP reproduces on 33/71.

The full run therefore confirms the blind sample’s central finding: the old 41/72 row ledger is not an adoption or migration basis. It also adds the evidence the sample could not provide: after every row is classified with k=3 and explicit counterfactuals, two substantial reproducible cohorts remain.

New rows

ID	edge	votes	majority	status
F032	kb/notes/reflection-buys-addressability.md:68 → kb/notes/localized-retention-pays-where-change-is-sparse-in-a-matching.md	EN/EN/EX	EN	contested
F068	kb/notes/the-bitter-lesson-selects-production-methods-not-representational.md:70 → kb/notes/localized-retention-pays-where-change-is-sparse-in-a-matching.md	EX/EX/EX	EX	unanimous
F091	kb/notes/a-consumption-channel-delivers-force-without-the-history-that.md:57 → kb/notes/llm-context-is-a-homoiconic-medium.md	EX/EX/EX	EX	unanimous

Evidence for and against retaining the split

For

• EX and OP are not marginal residue: 65 and 40 stable rows respectively, spanning 58 and 33 source artifacts.
• The distinction survives origin changes. EX includes 37 active-mechanism and 28 deferred-grounds rows; OP includes 33 and 7.
• The most reproducible cores are large: 56 unanimous EX rows and 31 unanimous OP rows.
• Classifier counterfactuals repeatedly produce different maintenance consequences: rejecting an EX target reopens the source explanation; changing an OP target reopens interface, behavior, or operational fit.
• One broad current candidate would be false for a material cohort: explained-by would misstate literal execution paths, while operates-through would misstate explanatory principles.

Against and limits

• The old mechanism surface is not binary: 18 stable OTHER rows, 4 EN rows, and 2 unstable rows require other treatment.
• Twenty-six stable rows are only 2/3 majorities. The protocol deliberately withholds them from automatic migration.
• OTHER is a coarse catch-all in this run. Its rationales must be mapped to exact existing relations or prose during adjudication; an OTHER majority is not itself a successor label.
• The three classifiers per row were fresh isolated contexts but all reported the Codex/GPT-5 family. The run measures context-level independence, not cross-model-family agreement.
• The four-class prompt tests this boundary, not every neighbouring registered label simultaneously. Exact mapping to extends, exemplifies, premised-on, evidence labels, or removal still requires adjudication where relevant.

Reader and revision consequences

• EX candidate: source is explained-by target. Follow when the reader needs the account or principle that explains why/how the source occurs. Rejecting or materially revising the target prompts re-reading the source causal argument; it does not by itself assert an implementation change.
• OP candidate: source operates-through target. Follow when the reader needs the actual process, component, control path, artifact, or rule producing the source effect. Changing the target prompts interface, behavioral, or operational-fit review.

The full run supports taking these two consequences back to maintainer adjudication as distinct registered relations. It does not settle the spellings or authorize the one reference→notes pairing. Those remain maintainer choices.

Exact cohorts

• EX unanimous (56): F001, F002, F003, F007, F008, F009, F011, F012, F016, F017, F018, F019, F021, F022, F023, F024, F029, F036, F039, F049, F056, F058, F059, F064, F065, F067, F068, F069, F073, F074, F075, F078, F079, F081, F082, F083, F087, F089, F090, F091, F093, F096, F097, F098, F099, F100, F101, F105, F106, F110, F114, F117, F120, F123, F126, F128
• EX contested (9): F026, F027, F035, F040, F044, F050, F052, F063, F071
• OP unanimous (31): F004, F005, F010, F014, F020, F025, F033, F034, F037, F038, F041, F045, F047, F054, F055, F060, F061, F070, F072, F076, F077, F084, F085, F108, F109, F118, F121, F124, F125, F127, F129
• OP contested (9): F013, F046, F053, F057, F086, F104, F111, F116, F119
• EN unanimous (2): F094, F107
• EN contested (2): F032, F092
• OTHER unanimous (12): F015, F030, F031, F043, F051, F062, F066, F095, F103, F112, F113, F122
• OTHER contested (6): F006, F042, F048, F080, F088, F102
• UNSTABLE (2): F028, F115

Per-row replacement ledger

Votes are class/confidence. Full boundary-test answers and justifications for every vote are in the linked 387-vote TSV.

ID	origin	source → target	prior	A	B	C	majority	status
F001	mechanism	kb/notes/stale-self-description-conceals-its-own-staleness.md:70 → kb/notes/reflection-buys-addressability.md	OP	EX/high	EX/medium	EX/high	EX	unanimous
F002	grounds-deferred	kb/notes/agent-memory-needs-discoverable-composable-trusted-knowledge-under.md:36 → kb/notes/claw-learning-loops-must-improve-action-capacity-not-just-retrieval.md	P	EX/high	EX/high	EX/high	EX	unanimous
F003	mechanism	kb/notes/raw-accumulation-does-not-create-usable-memory.md:31 → kb/notes/constraining-and-extraction-both-trade-generality-for-reliability.md	EX	EX/medium	EX/medium	EX/medium	EX	unanimous
F004	mechanism	kb/notes/links-encode-conditional-possibilities-not-obligations.md:107 → kb/notes/inbound-and-outbound-links-serve-asymmetric-reader-needs.md	OP	OP/high	OP/medium	OP/high	OP	unanimous
F005	mechanism	kb/notes/a-proposal-selection-loop-requires-search-evaluation-and-retention.md:102 → kb/notes/theory-and-methodology-form-a-two-layer-execution-system.md	OP	OP/high	OP/high	OP/high	OP	unanimous
F006	mechanism	kb/notes/selective-revision-needs-a-faithful-rationale-not-just-a-legible-one.md:51 → kb/notes/false-positive-generation-is-filtered-before-retention.md	EX	OTHER/medium	OTHER/medium	OP/medium	OTHER	contested
F007	grounds-deferred	kb/notes/runtime-structure-determines-governance-control-surfaces.md:56 → kb/notes/inspectable-artifact-not-supervision-defeats-the-blackbox-problem.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F008	mechanism	kb/notes/the-framework-is-often-larger-than-the-durable-contribution.md:63 → kb/notes/conjecture-is-seeing-the-particular-as-an-instance-of-the-general.md	X	EX/high	EX/high	EX/medium	EX	unanimous
F009	mechanism	kb/notes/pointer-design-tradeoffs-in-progressive-disclosure.md:80 → kb/notes/theory-and-methodology-form-a-two-layer-execution-system.md	OP	EX/medium	EX/medium	EX/medium	EX	unanimous
F010	mechanism	kb/notes/commitment-not-derivation-creates-new-ground-truth.md:92 → kb/notes/progressive-constraining-commits-only-after-patterns-stabilize.md	OP	OP/high	OP/high	OP/high	OP	unanimous
F011	mechanism	kb/notes/design-for-the-first-time-human-except-on-access-cost.md:28 → kb/notes/agents-navigate-by-deciding-what-to-read-next.md	OP	EX/medium	EX/medium	EX/medium	EX	unanimous
F012	grounds-deferred	kb/notes/scenario-decomposition-drives-architecture.md:86 → kb/notes/skills-derive-from-methodology.md	OP	EX/high	EX/high	EX/high	EX	unanimous
F013	mechanism	kb/notes/verification-needs-a-typed-target-before-it-needs-an-oracle.md:58 → kb/notes/axes-of-artifact-analysis.md	OP	OP/high	EX/medium	OP/high	OP	contested
F014	mechanism	kb/notes/definitions/context-engineering.md:64 → kb/notes/llm-context-is-composed-without-scoping.md	OP	OP/high	OP/high	OP/high	OP	unanimous
F015	mechanism	kb/notes/the-framework-is-often-larger-than-the-durable-contribution.md:62 → kb/notes/system-definition-artifacts-are-crystallized-reasoning-under-context.md	OP	OTHER/medium	OTHER/high	OTHER/high	OTHER	unanimous
F016	grounds-deferred	kb/notes/mechanistic-constraints-make-popperian-kb-recommendations-actionable.md:56 → kb/notes/llm-context-is-composed-without-scoping.md	OP	EX/high	EX/high	EX/high	EX	unanimous
F017	grounds-deferred	kb/notes/llm-executed-methodologies-are-metacircular-interpreters.md:38 → kb/notes/definitions/system-definition-artifact.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F018	grounds-deferred	kb/notes/formal-systems-assess-explanatory-reach-through-causal-and-proof.md:47 → kb/notes/definitions/representational-form.md	EX	EX/high	EX/high	EX/medium	EX	unanimous
F019	mechanism	kb/notes/verifiable-subroles-before-reviewer-identity.md:58 → kb/notes/the-augmentation-automation-boundary-is-discrimination-not-accuracy.md	EN	EX/high	EX/high	EX/high	EX	unanimous
F020	mechanism	kb/notes/the-practical-scheduler-is-the-host-language.md:66 → kb/notes/agent-is-a-tool-loop.md	OP	OP/high	OP/high	OP/high	OP	unanimous
F021	grounds-deferred	kb/notes/a-derived-copy-of-recomputable-truth-must-be-checked-or-absent.md:70 → kb/notes/text-testing-framework.md	OP	EX/high	EX/high	EX/medium	EX	unanimous
F022	mechanism	kb/notes/verifiable-subroles-before-reviewer-identity.md:60 → kb/notes/error-correction-works-above-chance-oracles-with-decorrelated-checks.md	OP	EX/high	EX/medium	EX/high	EX	unanimous
F023	grounds-deferred	kb/notes/increasing-computational-autonomy-relocates-human-effort.md:59 → kb/notes/the-boundary-of-automation-is-the-boundary-of-verification.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F024	mechanism	kb/notes/frontloading-spares-execution-context.md:56 → kb/notes/frontloading-is-partial-evaluation-not-divide-and-conquer.md	EX	EX/high	EX/medium	EX/medium	EX	unanimous
F025	mechanism	kb/notes/llm-context-is-composed-without-scoping.md:75 → kb/notes/agent-statelessness-means-the-context-engine-should-inject-context.md	OP	OP/high	OP/medium	OP/high	OP	unanimous
F026	grounds-deferred	kb/notes/context-contamination-operates-below-an-agents-compliance-reasoning.md:57 → kb/notes/agent-orchestration-needs-coordination-guarantees-not-just.md	OP	OTHER/medium	EX/medium	EX/medium	EX	contested
F027	mechanism	kb/notes/rlm-tendril-and-llm-do-place-symbolic-work-at-different-persistence.md:53 → kb/notes/deploy-time-learning-is-the-missing-middle.md	OP	OTHER/medium	EX/high	EX/high	EX	contested
F028	mechanism	kb/notes/mechanistic-constraints-make-popperian-kb-recommendations-actionable.md:60 → kb/notes/deploy-time-learning-is-the-missing-middle.md	OP	OP/high	EX/medium	OTHER/high	UNSTABLE	UNSTABLE
F029	mechanism	kb/notes/session-history-should-not-be-the-default-next-context.md:82 → kb/notes/theory-and-methodology-form-a-two-layer-execution-system.md	OP	EX/medium	EX/medium	EX/medium	EX	unanimous
F030	grounds-deferred	kb/notes/ephemerality-is-safe-where-embedded-operational-knowledge-has-low.md:64 → kb/notes/codification-and-relaxing-navigate-the-bitter-lesson-boundary.md	EX	OTHER/medium	OTHER/high	OTHER/high	OTHER	unanimous
F031	mechanism	kb/notes/design-for-the-first-time-human-except-on-access-cost.md:26 → kb/notes/feasibility-is-the-heaviest-forks-net-load.md	OP	OTHER/high	OTHER/high	OTHER/high	OTHER	unanimous
F032	mechanism	kb/notes/reflection-buys-addressability.md:68 → kb/notes/localized-retention-pays-where-change-is-sparse-in-a-matching.md	NEW	EN/high	EN/medium	EX/high	EN	contested
F033	mechanism	kb/notes/feasibility-is-the-heaviest-forks-net-load.md:38 → kb/notes/session-history-should-not-be-the-default-next-context.md	OP	OP/high	OP/high	OP/high	OP	unanimous
F034	mechanism	kb/notes/bounded-context-orchestration-model.md:92 → kb/notes/llm-context-is-composed-without-scoping.md	OP	OP/high	OP/high	OP/high	OP	unanimous
F035	grounds-deferred	kb/notes/compiling-coordination-preserves-primitive-not-aggregate-authority.md:58 → kb/notes/bounded-context-orchestration-model.md	OP	OP/high	EX/high	EX/medium	EX	contested
F036	mechanism	kb/notes/a-proximate-target-is-checked-for-achievement-not-for-warrant.md:69 → kb/notes/weakly-discriminated-qualities-tend-to-be-underselected.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F037	mechanism	kb/notes/structure-inference-needs-capture-at-the-decision-surface.md:37 → kb/notes/spec-mining-as-codification.md	OP	OP/high	OP/high	OP/high	OP	unanimous
F038	grounds-deferred	kb/notes/ad-hoc-prompts-extend-the-system-without-schema-changes.md:64 → kb/notes/llm-context-is-composed-without-scoping.md	OP	OP/high	OP/high	OP/high	OP	unanimous
F039	grounds-deferred	kb/notes/reverse-compression-is-when-llm-output-expands-without-adding.md:43 → kb/notes/information-value-is-observer-relative.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F040	mechanism	kb/notes/retrieval-failure-is-reflection-failure.md:40 → kb/notes/stale-indexes-are-worse-than-no-indexes.md	EX	EX/high	EX/high	OP/medium	EX	contested
F041	mechanism	kb/notes/entropy-management-must-scale-with-generation-throughput.md:31 → kb/notes/spec-mining-as-codification.md	OP	OP/high	OP/high	OP/high	OP	unanimous
F042	grounds-deferred	kb/notes/semantic-review-catches-content-errors-that-structural-validation.md:49 → kb/notes/text-testing-framework.md	OP	OTHER/high	OP/medium	OTHER/high	OTHER	contested
F043	mechanism	kb/notes/bounded-context-orchestration-model.md:97 → kb/notes/theory-and-methodology-form-a-two-layer-execution-system.md	OP	OTHER/high	OTHER/high	OTHER/high	OTHER	unanimous
F044	mechanism	kb/notes/error-messages-that-teach-are-a-constraining-technique.md:28 → kb/notes/frontloading-spares-execution-context.md	OP	OP/high	EX/high	EX/high	EX	contested
F045	mechanism	kb/notes/feasibility-is-the-heaviest-forks-net-load.md:36 → kb/notes/frontloading-spares-execution-context.md	OP	OP/high	OP/high	OP/high	OP	unanimous
F046	grounds-deferred	kb/notes/codify-versus-llm-decision-heuristics.md:120 → kb/notes/ephemeral-computation-prevents-accumulation.md	EX	OP/medium	OP/high	EX/high	OP	contested
F047	mechanism	kb/notes/definitions/context-engineering.md:63 → kb/notes/instruction-specificity-should-match-loading-frequency.md	OP	OP/high	OP/high	OP/high	OP	unanimous
F048	grounds-deferred	kb/notes/llm-context-is-composed-without-scoping.md:72 → kb/notes/instruction-specificity-should-match-loading-frequency.md	OP	OTHER/high	OTHER/high	OP/medium	OTHER	contested
F049	mechanism	kb/notes/system-definition-artifacts-are-crystallized-reasoning-under-context.md:60 → kb/notes/llm-context-is-a-homoiconic-medium.md	OP	EX/high	EX/high	EX/medium	EX	unanimous
F050	mechanism	kb/notes/orchestration-strategies-and-run-state-have-opposite-persistence.md:51 → kb/notes/verifiability-gradient.md	OP	OTHER/high	EX/medium	EX/medium	EX	contested
F051	grounds-deferred	kb/notes/definitions/reach-assessment.md:66 → kb/notes/abstract-an-experience-only-when-you-can-state-the-boundary.md	EX	OTHER/high	OTHER/medium	OTHER/medium	OTHER	unanimous
F052	mechanism	kb/notes/technical-constraints-make-kb-objective-choice-engineering.md:84 → kb/notes/oracle-strength-spectrum.md	EX	EX/medium	OTHER/medium	EX/high	EX	contested
F053	mechanism	kb/notes/orchestration-strategies-and-run-state-have-opposite-persistence.md:49 → kb/notes/deploy-time-learning-is-the-missing-middle.md	OP	OP/medium	EX/medium	OP/high	OP	contested
F054	mechanism	kb/notes/verifiable-subroles-before-reviewer-identity.md:62 → kb/notes/structured-output-is-easier-for-humans-to-review.md	OP	OP/high	OP/medium	OP/high	OP	unanimous
F055	mechanism	kb/notes/stale-self-description-conceals-its-own-staleness.md:71 → kb/notes/link-graph-plus-timestamps-enables-make-like-staleness-detection.md	OP	OP/medium	OP/high	OP/medium	OP	unanimous
F056	grounds-deferred	kb/notes/an-outcome-check-licenses-replay-a-rule-needs-the-process-verified.md:39 → kb/notes/first-principles-reasoning-selects-for-explanatory-reach-over.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F057	grounds-deferred	kb/notes/artifacts-produced-from-sources-need-lineage-recorded-at-the-source.md:45 → kb/notes/frontloading-spares-execution-context.md	OP	OP/high	OP/medium	OTHER/high	OP	contested
F058	grounds-deferred	kb/notes/adversarial-loop-can-reconstruct-the-writing-is-thinking-filter.md:28 → kb/notes/llm-generation-relaxes-goals-where-human-writing-stalls.md	OP	EX/high	EX/high	EX/high	EX	unanimous
F059	grounds-deferred	kb/notes/reflection-makes-retained-lessons-second-order.md:49 → kb/notes/reflection-buys-addressability.md	OP	EX/high	EX/high	EX/high	EX	unanimous
F060	mechanism	kb/notes/use-tests-a-decomposition-locally-rationale-makes-transfer-testable.md:63 → kb/notes/scenario-decomposition-drives-architecture.md	OP	OP/medium	OP/medium	OP/medium	OP	unanimous
F061	mechanism	kb/notes/selective-revision-needs-a-faithful-rationale-not-just-a-legible-one.md:46 → kb/notes/reflection-makes-retained-lessons-second-order.md	OP	OP/high	OP/high	OP/high	OP	unanimous
F062	mechanism	kb/notes/task-fitted-structure-costs-cross-task-reuse.md:71 → kb/notes/orchestration-strategies-and-run-state-have-opposite-persistence.md	EX	OTHER/high	OTHER/high	OTHER/high	OTHER	unanimous
F063	grounds-deferred	kb/notes/abstract-an-experience-only-when-you-can-state-the-boundary.md:61 → kb/notes/diagnostic-richness-constrains-outer-loop-learning-quality.md	EN	EX/medium	EX/high	EN/medium	EX	contested
F064	mechanism	kb/notes/continual-learning-open-problem-is-behaviour-not-knowledge.md:28 → kb/notes/llm-context-is-a-homoiconic-medium.md	OP	EX/high	EX/high	EX/high	EX	unanimous
F065	grounds-deferred	kb/notes/an-action-model-matters-only-through-its-consumption-path.md:24 → kb/notes/axes-of-artifact-analysis.md	OP	EX/high	EX/high	EX/high	EX	unanimous
F066	grounds-deferred	kb/notes/evidence/single-artifact-review-bundles-still-cut-claude-costs-substantially.md:72 → kb/notes/oracle-strength-spectrum.md	E	OTHER/high	OTHER/high	OTHER/high	OTHER	unanimous
F067	mechanism	kb/notes/theory-mediated-self-improvement-needs-interpretation-and-retention.md:59 → kb/notes/reflection-makes-retained-lessons-second-order.md	OP	EX/high	EX/high	EX/high	EX	unanimous
F068	mechanism	kb/notes/the-bitter-lesson-selects-production-methods-not-representational.md:70 → kb/notes/localized-retention-pays-where-change-is-sparse-in-a-matching.md	NEW	EX/high	EX/high	EX/high	EX	unanimous
F069	mechanism	kb/notes/a-methodology-governs-its-own-extension-only-as-far-as-it-settles.md:70 → kb/notes/the-boundary-of-automation-is-the-boundary-of-verification.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F070	mechanism	kb/notes/reflection-makes-retained-lessons-second-order.md:53 → kb/notes/retrieval-failure-is-reflection-failure.md	OP	OP/high	OP/high	OP/high	OP	unanimous
F071	mechanism	kb/notes/weakly-discriminated-qualities-tend-to-be-underselected.md:81 → kb/notes/error-correction-works-above-chance-oracles-with-decorrelated-checks.md	EN	OTHER/high	EX/high	EX/high	EX	contested
F072	grounds-deferred	kb/notes/llm-context-is-a-homoiconic-medium.md:45 → kb/notes/methodology-enforcement-is-constraining.md	EX	OP/high	OP/high	OP/high	OP	unanimous
F073	mechanism	kb/notes/decomposition-heuristics-for-bounded-context-scheduling.md:91 → kb/notes/theory-and-methodology-form-a-two-layer-execution-system.md	OP	EX/medium	EX/medium	EX/medium	EX	unanimous
F074	mechanism	kb/notes/improving-an-agentic-system-crosses-the-prose-symbolic-boundary.md:45 → kb/notes/scheduler-llm-separation-exploits-an-error-correction-asymmetry.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F075	mechanism	kb/notes/llm-executed-methodologies-are-metacircular-interpreters.md:36 → kb/notes/methodology-enforcement-is-constraining.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F076	mechanism	kb/notes/the-four-field-record-exposes-an-efficiency-security-and-sovereignty.md:46 → kb/notes/definitions/codification.md	OP	OP/high	OP/high	OP/high	OP	unanimous
F077	mechanism	kb/notes/theory-mediated-self-improvement-needs-interpretation-and-retention.md:60 → kb/notes/retrieval-failure-is-reflection-failure.md	OP	OP/high	OP/high	OP/high	OP	unanimous
F078	mechanism	kb/notes/the-augmentation-automation-boundary-is-discrimination-not-accuracy.md:56 → kb/notes/error-correction-works-above-chance-oracles-with-decorrelated-checks.md	EN	EX/high	EX/high	EX/high	EX	unanimous
F079	mechanism	kb/notes/methodological-and-computational-closure-track-different-changes.md:80 → kb/notes/reflection-buys-addressability.md	OP	EX/medium	EX/high	EX/high	EX	unanimous
F080	grounds-deferred	kb/notes/agent-memory-needs-discoverable-composable-trusted-knowledge-under.md:40 → kb/notes/agent-memory-is-a-crosscutting-concern-not-a-separable-niche.md	OP	OTHER/high	EX/high	OTHER/high	OTHER	contested
F081	mechanism	kb/notes/minimum-viable-vocabulary-is-the-naming-set-that-most-reduces.md:38 → kb/notes/conjecture-is-seeing-the-particular-as-an-instance-of-the-general.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F082	mechanism	kb/notes/generality-bought-to-avoid-counterexamples-is-paid-for-in.md:59 → kb/notes/llm-generation-confidence-tracks-typicality-not-soundness.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F083	grounds-deferred	kb/notes/an-enforced-tag-readme-is-a-moc-with-a-machine-checked-contract.md:49 → kb/notes/stale-indexes-are-worse-than-no-indexes.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F084	mechanism	kb/notes/feasibility-is-the-heaviest-forks-net-load.md:37 → kb/notes/agents-navigate-by-deciding-what-to-read-next.md	OP	OP/medium	OP/medium	OP/high	OP	unanimous
F085	mechanism	kb/notes/definitions/context-engineering.md:65 → kb/notes/agents-navigate-by-deciding-what-to-read-next.md	OP	OP/high	OP/high	OP/high	OP	unanimous
F086	mechanism	kb/notes/mechanistic-constraints-make-popperian-kb-recommendations-actionable.md:59 → kb/notes/definitions/constraining.md	OP	EX/high	OP/high	OP/high	OP	contested
F087	mechanism	kb/notes/deploy-time-learning-is-the-missing-middle.md:56 → kb/notes/llm-context-is-a-homoiconic-medium.md	OP	EX/high	EX/high	EX/high	EX	unanimous
F088	grounds-deferred	kb/notes/reasoning-production-is-not-reasoning-evaluation.md:44 → kb/notes/process-structure-and-output-structure-are-independent-levers.md	OP	OTHER/medium	OTHER/high	EX/medium	OTHER	contested
F089	grounds-deferred	kb/notes/llm-context-is-composed-without-scoping.md:78 → kb/notes/scheduler-llm-separation-exploits-an-error-correction-asymmetry.md	OP	EX/high	EX/high	EX/high	EX	unanimous
F090	grounds-deferred	kb/notes/selective-revision-needs-a-faithful-rationale-not-just-a-legible-one.md:49 → kb/notes/reasoning-production-is-not-reasoning-evaluation.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F091	mechanism	kb/notes/a-consumption-channel-delivers-force-without-the-history-that.md:57 → kb/notes/llm-context-is-a-homoiconic-medium.md	NEW	EX/high	EX/high	EX/high	EX	unanimous
F092	grounds-deferred	kb/notes/a-derived-copy-of-recomputable-truth-must-be-checked-or-absent.md:64 → kb/notes/the-boundary-of-automation-is-the-boundary-of-verification.md	EN	EN/high	EN/high	EX/high	EN	contested
F093	grounds-deferred	kb/notes/topology-isolation-and-verification-form-a-causal-chain-for-reliable.md:73 → kb/notes/synthesis-is-not-error-correction.md	EX	EX/medium	EX/medium	EX/high	EX	unanimous
F094	grounds-deferred	kb/notes/an-outcome-check-licenses-replay-a-rule-needs-the-process-verified.md:40 → kb/notes/diagnostic-richness-constrains-outer-loop-learning-quality.md	EN	EN/high	EN/high	EN/high	EN	unanimous
F095	grounds-deferred	kb/notes/compiling-coordination-preserves-primitive-not-aggregate-authority.md:61 → kb/notes/rlm-has-the-model-write-ephemeral-orchestrators-over-sub-agents.md	OP	OTHER/high	OTHER/high	OTHER/high	OTHER	unanimous
F096	mechanism	kb/notes/brainstorming-maintainability-oracles-for-agentic-development.md:191 → kb/notes/error-correction-works-above-chance-oracles-with-decorrelated-checks.md	EN	EX/high	EX/high	EX/high	EX	unanimous
F097	grounds-deferred	kb/notes/agent-memory-is-a-crosscutting-concern-not-a-separable-niche.md:66 → kb/notes/agent-runtimes-decompose-into-scheduler-context-engine-and-execution.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F098	grounds-deferred	kb/notes/automated-synthesis-is-missing-good-oracles.md:58 → kb/notes/the-boundary-of-automation-is-the-boundary-of-verification.md	X	EX/high	EX/high	EX/high	EX	unanimous
F099	grounds-deferred	kb/notes/a-derived-copy-of-recomputable-truth-must-be-checked-or-absent.md:63 → kb/notes/stale-indexes-are-worse-than-no-indexes.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F100	mechanism	kb/notes/the-four-field-record-exposes-an-efficiency-security-and-sovereignty.md:48 → kb/notes/opacity-is-a-scale-threshold.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F101	mechanism	kb/notes/spec-mining-as-codification.md:59 → kb/notes/oracle-strength-spectrum.md	OP	EX/high	EX/high	EX/high	EX	unanimous
F102	grounds-deferred	kb/notes/a-consumption-channel-delivers-force-without-the-history-that.md:59 → kb/notes/the-four-field-record-exposes-an-efficiency-security-and-sovereignty.md	OP	EX/medium	OTHER/high	OTHER/high	OTHER	contested
F103	mechanism	kb/notes/orchestration-strategies-and-run-state-have-opposite-persistence.md:50 → kb/notes/definitions/codification.md	OP	OTHER/high	OTHER/high	OTHER/high	OTHER	unanimous
F104	mechanism	kb/notes/rlm-tendril-and-llm-do-place-symbolic-work-at-different-persistence.md:52 → kb/notes/ephemeral-computation-prevents-accumulation.md	EX	OP/high	EX/high	OP/high	OP	contested
F105	mechanism	kb/notes/false-positive-generation-is-filtered-before-retention.md:69 → kb/notes/the-boundary-of-automation-is-the-boundary-of-verification.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F106	grounds-deferred	kb/notes/reasoning-production-is-not-reasoning-evaluation.md:43 → kb/notes/the-augmentation-automation-boundary-is-discrimination-not-accuracy.md	EX	EX/high	EX/medium	EX/medium	EX	unanimous
F107	mechanism	kb/notes/an-outcome-check-licenses-replay-a-rule-needs-the-process-verified.md:38 → kb/notes/abstract-an-experience-only-when-you-can-state-the-boundary.md	EN	EN/high	EN/high	EN/high	EN	unanimous
F108	grounds-deferred	kb/notes/evaluation-automation-is-phase-gated-by-comprehension.md:50 → kb/notes/spec-mining-as-codification.md	OP	OP/high	OP/high	OP/high	OP	unanimous
F109	mechanism	kb/notes/technical-constraints-make-kb-objective-choice-engineering.md:81 → kb/notes/codify-versus-llm-decision-heuristics.md	EX	OP/high	OP/high	OP/high	OP	unanimous
F110	grounds-deferred	kb/notes/adversarial-loop-can-reconstruct-the-writing-is-thinking-filter.md:30 → kb/notes/the-boundary-of-automation-is-the-boundary-of-verification.md	OP	EX/high	EX/high	EX/high	EX	unanimous
F111	mechanism	kb/notes/reflective-coverage-is-graded-across-representational-forms.md:100 → kb/notes/unified-calling-conventions-enable-bidirectional-refactoring.md	OP	OP/high	OP/medium	OTHER/high	OP	contested
F112	mechanism	kb/notes/rlm-tendril-and-llm-do-place-symbolic-work-at-different-persistence.md:54 → kb/notes/verifiability-gradient.md	EX	OTHER/high	OTHER/high	OTHER/high	OTHER	unanimous
F113	grounds-deferred	kb/notes/conjecture-is-seeing-the-particular-as-an-instance-of-the-general.md:62 → kb/notes/minimum-viable-vocabulary-is-the-naming-set-that-most-reduces.md	EX	OTHER/high	OTHER/high	OTHER/high	OTHER	unanimous
F114	grounds-deferred	kb/notes/brainstorming-maintainability-oracles-for-agentic-development.md:188 → kb/notes/weakly-discriminated-qualities-tend-to-be-underselected.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F115	mechanism	kb/reference/adr/040-scripts-directory-is-the-accumulation-substrate-for-ad-hoc-tooling.md:46 → kb/notes/spec-mining-as-codification.md	OP	OP/high	OTHER/high	EX/medium	UNSTABLE	UNSTABLE
F116	mechanism	kb/notes/bounded-context-orchestration-model.md:90 → kb/notes/frontloading-spares-execution-context.md	X	OTHER/medium	OP/high	OP/high	OP	contested
F117	grounds-deferred	kb/notes/frontloading-spares-execution-context.md:53 → kb/notes/agent-context-is-constrained-by-soft-degradation-not-hard-token-limits.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F118	mechanism	kb/notes/enforcement-without-structured-recovery-is-incomplete.md:69 → kb/notes/error-correction-works-above-chance-oracles-with-decorrelated-checks.md	OP	OP/high	OP/medium	OP/high	OP	unanimous
F119	grounds-deferred	kb/notes/always-loaded-context-mechanisms-in-agent-harnesses.md:89 → kb/notes/frontloading-spares-execution-context.md	OP	OP/medium	EX/high	OP/high	OP	contested
F120	mechanism	kb/notes/raw-accumulation-does-not-create-usable-memory.md:32 → kb/notes/conjecture-is-seeing-the-particular-as-an-instance-of-the-general.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F121	mechanism	kb/notes/orchestration-strategies-and-run-state-have-opposite-persistence.md:48 → kb/notes/ephemeral-computation-prevents-accumulation.md	OP	OP/high	OP/high	OP/high	OP	unanimous
F122	mechanism	kb/notes/link-graph-plus-timestamps-enables-make-like-staleness-detection.md:61 → kb/notes/traversal-improvements-should-be-deferred-via-logging-to-avoid-mid.md	OP	OTHER/high	OTHER/high	OTHER/high	OTHER	unanimous
F123	mechanism	kb/notes/agent-context-is-constrained-by-soft-degradation-not-hard-token-limits.md:62 → kb/notes/indirection-is-costly-in-llm-instructions.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F124	mechanism	kb/notes/llm-mediated-schedulers-are-a-degraded-variant-of-the-clean-model.md:35 → kb/notes/session-history-should-not-be-the-default-next-context.md	OP	OP/high	OP/high	OP/high	OP	unanimous
F125	mechanism	kb/notes/trace-extracted-memory-earns-authority-per-operation-not-at-capture.md:63 → kb/notes/abstract-an-experience-only-when-you-can-state-the-boundary.md	OP	OP/high	OP/high	OP/high	OP	unanimous
F126	grounds-deferred	kb/notes/codification-and-relaxing-navigate-the-bitter-lesson-boundary.md:55 → kb/notes/bitter-lesson-selects-against-unearned-reach-not-against-structure.md	EX	EX/high	EX/high	EX/high	EX	unanimous
F127	mechanism	kb/notes/brainstorming-maintainability-oracles-for-agentic-development.md:190 → kb/notes/evaluation-automation-is-phase-gated-by-comprehension.md	EN	OP/high	OP/high	OP/high	OP	unanimous
F128	mechanism	kb/notes/theory-mediated-learning-may-improve-sample-efficiency-under-shifts.md:112 → kb/notes/warranted-autonomy-is-bounded-by-oracle-domain.md	EN	EX/high	EX/medium	EX/high	EX	unanimous
F129	grounds-deferred	kb/notes/orchestration-strategies-and-run-state-have-opposite-persistence.md:46 → kb/notes/bounded-context-orchestration-model.md	EX	OP/high	OP/high	OP/high	OP	unanimous

Contested and unstable diagnostics

Every non-unanimous row is repeated here with the three classifier justifications. The full counterfactual records remain in the vote TSV.

F006 — OTHER (contested), prior EX

kb/notes/selective-revision-needs-a-faithful-rationale-not-just-a-legible-one.md:51 → kb/notes/false-positive-generation-is-filtered-before-retention.md

• A — OTHER/medium: The target categorizes the failure but does not explain or implement the rationale-specific causal chain.
• B — OTHER/medium: The source analogizes an unassessed rationale to false-positive acceptance, but its wrong-premise mechanism is self-contained.
• C — OP/medium: The target's evaluation-and-retention control path is how an unassessed revision acquires force, though the source also gives a self-contained error account.

F013 — OP (contested), prior OP

kb/notes/verification-needs-a-typed-target-before-it-needs-an-oracle.md:58 → kb/notes/axes-of-artifact-analysis.md

• A — OP/high: Artifact analysis is a concrete ontology through which heterogeneous retained artifacts become typed verification targets.
• B — EX/medium: The target is the conceptual ontology that explains the heterogeneous authority paths the source says verification must distinguish.
• C — OP/high: The target is the classification artifact through which heterogeneous verification targets can be typed.

F026 — EX (contested), prior OP

kb/notes/context-contamination-operates-below-an-agents-compliance-reasoning.md:57 → kb/notes/agent-orchestration-needs-coordination-guarantees-not-just.md

• A — OTHER/medium: The target situates the phenomenon but does not supply its specific causal account or control path.
• B — EX/medium: The target supplies the broader uncoordinated-composition account under which the observed drift is contamination from absent isolation.
• C — EX/medium: The target generalizes the observed stance drift as contamination caused by composition without a scoping guarantee.

F027 — EX (contested), prior OP

kb/notes/rlm-tendril-and-llm-do-place-symbolic-work-at-different-persistence.md:53 → kb/notes/deploy-time-learning-is-the-missing-middle.md

• A — OTHER/medium: The target names the learning category into which Tendril falls but does not explain why the compared systems chose different persistence boundaries.
• B — EX/high: The target explains how cross-session executable artifacts constitute deploy-time behavior change.
• C — EX/high: Deploy-time learning explains why Tendril's retained generated capability is durable behavior change rather than ordinary memory.

F028 — UNSTABLE (unstable), prior OP

kb/notes/mechanistic-constraints-make-popperian-kb-recommendations-actionable.md:60 → kb/notes/deploy-time-learning-is-the-missing-middle.md

• A — OP/high: Deploy-time artifact updates are the operational path by which a rejected interpretation becomes a lasting behavior change.
• B — EX/medium: Deploy-time learning explains how a rejected interpretation retained in an instruction or test can alter later-session behavior.
• C — OTHER/high: The target names the broader adaptation timescale into which the source's persistent correction falls.

F032 — EN (contested), prior NEW

kb/notes/reflection-buys-addressability.md:68 → kb/notes/localized-retention-pays-where-change-is-sparse-in-a-matching.md

• A — EN/high: The target gives the change-topology conditions under which selective revision and rollback actually remain local.
• B — EN/medium: The target supplies the applicability conditions under which selective revision and related addressability advantages remain local.
• C — EX/high: The target causally bounds the payoff by explaining how edit and validation locality arise.

F035 — EX (contested), prior OP

kb/notes/compiling-coordination-preserves-primitive-not-aggregate-authority.md:58 → kb/notes/bounded-context-orchestration-model.md

• A — OP/high: The target specifies the scheduler and bounded-call control path whose selection logic is lifted into the compiled workflow.
• B — EX/high: The target explains the choose-next-step structure whose symbolic extraction the source calls compilation.
• C — EX/medium: The target explains the coordination structure whose selection logic is lifted into a persistent symbolic artifact.

F040 — EX (contested), prior EX

kb/notes/retrieval-failure-is-reflection-failure.md:40 → kb/notes/stale-indexes-are-worse-than-no-indexes.md

• A — EX/high: The target explains how an authoritative stale index makes missing knowledge invisible by stopping the fallback search.
• B — EX/high: A seemingly exhaustive but incomplete index explains how discovery is suppressed and the reflective causal wire silently breaks.
• C — OP/medium: A trusted incomplete index is the artifact that literally suppresses fallback discovery and breaks retrieval.

F042 — OTHER (contested), prior OP

kb/notes/semantic-review-catches-content-errors-that-structural-validation.md:49 → kb/notes/text-testing-framework.md

• A — OTHER/high: The target classifies the checks as Level B but is not the mechanism that performs them.
• B — OP/medium: The target is the testing process artifact that the source operationalizes as a Level B semantic-review layer.
• C — OTHER/high: The target supplies a testing taxonomy and catalogue into which the source places its checks, not an explanation or operative mechanism.

F044 — EX (contested), prior OP

kb/notes/error-messages-that-teach-are-a-constraining-technique.md:28 → kb/notes/frontloading-spares-execution-context.md

• A — OP/high: A teaching error message performs frontloading by delivering the remediation instead of making the agent derive it.
• B — EX/high: Frontloading explains why inserting the known remediation into the error channel removes repeated runtime diagnosis work.
• C — EX/high: Frontloading explains why inserting the fix avoids an in-context diagnosis step.

F046 — OP (contested), prior EX

kb/notes/codify-versus-llm-decision-heuristics.md:120 → kb/notes/ephemeral-computation-prevents-accumulation.md

• A — OP/medium: The target's discard path literally produces the no-accumulation cost used by the heuristic.
• B — OP/high: Generate, execute, and discard is the literal operating path that makes the leave-for-LLM side lose cross-run accumulation.
• C — EX/high: The target explains the structural cost of leaving work ephemeral: useful patterns cannot accumulate across runs.

F048 — OTHER (contested), prior OP

kb/notes/llm-context-is-composed-without-scoping.md:72 → kb/notes/instruction-specificity-should-match-loading-frequency.md

• A — OTHER/high: The target organizes progressive disclosure, while the source diagnoses absent local scope and prescribes fresh frames; neither produces the other.
• B — OTHER/high: The target is a policy for reducing always-loaded material, not the cause, control path, or prerequisite of flat context.
• C — OP/medium: The loading hierarchy is the operational policy that selects which instruction bodies enter the otherwise global context.

F050 — EX (contested), prior OP

kb/notes/orchestration-strategies-and-run-state-have-opposite-persistence.md:51 → kb/notes/verifiability-gradient.md

• A — OTHER/high: The target supplies placement vocabulary rather than the test-gated promotion mechanism.
• B — EX/medium: The verifiability gradient explains why test-gated promotion makes a reusable strategy cheaper and safer to trust.
• C — EX/medium: The verifiability gradient explains why tested promotion can support tighter trust and reuse than ephemeral code.

F052 — EX (contested), prior EX

kb/notes/technical-constraints-make-kb-objective-choice-engineering.md:84 → kb/notes/oracle-strength-spectrum.md

• A — EX/medium: Oracle strength explains why descriptive, prescriptive, and theoretical objectives admit different degrees and costs of verification.
• B — OTHER/medium: The target names grades used to describe profile-specific judges but does not produce or causally explain their differing evidence surfaces.
• C — EX/high: Oracle strength is the general principle explaining differences among profile-specific judges.

F053 — OP (contested), prior OP

kb/notes/orchestration-strategies-and-run-state-have-opposite-persistence.md:49 → kb/notes/deploy-time-learning-is-the-missing-middle.md

• A — OP/medium: Test-gated promotion of recurring scheduler fragments is a concrete deploy-time learning loop through durable repository artifacts.
• B — EX/medium: The target gives the general account under which promoting tested control fragments into a durable repository constitutes cross-session learning.
• C — OP/high: Test-gated strategy promotion is literally a deploy-time learning loop over durable artifacts.

F057 — OP (contested), prior OP

kb/notes/artifacts-produced-from-sources-need-lineage-recorded-at-the-source.md:45 → kb/notes/frontloading-spares-execution-context.md

• A — OP/high: Frontloading literally produces the source-dependent, use-shaped artifact for which the source note requires lineage.
• B — OP/medium: Frontloading is a literal artifact-production path that can hide source material from the consumer while preserving a maintenance dependency on it.
• C — OTHER/high: The target supplies an illustrative production mode, not the explanation, operating path, or prerequisite for change-time lineage signaling.

F063 — EX (contested), prior EN

kb/notes/abstract-an-experience-only-when-you-can-state-the-boundary.md:61 → kb/notes/diagnostic-richness-constrains-outer-loop-learning-quality.md

• A — EX/medium: The target explains how retained traces and failure detail let a learner infer mechanisms and boundaries instead of learning only from outcomes.
• B — EX/high: The target explains how preserved traces and diagnostics let a learner infer the condition that bounds a lesson.
• C — EN/medium: Diagnostic evidence is a necessary available input to boundary formation, while the source supplies the abstraction rule itself.

F071 — EX (contested), prior EN

kb/notes/weakly-discriminated-qualities-tend-to-be-underselected.md:81 → kb/notes/error-correction-works-above-chance-oracles-with-decorrelated-checks.md

• A — OTHER/high: The target addresses how to mitigate weak selection, not why weakly discriminated qualities are underselected.
• B — EX/high: The target explains when several weak quality checks can exert stronger selection pressure together.
• C — EX/high: The target explains when multiple partial quality checks can overcome weak individual discrimination instead of amplifying shared blind spots.

F080 — OTHER (contested), prior OP

kb/notes/agent-memory-needs-discoverable-composable-trusted-knowledge-under.md:40 → kb/notes/agent-memory-is-a-crosscutting-concern-not-a-separable-niche.md

• A — OTHER/high: The target locates the note's artifact-quality triad inside a broader memory-system decomposition rather than causing or enabling it.
• B — EX/high: The target explains why discoverability, composability, and trust cover remembered material while memory remains a crosscutting system concern.
• C — OTHER/high: The target situates the source's remembered-material triad inside a broader storage, activation, and learning stack.

F086 — OP (contested), prior OP

kb/notes/mechanistic-constraints-make-popperian-kb-recommendations-actionable.md:59 → kb/notes/definitions/constraining.md

• A — EX/high: Constraining provides the general semantic mechanism by which a correction rules out future readings.
• B — OP/high: Constraining is the literal operation by which a rejected interpretation becomes a durable instruction test.
• C — OP/high: Constraining is the process through which a rejected interpretation becomes a durable narrowing of future behavior.

F088 — OTHER (contested), prior OP

kb/notes/reasoning-production-is-not-reasoning-evaluation.md:44 → kb/notes/process-structure-and-output-structure-are-independent-levers.md

• A — OTHER/medium: The target offers an analogous distinction between process and result structure, but it does not cause the production-evaluation gap.
• B — OTHER/high: The target helps name a process-versus-result distinction but neither produces nor explains the evaluator's conclusion-agreement failure.
• C — EX/medium: The general independence principle explains why a correct destination cannot verify the submitted route.

F092 — EN (contested), prior EN

kb/notes/a-derived-copy-of-recomputable-truth-must-be-checked-or-absent.md:64 → kb/notes/the-boundary-of-automation-is-the-boundary-of-verification.md

• A — EN/high: Verification availability is the condition that permits automation of the re-derive-and-compare branch; otherwise the copy must be omitted.
• B — EN/high: Verification availability is the necessary condition separating enforceable checked copies from the omission-only case.
• C — EX/high: Verification cost explains the boundary between enforceable copies and judgment-heavy managed staleness.

F102 — OTHER (contested), prior OP

kb/notes/a-consumption-channel-delivers-force-without-the-history-that.md:59 → kb/notes/the-four-field-record-exposes-an-efficiency-security-and-sovereignty.md

• A — EX/medium: The target supplies the general security account that the source refines into authorization bound to exact content and use.
• B — OTHER/high: The source develops one security boundary and borrows rollback vocabulary from the triad; the triad does not cause that boundary.
• C — OTHER/high: The target supplies the larger security and sovereignty framing that the source specializes into an authorization-to-force boundary.

F104 — OP (contested), prior EX

kb/notes/rlm-tendril-and-llm-do-place-symbolic-work-at-different-persistence.md:52 → kb/notes/ephemeral-computation-prevents-accumulation.md

• A — OP/high: Ephemeral generate-execute-discard is the operational path that gives RLM simplicity while preventing cross-run accumulation.
• B — EX/high: The target explains how discarding generated orchestration removes lifecycle burden and simultaneously prevents cross-run learning.
• C — OP/high: Discarding task-local code is the literal process that separates RLM's nonaccumulating boundary from Tendril's persistence.

F111 — OP (contested), prior OP

kb/notes/reflective-coverage-is-graded-across-representational-forms.md:100 → kb/notes/unified-calling-conventions-enable-bidirectional-refactoring.md

• A — OP/high: The target supplies a concrete interface-level transfer path across neural and symbolic implementations while leaving authority and lineage separate.
• B — OP/medium: The target is a concrete interface path across neural and symbolic implementations, while coverage, authority, and lineage remain separate obligations.
• C — OTHER/high: Unified calling is one cross-form mapping affordance, not an explanation or requirement for reflective coverage.

F115 — UNSTABLE (unstable), prior OP

kb/reference/adr/040-scripts-directory-is-the-accumulation-substrate-for-ad-hoc-tooling.md:46 → kb/notes/spec-mining-as-codification.md

• A — OP/high: The ADR directly adopts spec mining's observe-stability-then-codify rule as the operational trigger for promoting a script.
• B — OTHER/high: Moving an existing symbolic script into a packaged command reuses spec mining's stability signal but is not itself behavior-to-spec codification.
• C — EX/medium: Spec mining explains why repeated unchanged core logic is evidence that an ad hoc script is ready for durable command promotion.

F116 — OP (contested), prior X

kb/notes/bounded-context-orchestration-model.md:90 → kb/notes/frontloading-spares-execution-context.md

• A — OTHER/medium: Frontloading is a one-step instance subsumed by the iterative model, not a dependency or mechanism that produces the whole model.
• B — OP/high: Frontloading is the literal one-step scheduler action that removes already-solvable work before a bounded LLM call.
• C — OP/high: Frontloading is the one-step scheduling path that the orchestration loop generalizes.

F119 — OP (contested), prior OP

kb/notes/always-loaded-context-mechanisms-in-agent-harnesses.md:89 → kb/notes/frontloading-spares-execution-context.md

• A — OP/medium: The target supplies the precompute-and-insert operation that build-time configuration injection performs.
• B — EX/high: Frontloading explains why pre-resolving installation values saves live execution context.
• C — OP/high: Frontloading is the literal operational process used by build-time and session-start configuration injection.

Classifier provenance

Requested model: no explicit override; each pass used a fresh default Codex collaboration subagent in an isolated context. Actual model information reported by each pass follows verbatim:

batch	pass	reported model information
1	A	Codex agent based on GPT-5.
1	B	Codex agent based on GPT-5.
1	C	Codex agent based on GPT-5.
2	A	Codex, an agent based on GPT-5; no more specific model identifier was exposed.
2	B	Codex agent based on GPT-5; exact model identifier is not exposed.
2	C	Codex, based on GPT-5.
3	A	Codex agent based on GPT-5; no more specific model identifier was exposed in this run.
3	B	Codex, based on GPT-5.
3	C	Codex, based on GPT-5; the exact model identifier was not exposed.
4	A	Codex agent based on GPT-5.
4	B	Codex agent based on GPT-5.
4	C	Codex agent based on GPT-5; exact deployed model identifier not exposed.

No stronger provider, checkpoint, or parameter provenance was exposed, so none is claimed.

Changed paths and verification

This run adds:

• mechanism-full-reclassification-protocol.md;
• mechanism-full-reclassification-manifest.tsv;
• mechanism-full-reclassification-votes.tsv;
• mechanism-full-reclassification-results.md.

It changes no corpus edge, collection contract, catalogue entry, ADR, durable instruction, or prior workshop result. The result, protocol, manifest, and vote ledger are checked for completeness and whitespace; the Markdown artifacts are validated with commonplace-validate.