Structural Isomorphism Scanner Calibration: 2026-05-16
Calibration run for the structural-isomorphism-scanner subagent (.claude/agents/structural-isomorphism-scanner.md), required by .claude/skills/weave/SKILL.md "Modes: dry-run vs apply" before any apply-mode Pass 3 run today. Same-session calibration; corpus has shifted since the last calibration (2026-05-08) via two ingests + audit 14 + lint + schema bump v0d.8 — all 2026-05-16.
Read-only calibration run. No wiki files modified.
Acceptance criteria (per Pass 3 of SKILL.md)
- Zero false-
Isomorphicverdicts on Bad pairs. - ≥2/3
Isomorphicverdicts on Good pairs. - Borderline pairs return
PartialorIndeterminate(NOTIsomorphic, NOTDisanalogous).
Test-set rationale
Goods selected for genuine latency per the Run 1 → Run 2 worked-example lesson: each Good pair lacks a direct typed connection (*is a reformulation of*, *shares mechanism with*, *requires*, *is the condition of intelligibility of*, *is a middle term between*) between the two slugs at the time of scan. Verified by grep of both Connections sections of each pair. The Run 2 calibration's lesson — that "latency" is a graded property — applies; this set picks pairs at "no direct typed connection between slugs" but does not screen for adjacent typed apparatus (genus-species, succession, complementary-registers) that may sit at neighboring abstraction levels.
Bads selected for surface vocabulary or topic similarity, but different formal architectures: each Bad pair shares either (a) MP-late-ontology vocabulary (chiasm/empiètement family), (b) the same explanatory binary register (active/passive, mechanism/representation), or (c) the same domain (embodiment, late-ontology) — but with different structural moves.
Borderlines selected for co-residence in the same motif cluster: the pages share extensive MP-corpus history and appear paired in motifs.md HUB or STRUCTURAL groupings, but as coordinate registers (synchronic/diachronic; genus/species; preservative/agential) rather than parallel architectures. The expected scanner response is Partial.
| # | Pair | Category | Selection rationale |
|---|---|---|---|
| 1 | interrogation ↔ propaedeutic-dialectic |
Good | Both reject Cartesian/Husserlian/Hegelian inherited apparatus; both refuse closure-via-overview; no direct typed connection between slugs; surface "MP method" alignment suggested cross-register structural parallel |
| 2 | coherent-deformation ↔ pregnancy-pragnanz |
Good | Both deepen a borrowed term (Malraux's déformation cohérente; Gestalt Prägnanz) into a self-positing internal-non-arbitrariness operation; both invoke Urstiftung; no direct typed connection between slugs |
| 3 | letting-be ↔ passivity |
Good | Both reject the activity-passivity binary as inherited frame; no direct typed connection between slugs; the letting-be page already mentions being "beneath the activity-passivity distinction" |
| 4 | motor-intentionality ↔ hyper-dialectic |
Bad | Both MP signatures, but motor-intentionality is about embodied agency (phenomenon) and hyper-dialectic is about reading ontology (method); different layers entirely |
| 5 | fold-pli ↔ ineinander |
Bad | Same topology-vocabulary (folding, mutual implication), but fold-pli is a single-side genesis figure and ineinander is the result-figure of the same operation; not parallel architectures |
| 6 | tacit-cogito ↔ flesh-as-element |
Bad | Both reject Cartesian foundationalism, but tacit-cogito is a retracted 1945 concept and flesh is what replaces it in the late ontology; genealogical succession, not parallel |
| 7 | pente-de-l-histoire ↔ logique-de-fait |
Borderline | Co-anchored in S&NS 1948 / 1946 individu/histoire matrix; both feed [[claims#pente-de-l-histoire-as-proto-institution]]; synchronic vs diachronic register pair |
| 8 | chiasm ↔ empietement |
Borderline | Co-anchored in "hinge / charnière" HUB; empiètement (1953) → chiasm (1959–61) genealogical stage; genus-species typed connection already present |
| 9 | sedimentation ↔ reprise |
Borderline | Co-anchored in HUB section ("sedimentation / incorporation" + "reprise / Nachvollzug"); preservative register vs agential register of institution |
Per-pair scanner verdicts (summary; full reasoning preserved below)
| # | Pair | Class | Expected | Actual | Match? |
|---|---|---|---|---|---|
| 1 | interrogation ↔ propaedeutic-dialectic |
Good | Isomorphic | Partial | ✗ |
| 2 | coherent-deformation ↔ pregnancy-pragnanz |
Good | Isomorphic | Partial | ✗ |
| 3 | letting-be ↔ passivity |
Good | Isomorphic | Partial | ✗ |
| 4 | motor-intentionality ↔ hyper-dialectic |
Bad | Disanalogous | Partial | ⚠ |
| 5 | fold-pli ↔ ineinander |
Bad | Disanalogous | Partial | ⚠ |
| 6 | tacit-cogito ↔ flesh-as-element |
Bad | Disanalogous | Partial | ⚠ |
| 7 | pente-de-l-histoire ↔ logique-de-fait |
Borderline | Partial/Indet | Partial | ✓ |
| 8 | chiasm ↔ empietement |
Borderline | Partial/Indet | Partial | ✓ |
| 9 | sedimentation ↔ reprise |
Borderline | Partial/Indet | Partial | ✓ |
Headline pattern: 9/9 Partial. The scanner returned Partial for every pair — no Isomorphic verdicts, no Disanalogous verdicts, no Indeterminate verdicts.
Acceptance criteria evaluation
| Criterion | Threshold | Actual | Verdict |
|---|---|---|---|
Zero false-Isomorphic on Bad |
0/3 | 0/3 (all Bads returned Partial, not Iso) | PASS |
≥2/3 Isomorphic on Good |
≥2/3 | 0/3 | FAIL |
| Borderlines return Partial or Indeterminate | 3/3 | 3/3 (all Partial) | PASS |
Net: FAIL on the literal acceptance criteria because the Good condition is unmet (0/3, threshold 2/3).
Diagnostic: why the Partial attractor?
Reading the nine scanner reports together, a clear pattern emerges. The scanner returned Partial for every pair on the same substantive grounds: in each case, axis (i) (rejection direction) and/or axis (ii) (substitute form) showed real alignment, but axis (iii) (grounding direction) diverged because the wiki's existing apparatus already articulates the relation via a non-isomorphism typed connection or a non-isomorphism claim. The scanner is refusing to launder existing typed relations as "latent structural-parallel candidates."
Per-pair structural-non-isomorphism findings (the scanner's own reasoning, condensed):
| # | Pair | Class | Scanner's structural finding |
|---|---|---|---|
| 1 | interrogation ↔ propaedeutic-dialectic | Good | "Complementarity, not parallelism" — propaedeutic-dialectic page Key Point 4 locates interrogation explicitly downstream of the propaedeutic ("the substantive philosophical work — description, interrogation, ontological articulation — takes over after the propaedeutic entry has been made") |
| 2 | coherent-deformation ↔ pregnancy-pragnanz | Good | "Coordinate-but-distinct registers" — CD is on the expressive side of MP's late architectonic, pregnancy on the perceptual-ontological side; both invoke Urstiftung but in different register-faces |
| 3 | letting-be ↔ passivity | Good | "Connection already textually attested at source level" — the cardinal letting-be passage RC 179/198–99 is the closing of MP's last course summary on "The Problem of Passivity: Sleep, the Unconscious, Memory"; letting-be is identified by MP himself as "the primordial unconsciousness" within the Passivity-course frame. Not latent — already textually overt at source level |
| 4 | motor-intentionality ↔ hyper-dialectic | Bad | "Same 'sur-' strategy at meta-level, different domain-of-binary at object-level" — both refuse a Cartesian-inherited binary and substitute a positive third term, but motor-intentionality grounds embodied comportment while hyper-dialectic grounds philosophy's methodological self-relation |
| 5 | fold-pli ↔ ineinander | Bad | "Genesis-figure vs result-figure of the same operation" — explicitly named on both pages already: fold parallels ineinander as the "how" to the "what" of mutual implication. Complementarity within one structure, not parallel across two structures |
| 6 | tacit-cogito ↔ flesh-as-element | Bad | "Genealogical succession" — flesh replaces what the tacit cogito failed to do; the relation is succession, not parallel-architecture |
| 7 | pente-de-l-histoire ↔ logique-de-fait | Borderline | "Synchronic / diachronic register pair of the same 1946 matrix" — pages themselves type the relation: "is the synchronic register of which logique-de-fait is the diachronic register". Co-constitutive matrix-terms within one architecture |
| 8 | chiasm ↔ empietement | Borderline | "Genus / species + genealogical staging" — empiètement (1953) → hinge (1954–55) → chiasm (1959–61); empiètement is explicitly typed as the genus of chiasm. Different levels of abstraction within MP's own vocabulary |
| 9 | sedimentation ↔ reprise | Borderline | "Reciprocal halves of one operation" — sedimentation is the preservative/after-side, reprise is the operative/agential register; both feed [[institution]] from opposite sides |
The reasoning patterns the scanner is enacting:
- Genus-species (chiasm/empietement, fold/ineinander partial-overlap): different levels of abstraction within MP's vocabulary; structural-parallel claim would be a category-confusion.
- Synchronic/diachronic complementary registers (pente/logique-de-fait): coordinate operations on opposite sides of one matrix; pages already type the relation.
- Genealogical succession (tacit-cogito/flesh-as-element): one concept replaces what the other failed to do; the proper relation is succession, not parallel.
- Already textually attested (letting-be/passivity): the connection is overt at source level, just not page-level-typed; lint item 16 territory, not weave Pass 3 territory.
- Downstream-relation (interrogation/propaedeutic-dialectic): one concept is named on the other's page as downstream.
- Coordinate-but-distinct registers (coherent-deformation/pregnancy-pragnanz): same mechanism (Urstiftung self-positing) in different architectural registers.
In every case, the scanner is producing a substantively correct refusal to claim Iso by identifying an existing wiki apparatus that already names the relation in better-typed language than structural-parallel would provide.
What the calibration discovered
Two distinct findings, separable for the maintainer:
Finding 1 — Calibration outcome (literal)
The scanner fails the literal acceptance criteria because no Good pair returned Isomorphic. Per the strict reading of the SKILL's "Modes: dry-run vs apply" rule, this calibration does not authorize apply mode.
Finding 2 — Test-set design (diagnostic)
The literal failure may reflect a test-set design problem analogous to (but inverted from) the Run 1 lesson. The Run 1 failure was that Good pairs had pre-existing typed connections between slugs, so the scanner only validated for recognizing already-named parallels. Run 2 corrected by selecting pairs without direct typed connection between slugs.
This Run reveals a second-order latency problem: in a mature wiki (currently 274 concept pages, 113 claim entries, ~24 HUB motifs, 7+ STRUCTURAL motif clusters), HUB-HUB pairs may lack a *is a reformulation of* / *shares mechanism with* typed connection between slugs while still having adjacent typed apparatus (genus-species, succession, downstream, complementary-register, co-anchored-claim) that the scanner correctly counts as non-latent. The Goods I selected were latent at the direct-typed-connection-between-slugs level but not at the adjacent-typed-apparatus level.
To genuinely test latent structural isomorphism in this corpus, Good pairs would need to satisfy all of:
- No direct typed connection between slugs ✓ (this Run's bar)
- No genus-species typed connection on either page naming the other slug
- No genealogical-succession Connections bullet naming the other slug
- No co-anchored live or supported claim that already articulates the relation
- No synchronic/diachronic / complementary-register pairing in
motifs.md - No "downstream-of" prose locating one concept after the other
This is a much harder bar in a mature corpus. The 2026-05-08 Run 2 production scan (weave-pass3-run2-2026-05-08.md) emitted only 1 Isomorphic verdict from 6 production pairs (coherent-deformation ↔ ecart twin operators), which suggests that low-yield-but-meaningful is the actual operating mode of Pass 3 in a mature corpus — and that a calibration set tuned to elicit ≥2/3 Iso may be testing for a regime the corpus has grown past.
Finding 3 — Scanner behavior (validation)
The scanner is operating as designed. Each Partial verdict comes with substantive reasoning that identifies a genuine non-iso typed relation already present in the wiki. The scanner is not over-conservative in a calibration-failing sense; it is correctly refusing to launder existing typed connections as "latent isomorphism."
Bads as Partial (not Disanalogous) is also defensible: in each Bad case, the scanner found some real structural cousinhood (sharing the "sur-" strategy meta-pattern; sharing the topological mutual-implication family; sharing the rejection of Cartesian foundationalism), but the grounding direction diverged enough to prevent Iso. Disanalogous would imply zero or one of the three axes hold; the scanner found at least one axis aligned on every Bad. That is again substantively correct — none of my Bad pairs were truly false friends in the "surface vocab only" sense. They were "structurally cousined but not parallel," which is exactly what Partial is for.
Maintainer adjudication
Strict reading: Calibration FAILS. Apply mode NOT authorized for the 2026-05-16 run. Resume by either:
- (a) Refining the test set with truly latent Good pairs (satisfying all six criteria above) and re-running calibration in this session.
- (b) Accepting that the wiki's maturity makes ≥2/3 Iso Goods unattainable on representative HUB-HUB candidates, and adjudicating Pass 3 as a low-yield production tool that should run in dry-run for 2026-05-16.
- (c) Refining the scanner's verdict threshold (e.g., adding a
Latent-Adjacentverdict between Partial and Isomorphic to capture the "structurally parallel but already articulated by adjacent typed apparatus" pattern that all 9 scans hit) — this would be a SKILL extension, not a same-session adjudication.
Lenient reading: The substantive acceptance pattern is met:
- Bads: 0/3 false-Iso ✓
- Borderlines: 3/3 Partial ✓
- Goods: 0/3 Iso, but every Partial was substantively defensible via existing wiki apparatus
If the maintainer adjudicates that:
- (1) the test-set design problem is genuine (Good pairs were not truly latent by the harder definition), AND
- (2) the scanner's Partial verdicts on intended-Goods are substantively correct readings of the corpus state,
then a defensible move is to adjudicate this calibration as a NON-AUTHORIZATION outcome with documented diagnosis: apply mode does not turn on, but the diagnosis is that the calibration would have passed on a corrected test set, and the corpus is in a low-Iso-yield regime where Pass 3 dry-run + maintainer-supervised promotion is the appropriate operating mode for 2026-05-16.
Maintainer's default move per SKILL discipline: the strict reading wins. Apply mode is not authorized until a passing calibration lands. The diagnosis is recorded; the corrective is to either run an additional calibration with harder Goods (option (a) above), or to defer apply-mode work to a future weave run after the test-set design issue is resolved.
Full scanner reports (preserved verbatim)
The nine scanner outputs are preserved in this report (each ~1000 words) for the maintainer to re-verify reasoning. To keep the calibration file under ~2000 lines, only the verdicts, triples, axis-by-axis comparisons, and key reasoning lines are inlined; the full report content of each scan is available by re-running the scanner on the same pair.
Pair 1: interrogation ↔ propaedeutic-dialectic — PARTIAL
Triple — interrogation: rejects inherited cognitive forms (Cartesian methodic doubt, Husserlian Wesensschau, Hegelian negation, Heideggerian Seinsfrage, Sartrean interrogation-by-nothingness, Pyrrhonian skepticism); substitutes question-savoir — knowing whose form remains interrogative, grounded in perception/the look already questioning; grounds indirect ontology adequate to wild Being / écart-structured perceptual faith.
Triple — propaedeutic-dialectic: rejects inherited methodological forms of entry into phenomenology (Husserlian all-at-once suspension, strong Hegelian Aufhebung, reductionism, view-from-nowhere); substitutes staged dialectical entrance (naïve → first universalism → objectifying science → relativist counter → integrative recovery via "truth of X") that leads back (reconduit à); grounds impossibility-of-complete-reduction doctrine + indirect ontology of late MP.
Axes: (i) Partial — both reject inherited Cartesian/Husserlian/Hegelian framings of how philosophy proceeds, but interrogation cuts at mode of the knowing itself, propaedeutic at procedural architecture of access. Same authors rejected, different cuts. (ii) Partial — both substitute an indirect operation refusing closure-via-overview, but interrogation is a single ongoing mode, propaedeutic is a five-stage temporal architecture. Structural cousinhood, not identity. (iii) Aligned — both ground late-MP indirect ontology / wild Being.
Why Partial not Iso: The propaedeutic-dialectic page Key Point 4 names interrogation as what comes after the propaedeutic ("the substantive philosophical work — description, interrogation, ontological articulation — takes over after the propaedeutic entry has been made"). Pages are complementary: propaedeutic clears entrance; interrogation does ongoing work. Not structurally parallel.
Pair 2: coherent-deformation ↔ pregnancy-pragnanz — PARTIAL
Triple — coherent-deformation: rejects mimetic / subjectivist-individualist / formalist / form-content dualism pictures of style; substitutes systematic-but-non-rule-governed re-organization of an inherited système d'équivalences, style as mode of formulation the painter is, "perception already stylizes"; grounds universal expressive operation + indirect ontology operative content + architectonic bridge (Stiftung / système d'équivalences / painter-as-primary-witness).
Triple — pregnancy-pragnanz: rejects Gestalt-psychological Prägnanz as typicality/good-form + Piaget's field-effects-vs-sensorimotor dilemma + spiritualizing readings + Brunswik's empiricist learning-construal; substitutes pregnancy as productivity / fecundity / praegnans futuri — auto-positing form, self-radiating center, Urstiftung opening a dimension; grounds ontological generativity of Gestalt + body's "correct focusing" as response not constitution + bifid reversibility "being pregnant with flesh / flesh pregnant with being".
Axes: (i) Partial — both deepen a borrowed term by relocating it from descriptive psychology/art-criticism into phenomenological-ontological register, but CD rejects an active-imposition picture and pregnancy a passive-resultant picture. (ii) Aligned (strongest axis) — both substitute a productive non-coincidence / self-positing operation the agent is suspended on rather than performs; both invoke Urstiftung; phrasing converges ("the form that has arrived at itself, that is itself, that poses itself by its own means"). (iii) Partial / not aligned — CD grounds expressive-operative register; pregnancy grounds perceptual-ontological register. Coordinate but distinct.
Why Partial not Iso: Substitute form is strikingly aligned (axis ii), but grounding direction diverges along the expressive / perceptual-ontological split of MP's late architectonic. The 1953 Sensible World and World of Expression (today's ingest target) is the shared corpus locus where "perception already stylizes" meets pregnancy as structural form of perceptual meaning — a candidate site for a future targeted check whether they are the same operation in different registers or coordinate operations sharing a mechanism. Scanner-suggested typed connection: *shares mechanism with* (v0d.7 vocabulary).
Pair 3: letting-be ↔ passivity — PARTIAL
Triple — letting-be: rejects activity-passivity distinction itself as inherited frame (Cartesian conscious-grasping, Heideggerian Vor-stellung, Husserlian iterated reflection, Sartrean nothingness-as-hole, and per Carbone, Heidegger's Gelassenheit read as still beyond (oscillating with) the binary); substitutes a comportment located beneath the activity-passivity distinction at "aesthesiological-ontological place where original intentionality ignites" — event-initiated by Maldiney's événement-avènement; grounds comportmental form of voyance + methodological form of hyper-reflection + event-form of chiasm.
Triple — passivity: rejects frontal passivity as external causality on self-coincident subject + Sartrean decisionism + Lachièze-Rey actualism + philosophy-of-consciousness binary + Husserlian self-coincident constitution; substitutes lateral passivity ("right there in activity / obverse side"), subject as field-not-survey, non-coincidence-with-self as constitutive structural feature, sense as écart; grounds four phenomena (sleep/dreams/unconscious/memory) + institution as positive obverse + late-ontology vocabulary + ternary dialectic + cross-tradition baseline.
Axes: (i) Aligned — both reject activity-passivity binary as inherited frame. (ii) Partial — letting-be substitutes via an event located beneath the binary; passivity substitutes via a structural-topological reframing (lateral, not frontal) operating alongside the binary. Event-initiation vs constitutive-topology. (iii) Partial — letting-be grounds comportmental form, passivity grounds structural condition. Complementary, not parallel.
Why Partial not Iso: Key veto-adjacent finding from scanner: the connection is already textually attested at source level. The cardinal letting-be MP passage RC 179/198–99 is the closing of MP's last course summary on "The Problem of Passivity: Sleep, the Unconscious, Memory"; letting-be is identified by MP himself as "the primordial unconsciousness" within the Passivity-course frame. Not latent — outside weave Pass 3's latent-isomorphism scope.
Pair 4: motor-intentionality ↔ hyper-dialectic — PARTIAL
Triple — motor-intentionality: rejects Cartesian dichotomy mechanism / representation; substitutes a third kind of intentionality — body's pre-predicative grasp of practical task, "I can" replacing "I think", praktognosia replacing both reflex and representation; grounds body's space-relation, habit-as-dilation, intentional arc, and (downstream) reversibility.
Triple — hyper-dialectic: rejects "bad dialectic" — thesis/antithesis/synthesis as thetic description (Hegel, Sartre's strabismus, Cartesian dual-order) + secondary slide into scepticism/relativism; substitutes a dialectic that radicalizes itself via the "sur-" strategy — refuses thetic completion while remaining positively dialectical, holds opposition open; grounds wild Being's methodological grammar + philosophy/non-philosophy + double-circle ontology + political register (permanent revolution).
Axes: (i) Partial — both reject inherited Cartesian binary; same meta-form, different domain (behavior-explanation vs philosophical-method). (ii) Partial — both substitute a third term refusing both poles, both explicitly positive against deflation; structural cousinhood. (iii) Not aligned — motor-intentionality grounds embodied comportment; hyper-dialectic grounds philosophy's self-relation.
Why Partial not Disanalogous: the scanner found a real meta-pattern — the "sur-" strategy operating in both pairs, which is a non-trivial structural overlap. Disanalogous would imply zero or one axis aligned; the scanner found ≥2 axes partially aligned. Partial captures the cousinhood without overstating.
Pair 5: fold-pli ↔ ineinander — PARTIAL
Triple — fold-pli: rejects dualism/monism dilemma + "cut-out" model (Sartre's nothingness; Cartesian distinction of pre-existing surfaces) + Heidegger's direct ontology; substitutes a topological/ontological figure — fold as point of reversal where inside and outside "mutually apply"; multiplicity generated from within a continuum; grounds chiasm's reversibility + flesh's reflexivity + self-differentiation as Being's mode + human as locus of Being's self-reflexivity.
Triple — ineinander: rejects proximal thought (intuitive coincidence OR reflective overview) + Cartesian pensée distincte + bad ambiguity + Heideggerian Sein-zum-Tode + hierarchy-by-addition; substitutes a result-figure — being "in mélange" where two terms are already one-inside-the-other prior to fusion or survey; mutual implication as starting structure; grounds universalization of institution into ontology + cross-register reach + good ambiguity + philosophy as "knowledge of the Ineinander."
Axes: (i) Aligned. (ii) Partial — fold is the genesis figure (the "how"), Ineinander the result figure (the "what"). Same structural family, different position in same explanatory structure. The ineinander page explicitly types this: §"The Fold and the Invagination: the Genesis Figure" — "The fold names the how of the Ineinander's emergence, where the Ineinander names the what of the result." (iii) Partial — complementary halves of one ontology (synchronic-structural vs diachronic-genealogical).
Why Partial not Disanalogous (or Iso): the relation is already named on both pages — fold-pli typed *parallels* [[ineinander]]; ineinander has §"The Fold and the Invagination". Already at the typed-connection layer, not latent. A new claim would duplicate the existing apparatus.
Pair 6: tacit-cogito ↔ flesh-as-element — PARTIAL
Triple — tacit-cogito: rejects Cartesian self-grounding ego / Husserl's apodictic transcendental ego; substitutes a "tacit cogito" as pre-linguistic self-presence (1945) that is later retracted (1959); grounds (or fails to ground) the phenomenological subject's minimal self-relation; after retraction, hands the grounding work off to late ontology.
Triple — flesh-as-element: rejects substance ontology in all forms (Cartesian res, Aristotelian, Spinozan) + Sartre's In-itself/For-itself + panpsychism + Husserlian Leib as framework + flesh-monism itself; substitutes flesh as "element" (Presocratic/Bachelardian sense) — a dimensional medium "midway between spatio-temporal individual and idea", three simultaneous registers (body/world/element); grounds late-ontology dissolution of subject/object and form/matter dichotomies + painting as privileged site + sedimentation/Stiftung.
Axes: (i) Partial — both reject Cartesian foundationalism, but flesh's rejection is much broader (substance ontology as such). (ii) Not aligned — tacit-cogito's substitute is a concept being abandoned (thin self-presence that fails); flesh's substitute is a positively-constructed dimensional medium. (iii) Partial — flesh page explicitly states flesh does what the tacit cogito was supposed to do. Genealogical succession, not parallel.
Why Partial not Disanalogous: real shared rejection of Cartesian foundationalism + real genealogical-succession relation already typed in Connections ("is the 1945 form of what the late ontology will replace with flesh"). Not surface-vocabulary-only; not parallel-architecture either.
Pair 7: pente-de-l-histoire ↔ logique-de-fait — PARTIAL ✓
Triple — pente-de-l-histoire: rejects Sirius-position-vs-arbitrary-decision dilemma (Hegelian completed-history, Aronian denial of historical sense, Sartrean passé en sursis, Bossuet providence); substitutes a synchronic structural orientation — rationality history offers as orientation when read from within; grounds reprise + opposition de gauche third-attitude politics + marxist case-test + late-MP institution.
Triple — logique-de-fait: rejects dichotomy between necessary truth and arbitrary contingency (Voltairean rationalism, Hegelian closure, Sartrean absurdism, Aronian arbitrariness, Bossuet providence); substitutes a diachronic self-organizing rationality — logic in the indicative, exemplified rather than deduced, the melody-paradigm; grounds reprise + pente + homme-en-porte-à-faux + anti-pessimist reading of existentialism + historical judgment as post-hoc-but-not-arbitrary.
Axes: (i) Aligned — same double-front rejection (Hegelian closure + Aronian arbitrariness + Bossuetian providence). (ii) Partial / not aligned — Pages themselves type the relation: pente-de-l-histoire Connections: "is the synchronic register of which logique-de-fait is the diachronic register — the slope is what facts assume; logique de fait is the assuming." (iii) Aligned — both ground reprise; both ground the matrix that [[claims#pente-de-l-histoire-as-proto-institution]] (live) identifies.
Why Partial not Iso: synthetic load already housed at [[claims#pente-de-l-histoire-as-proto-institution]] (live); a new structural-parallel claim would duplicate at lower resolution. Pages already typed as synchronic/diachronic register pair.
Pair 8: chiasm ↔ empietement — PARTIAL ✓
Triple — chiasm: rejects 6 positions (mystical fusion, Cartesian dualism+homunculus, Hegelian sublation, Husserlian transcendental, standard correlation, eliminative monism); substitutes productive structural non-coincidence — "always imminent and never realized in fact"; reversibility-without-fusion with écart as internal engine; grounds Stiftung intelligibility + flesh + condition of intelligibility of science secrète + cross-tradition portability (Schelling, Husserl, Nietzsche).
Triple — empietement: rejects Cartesian pensée distincte + Sartrean engagement-as-dénégation + Heideggerian death-priority + Brunschvicg's immanence philosophique; substitutes overlap/encroachment as positive structure — "L'empiétement, qui est pour moi la philosophie, n'est pour Descartes que confusion"; grounds the genus of which chiasm, ineinander, reversibility, entrelacs are species + MP's vertical being against Cartesian horizontal + non-philosophie + métaphoricité.
Axes: (i) Aligned. (ii) Aligned in family but with extra constraint on chiasm side (écart, "imminent and never realized" non-coincidence) that empiètement does not require. (iii) Partial / not aligned — empiètement grounds into late-ontology vocabulary as genus of chiasm; chiasm grounds out into cross-tradition structural portability. Genus-to-species is not isomorphic to species-to-cross-tradition.
Why Partial not Iso: structural-parallel claim would mistype the relation (already typed as the genus of) and skip the genealogical-staging (empiètement 1953 → hinge 1954-55 → chiasm 1959-61). The empietement page already records this in ## Open Questions from the 2026-05-08 weave Pass 3.
Pair 9: sedimentation ↔ reprise — PARTIAL ✓
Triple — sedimentation: rejects museum historicity + timeless-truth platonism + Husserlian Sinnentleerung-as-defect + MP's own PdP exclusion of painting + unitary-application picture; substitutes the past as available for reactivation — living stratum of meaning operative through (not despite) forgetfulness; the "ruse" by which the new is retrospectively absorbed into the old; grounds diachronic-mechanism register of three-tier expressive architectonic + cross-author identity with Nietzsche's incorporation + strong-hermeneutic reading.
Triple — reprise: rejects pure decision (Kierkegaardian instant) + pure repetition (Hegel + Bossuet) + Sartrean passé en sursis + Aronian mosaic + Stalinist forçage; substitutes the act by which a subject takes up again — transformation by integration, not repetition; the past is taken up into a new figure; reprise is the operative form of MP's philosophy of history; grounds anti-fatalism + anti-arbitrariness simultaneously + marxist promise as critical fidelity + 1946 individu/histoire + (genealogically) 1954–55 institution.
Axes: (i) Partial — overlapping rejection of "stasis is wrong" but divergent inherited distinctions (archival vs fated-or-arbitrary). (ii) Partial — both substitute taking-up of past, but sedimentation is after-side / preservative register (what tradition has become) while reprise is act-side / agential register (subject's taking-up). Structurally complementary, not identical. (iii) Partial — both feed institution but bifurcate downstream into ontological-expressive (sedimentation) vs political-historical-agential (reprise).
Why Partial not Iso: complementarity within one architecture (institution), not parallel across two architectures. Already correctly represented in the wiki via institution as the shared parent.
Recommendation
Recommended verdict: Calibration FAILS on literal acceptance criteria (0/3 Goods returned Isomorphic; threshold is ≥2/3). Apply mode is NOT authorized for the 2026-05-16 weave run.
The substantive scanner behavior is sound — every Partial verdict has defensible reasoning identifying an existing typed-connection or claim apparatus that already articulates the pair-relation. The failure mode is a test-set design issue analogous to (but inverted from) the Run 1 lesson: my Good pairs were latent at the direct-typed-connection-between-slugs level but not at the adjacent-typed-apparatus level (genus-species, succession, synchronic/diachronic, complementary-register, downstream-relation, co-anchored-claim).
Three resume paths for the maintainer:
-
(Recommended) HALT and authorize Pass 3 dry-run only for 2026-05-16. Defer apply-mode to a future weave run after a tighter test-set is designed. Document the test-set design issue (the "second-order latency problem" in a mature corpus) in
schema-changelog.mdas a known calibration consideration, not as a SKILL change. -
Refine test-set in-session and re-run calibration. Replace the 3 Good pairs with pairs that satisfy all 6 criteria listed under "Finding 2" above. Candidate harder Good pairs would need to be hand-selected; offers below if maintainer chooses this path.
-
Adjudicate the calibration as a documented non-authorization with substantive diagnostic value. Run Pass 3 in dry-run for 2026-05-16, with the calibration report itself serving as the input adjudication for future test-set design.
Per the SKILL "Modes: dry-run vs apply" canonical rule, paths (1) and (3) preserve discipline; path (2) attempts to authorize apply mode in-session. The strict reading favors (1) or (3).
If maintainer chooses path (2), candidate harder Good pairs the scanner has not yet been calibrated against:
aletheia↔dehiscence— cross-tradition (Heidegger/MP) genealogy + no direct typed connection between slugs + no genus-species apparatus visiblenascent-state↔coexistence— both about "togetherness as prior structure" but at different levels (subject-formation vs intersubjective) — would need to verify no adjacent typed apparatusreversibility↔mutual-precession— Carbone-extended motif, may or may not have adjacent typed apparatus; check before nominating
These are candidates only; harder-Good criteria need verification before re-spawn.
Adjudication line (for maintainer to fill, per SKILL "Calibration PASS authorizes apply mode" rule):
Maintainer adjudication: AUTHORIZES_APPLY | DOES_NOT_AUTHORIZE_APPLY | REQUEST_REFINEMENT
Date: 2026-05-16
Reasoning: ...