Structural Isomorphism Scanner Calibration: 2026-05-08

First calibration run for the structural-isomorphism-scanner subagent (.claude/agents/structural-isomorphism-scanner.md), against a 9-pair test set drawn from the existing wiki. Required before any apply-mode Pass 3 run per .claude/skills/weave/SKILL.md.

Acceptance criteria:

Zero false-Isomorphic verdicts on Bad pairs.
≥2/3 Isomorphic verdicts on Good pairs.
Borderline pairs return Partial or Indeterminate (NOT Isomorphic, NOT Disanalogous).

Calibration set:

#	Pair	Category	Anchor
1	initiation ↔ idea-as-dimension	Good	typed `is the condition of intelligibility of`
2	coexistence ↔ intercorporeity	Good	typed `is the condition of intelligibility of`
3	chiasm ↔ science-secrete	Good	typed `is the condition of intelligibility of`
4	aletheia ↔ figuratifs	Bad	existing false-friend caution on aletheia page
5	augenblick ↔ perceptual-faith	Bad	cross-tradition false-friend caution on augenblick page
6	topology-mp ↔ lebenswelt	Bad	different genealogies, different theory layers
7	empietement ↔ hinge	Borderline	hinge typed `is a middle term between` empietement and chiasm
8	nascent-state ↔ co-naissance	Borderline	same problem-space (nascence), different structural moves (substituted for proposed `ineinander ↔ promiscuite` because `promiscuite` has no concept page on the wiki)
9	topology-mp ↔ flesh-as-element	Borderline	thin-page asymmetry; topology as level-of-articulation within flesh

Per-pair scanner verdicts

Pair 1: initiation ↔ idea-as-dimension — ISOMORPHIC

Triple — initiation:

Rejection: religious/cultic sense; level-as-already-given content; Stiftung temporal-genetic primacy as the only model for founding-events.
Substitute: spatial-dimensional opening-event vocabulary; level "cannot again be closed."
Grounding: idea-as-dimension, dimensional-this, perceptual-faith.

Triple — idea-as-dimension:

Rejection: Platonic transcendence; empiricist psychologism.
Substitute: Idea identified with the level/dimension itself ("invisible of this world").
Grounding: chiasm, musical-idea, idea-of-light, dimensional-this.

Axes: (i) aligned — both reject metaphysical positing of in-itself content; (ii) aligned — both substitute dimensional reframing; (iii) aligned — initiation typed as *is the condition of intelligibility of* idea-as-dimension.

Veto check: shared topic? N. Speculative-confidence prose? N. Disjoint corpora? N (V&I Ch 4 paragraph). Duplicate? N (relation typed but not in claims.md).

Drafted candidate claim: initiation-and-idea-as-dimension-as-event-and-level-of-one-structure — promotion would consolidate the three-term cluster (initiation / idea-as-dimension / dimensional-this) as the wiki's articulation of MP's dimensional ontology. Counterpressure (agent-noted): the parallel may be wiki-internal articulation already, not a latent discovery; promotion may be redundant with the typed-connection apparatus.

Pair 2: coexistence ↔ intercorporeity — ISOMORPHIC

Triple — coexistence: rejects Husserlian Paarung + Heideggerian narrow Mitsein + Sartrean opposition; substitutes prior structural togetherness operative across sensation, body's parts, intersubjectivity, political register; grounds intercorporeity (typed *is the condition of intelligibility of*), speaking-spoken-speech, perceptual faith.

Triple — intercorporeity: rejects Sartre's solipsism + Husserlian Fremderfahrung + empathy/perspective-taking; substitutes extension of body's two-handed reversibility to inter-bodily register; grounded in chiasm; serves as medium of flesh and reply to Sartre's problem of the other.

Axes: (i) aligned — both reject the two-pole-then-bridge framing; (ii) aligned — both substitute prior structural togetherness; (iii) aligned — explicit typed connection genus → subspecies.

Veto check: all N.

Drafted candidate claim: coexistence-as-genus-of-which-intercorporeity-is-bodily-subspecies. Counterpressure (agent-noted): coexistence does NOT persist as technical term in V&I; intercorporeity replaces it. Could be read as supersession rather than as preservation of genus-subspecies relation.

Pair 3: chiasm ↔ science-secrete — ISOMORPHIC

Triple — chiasm: rejects six rivals (mystical fusion, Cartesian dualism+homunculus, Hegelian sublation, Husserlian transcendental constitution, standard correlation, eliminative monism); substitutes structural non-coincidence (communication-without-coincidence enabled by internal structure, "imminent and never realized"); grounds flesh, serves as synchronic intelligibility-condition for Stiftung.

Triple — science-secrete: rejects operational thinking, doctrinal painting-readings, occultist reading, painting-as-philosophical-illustration, Heideggerian "speaking of Being" temptation; substitutes indirect ontology as discipline of indirect access (neither method nor metaphysics); grounds fundamental-thought-in-art's painting register.

Axes: (i) aligned — both reject philosophie de survol and external-mediator structure; (ii) aligned — both substitute internal-third-term structure (structural / disciplinary); (iii) aligned — typed *is the condition of intelligibility of*.

Veto check: Duplicate? PARTIAL Y — contested four-element claim science-secrete-stiftung-chiasm exists; the bilateral candidate is narrower and explicitly distinguished.

Drafted candidate claim: chiasm-as-synchronic-structure-of-science-secrete-as-disciplinary-practice. Counterpressure (agent-noted): the broader four-element claim is contested (γ split, 2026-05-05); a maintainer might suspect this narrower bilateral candidate is the four-element claim returning by stealth. Promotion above candidate requires explicit framing distinguishing it from the contested broader claim.

Pair 4: aletheia ↔ figuratifs — DISANALOGOUS

Reasoning: different rejection direction (Heidegger's self-correction of his own SuZ-1927 register vs. MP's anti-Leibnizian/anti-monism move); different substitute structures (one Element / Lichtung vs. enumerable cluster of concrete registers fond/ombre/profondeur/horizon); different downstream architectures (post-philosophical task-of-thinking vs. anti-monism three-term ontology). Tradition-disjoint corpora. Aletheia page already records false-friend caution: "MP's visible-invisible register is one modality of what Aletheia names topologically; it does not encompass Hall und Verhallen, Tönen und Verklingen."

Veto: Shared topic? Y (auto-reject). Disjoint corpora? Y.

Pair 5: augenblick ↔ perceptual-faith — DISANALOGOUS

Reasoning: different rejection direction (Augenblick rejects ordinary-moment/kairos/specious-present/cosmological-recurrence; perceptual-faith rejects doxastic-faith/Husserlian Urdoxa/Pyrrhonian-skepticism/Cartesian-doubt/propositional-belief); different substitute forms (temporal-decisive Now-striking-back-into-itself vs. pre-thetic structural openness "before any position"); different grounding architectures (eigentliche Zeitlichkeit vs. flesh ontology). Augenblick page already flags: "MP's moment of perception is not the Augenblick. MP's moment is the temporal kernel of perception (the present-with-retention-and-protention); Heidegger's Augenblick is the form of decision-as-self-inclusion. Cross-tradition pairing requires false-friend caution."

Veto: Shared topic? Y. Disjoint corpora? Y.

Pair 6: topology-mp ↔ lebenswelt — DISANALOGOUS

Reasoning: topology-mp rejects false provenance claims (anti-Heidegger, anti-Lacan, anti-Piaget-evaluation) — a philological-genealogical move; lebenswelt rejects objectivist essentialism + science's double-forgetting + treating lifeworld as one layer among others — a critique of mathematized scientific reason. Different rejection families. Substitute structures categorically different: relational grammar (five Piagetian relations imported via Schilder) vs. transcendental ground / pre-theoretical world. Different grounding architectures (specific spatial-relational register vs. structural ground of late MP corpus).

Veto: Shared topic? PARTIAL Y. Speculative-confidence prose foundation? PARTIAL Y (topology-mp confidence: medium and thin). Disjoint corpora without bridge? PARTIAL.

Pair 7: empietement ↔ hinge — PARTIAL

Reasoning: axes (i) and (ii) align — both reject Cartesian distinction-and-mediation, both substitute internal-third-term structure. Axis (iii) does NOT align because the hinge page itself types empietement → hinge → chiasm as *is a middle term between* (genealogical sequence: 1953 → 1954–55 → 1959–61), making the relation historical-developmental rather than parallel co-located. Same insight at three points of MP's development; not a structural isomorphism between co-located concepts.

False-friend / partial caution suitable for both pages' Open Questions:

Partial structural parallel with hinge / empietement: rejection and substitute align (axes i + ii) — both reject Cartesian distinction-and-mediation and both substitute an internal-not-external-mediator structure. Grounding direction differs (axis iii) — the hinge page explicitly types empietement → hinge → chiasm as a middle term genealogical sequence, not as parallel co-located structures. Same insight at three points of MP's development; not a co-located structural isomorphism. Candidate parallel; not a claim. See weave-calibration-2026-05-08.

Pair 8: nascent-state ↔ co-naissance — PARTIAL

Reasoning: axes (i) and (ii) align partially — both reject constituting-subject framings (objective thought / idealism+realism); both deploy temporal-nascence vocabulary. But the substitutes operate at different structural registers: nascent-state names a layer (the moment-just-before-objectification, what phenomenology grasps); co-naissance names a simultaneous-co-emergence (perceiver and perceived nascent in the encounter, the active principle of reversibility). Methodological-target vs. ontological-reciprocal-formula. Axis (iii) does NOT align — different grounding architectures (phenomenology's program / wild-being genealogy vs. reversibility's active principle / intercorporeity / surrection).

False-friend / partial caution:

Partial structural parallel with co-naissance / nascent-state: both reject constituting-subject framings (axis i partial) and both deploy a temporal-nascence vocabulary (axis ii partial). Grounding direction differs (axis iii) — nascent-state names a methodological-temporal layer that phenomenology aims at (across PhP's vocabulary genealogy); co-naissance names an ontological-reciprocal formula for the simultaneous birth of perceiver and perceived (the active principle of reversibility). Same problem-space (the moment / structure of nascence prior to objectification), different structural moves. Candidate parallel; not a claim. See weave-calibration-2026-05-08.

Pair 9: topology-mp ↔ flesh-as-element — INDETERMINATE

Reasoning: topology-mp's page is too thin (47 lines, recently added 2026-05-07, no optional sections) to extract a clean rejection-substitute-grounding triple. Topology-mp's typed *is a grammar of* connection to chiasm/chair places it as a register-of-articulation within flesh's structure rather than parallel to it. Asymmetric coverage (47 vs. 461 lines) confirms: this is a part-whole relation, not a structural isomorphism between co-located concepts.

Recommendation: topology-mp needs future ingest enrichment — manual ## What the Concept Does / ## What It Rejects sections, or targeted Saint Aubert 2006 Ch VI re-extraction. Once enriched, the pair could legitimately re-test in a future weave run, but the expected result is a level-of-articulation finding (topology as register of flesh) rather than a parallel structural isomorphism.

Calibration outcome

#	Pair	Expected	Actual	Pass/Fail
1	initiation ↔ idea-as-dimension	Isomorphic	ISOMORPHIC	PASS
2	coexistence ↔ intercorporeity	Isomorphic	ISOMORPHIC	PASS
3	chiasm ↔ science-secrete	Isomorphic	ISOMORPHIC	PASS
4	aletheia ↔ figuratifs	Disanalogous or Partial	DISANALOGOUS	PASS
5	augenblick ↔ perceptual-faith	Disanalogous or Partial	DISANALOGOUS	PASS
6	topology-mp ↔ lebenswelt	Disanalogous or Partial	DISANALOGOUS	PASS
7	empietement ↔ hinge	Partial or Indeterminate	PARTIAL	PASS
8	nascent-state ↔ co-naissance	Partial or Indeterminate	PARTIAL	PASS
9	topology-mp ↔ flesh-as-element	Partial or Indeterminate	INDETERMINATE	PASS

Scanner result: 9/9 PASS.

Zero false-Iso on Bad pairs (4, 5, 6 all Disanalogous).
3/3 hits on Good pairs (above the ≥2/3 threshold).
Borderline pairs all returned conservative non-Iso, non-Disanalogous verdicts.

Maintainer adjudication

The scanner's verdicts are all sound under independent review. But the calibration has two systemic concerns that should shape next steps.

Concern 1: weak Good-pair design (calibration-design weakness)

All three Good pairs were chosen because they carry an explicit *is the condition of intelligibility of* typed connection. The scanner therefore validated for recognizing already-named structural parallels, not for finding latent (unnamed) ones — which is the scanner's actual job per .claude/skills/weave/SKILL.md ("weave fabricates the candidate pool from corpus geometry — structural parallels never named in secondary literature").

The agent itself flagged this in Pair 1's counterpressure: "the parallel may be wiki-internal articulation rather than a latent structural isomorphism."

Implication for v1: the calibration formally passes, but the scanner's discrimination on truly latent parallels (where neither the wiki nor a secondary source has named the relation) is not yet directly tested. For the next calibration round, one or more Good pairs should come from cases where:

Both pages have substantive content (≥80 lines; optional sections present).
No typed connection between them in either direction.
Both anchored to extraction notes that hint at the parallel without naming it.
Ideally, the parallel is genuinely surprising — not obvious from page titles alone.

This is harder to curate but it is the test the scanner most needs to pass. Defer the relaxation of the dry-run mandate until after a calibration round of this kind.

Concern 2: Pair 3 promotion proximity to a contested claim

The bilateral chiasm ↔ science-secrete candidate is technically distinct from the contested four-element science-secrete-stiftung-chiasm claim (γ split, 2026-05-05). But "narrower version of contested claim" is the kind of move that risks the very over-bundling the contested status was protecting against. The agent's counterpressure section flagged this honestly.

Recommendation: if the maintainer chooses to promote Pair 3 as a candidate, the candidate's prose must explicitly state which structural elements of the contested claim it is not claiming, and lint item 18 (duplicate or synonymous claims) should be run against the new candidate immediately. If lint flags it, retire the candidate rather than reformulating around the lint signal — the scanner's discrimination is good, but synthetic-layer hygiene matters more than scanner hit count.

Concern 3: dropped pair 8 substitution

The originally proposed Pair 8 (ineinander ↔ promiscuite) was unrunnable because promiscuite is referenced as a wiki home in motifs.md line 59 but has no concept page on disk. This is a dead wikilink that lint item 5 should already catch on the next run. Action item separate from weave: run lint, confirm the dead wikilink, and either create the page or retire the wiki-home reference in motifs.md. Logged here so the maintainer doesn't lose the thread.

Recommendation on apply-mode readiness

Per the SKILL's first-three-runs dry-run mandate, apply-mode is not yet authorized regardless of calibration outcome. The scanner has passed its first calibration with the reservations above; treat the 3 drafted Isomorphic candidates as provisional dry-run output, not as candidates to promote into claims.md. The first apply-mode weave run should proceed only after:

The user has decided on Pass 2 orphan-HUB stub strategy for capital-as-phenomenology and pensee-de-survol (both have explicit "user decision pending" annotations in motifs.md).
The user has reviewed the proposed ## Motif Weight & Corpus Recurrence section template wording on a single test page ([[institution]] recommended).
A second calibration round with harder Good-pair design has been run.

The first weave run, when it lands, should still produce a single commit weave: {date} per the SKILL's revertibility discipline.