All claims
  1. Multiple Natures
  2. Claims
  3. 09-tagging-and-clusters
  4. CLM-L033
09-tagging-and-clusters

Chromatrait clusters — the upstream authored layer activating careers via signature dimensions

  • CLM-L033
  • 🔒 Locked (legacy)
  • 🔍 Practitioner-grounded
  • Falsifiable ✓
  • 🔒 Practitioner

CLM-L033 — Chromatrait clusters (upstream authored lineage)

Status: 🔒 Locked (legacy) · 🔍 Practitioner-grounded · Falsifiable ✓ — operational in diagnostics/engine/careers/scripts/; not yet integrated into THEORY-OF-TRAITS.md

Topic: 09-tagging-and-clusters

CLAIM TEXT

The framework names a small, manually-authored upstream layer of cross-cutting trait clusters — the chromatrait archetypes — that sit above the 9×10 dimensional matrix and activate occupational entities through weighted signature dimensions. Sixteen chromatrait archetypes were authored by the MN team (preserved in the legacy xavigate_db.sql dump under insights.chromatrait_archetypes), translated into three languages (en/fr/ru), and migrated into the current schema as clusters with provenance code chromatrait-xavilms and tier gold.

The structure of a cluster:

  • Identity: name, essence, energy pattern, work style, thriving contexts, friction contexts, famous examples.
  • Signature dimensions: a weighted JSON object linking the cluster to the 19-dimensional MI/MN space. Three weight tiers from the chromatrait source:
  • signature_big_tigers → weight 1.0
  • signature_medium_tigers → weight 0.7
  • signature_small_tigers → weight 0.4

The activation relation between a cluster and a career:

activation_strength(cluster, career) =
    Σ (career_score[dim] × cluster_weight[dim])
    /
    (10 × Σ cluster_weights)

Pairs above a threshold (default 0.6) are emitted as junction rows in activated_clusters, linking a cluster to every career whose 19-dimensional profile resonates with it.

The framework's structural claims about clusters:

  • Clusters are derived, not tagged. No human assigns a cluster to a career. The activation relation falls out of the cluster's signature weights and the career's scores. This is the methodological mirror of CLM-L031: the rubric is reverse-engineered; the cluster membership is computed.
  • Clusters cut across the 9×10 matrix. A chromatrait archetype is not a single Nature or a single Intelligence — it is a recurring shape across multiple dimensions. This is what permits Quiet Depletion, Manufactured Energy, Structural Absence, Mode Collapse, Personalization Error to be recognized as cluster-level patterns rather than per-dimension diagnoses.
  • Clusters are upstream of marketing pillars. The pillars surfaced in marketing canon (MARKETING- pillar files) are the public-facing surfacing of cluster-level patterns; the theoretical justification for why the clusters cluster* lives at this layer.
  • Clusters are tier-graded by provenance. The 16 chromatrait archetypes are tier gold (manually authored by the MN team). Future clusters discovered via unsupervised methods on the 649-career corpus would enter at lower tiers and require curation to reach gold.

The framework's load-bearing methodological commitment:

> Cluster membership must be derivable from the dimensional layer below it; otherwise the cluster is a typology, not a pattern.

This is the discipline against the most common failure mode in trait-cluster work: clusters that look meaningful but cannot be operationally cashed out as functions of the underlying dimensions.

LOCATION (pre-adoption)

  • diagnostics/engine/careers/scripts/transform-chromatrait.py (chromatrait → clusters schema migration)
  • diagnostics/engine/careers/scripts/compute-chromatrait-activations.py (activation-strength computation)
  • diagnostics/engine/careers/scripts/seed-output/seed-clusters-chromatrait.sql (16 archetypes, 48 translation rows)
  • diagnostics/engine/careers/scripts/seed-output/seed-activations-chromatrait.sql (cluster-career junction rows)
  • theory/clusters/README.md (placeholder — empty pending theoretical articulation)
  • content-system/kernels/structural-kernels.jsonl and pain-kernels.jsonl (operational sibling layer; cluster-level patterns surfaced as kernels)

LOCATION (post-adoption, when integrated)

Not yet integrated into THEORY-OF-TRAITS.md. The theory/clusters/README.md notes this folder is empty pending theoretical articulation; this claim is the first formal articulation of the cluster layer for canon. Recommended cherry-pick: a Clusters & Lineage sub-section paired with CLM-L031 (frequency rubric) and CLM-L032 (tagger architecture), naming the chromatrait gold layer, the signature-dimension weights, the activation-strength formula, and the derived-not-tagged commitment.

EVIDENCE TYPES

[P] Phenomenological

Strong practitioner observation. The 16 chromatrait archetypes (authored over years of MN practitioner work) reproduce as recognizable career-cluster shapes in the 649-career corpus when activation strength is computed; the activated cluster lists for any given career match practitioner judgment about which archetypes fit. The pattern that "Healing-Educative-Intrapersonal" forms a recognizable cluster (rather than three independent dimensions) replicates across observers.

[E] Empirical

  • 16 manually authored archetypes × 3 languages = 48 rows preserved from the legacy database with full provenance.
  • Activation strengths computed for every cluster × career pair (16 × 649 = 10,384 pairs); the threshold-passing junction rows are auditable.
  • MISSING — empirical validation that activation-derived cluster membership matches independent practitioner judgment (a held-out cluster-tagging task).
  • MISSING — analysis of which clusters are robust across thresholds (0.5, 0.6, 0.7) and which collapse.
  • MISSING — unsupervised cluster-discovery analysis on the 649-career corpus for comparison with the 16 authored clusters.

[T] Theoretical

  • Compatible with CLM-L025 (combinatorial profile space): clusters are patterns within the combinatorial space, not types reducing it.
  • Compatible with CLM-L031 (frequency rubric) and CLM-L032 (tagger architecture): cluster prior in the tagger reads from this layer.
  • Compatible with the framework's structural-attribution canon (CLM-L020 personalization error): cluster diagnoses describe recurring structural shapes rather than individual deficits.
  • Convergent with archetype theories (Jung), career-typology research (Holland's RIASEC), and modern trait-cluster work (Big Five facet structures, NEO-PI-R clusters).

[C] Convergent

  • Holland (1959, 1985) — RIASEC career typology; structural parallel for occupational clusters cutting across personality dimensions, with the framework departing from Holland's six discrete types toward weighted multi-dimensional signatures.
  • Jung (1921) — psychological types and archetypes; convergent on the recurring-pattern claim, with the framework's clusters being empirically derivable rather than archetypally given.
  • Big Five facet research — Costa & McCrae's NEO-PI-R; clusters of facets that travel together across populations.
  • Career-cluster taxonomies — U.S. Department of Education 16 Career Clusters, O*NET Job Family taxonomy; structural parallel at the occupational level.
  • MISSING — convergent rs- entries on Holland, Jung archetypes, NEO-PI-R facet structure, and career-cluster taxonomy literatures.

UPSTREAM SOURCES

  • Steven Rudolph and the MN practitioner team (2018–2026). Chromatrait archetypes (insights.chromatrait_archetypes, legacy xavigate_db.sql.gz dump 2026-02-17).
  • diagnostics/engine/careers/scripts/transform-chromatrait.py (transformation governance).
  • diagnostics/engine/careers/governance/TAGGING-PHILOSOPHY.md v1.2 §Trait Clusters and §Lineage.

POSITIONING IN LITERATURE

  • Confirms: Holland on occupational clusters cutting across dimensions, Jung on recurring archetypal patterns, Big Five facet research, career-cluster taxonomies.
  • Extends: treats clusters as derived (computable from signature weights × dimensional scores) rather than given (assigned by typology). The framework's contribution: a tier-graded, provenance-tracked cluster layer with a published activation formula — clusters that can be challenged by recomputing rather than re-authoring.
  • Departs: from typology traditions (Jungian types, Holland's six categories) by refusing discrete-category cluster membership. A career belongs to multiple clusters with continuous activation strengths; clusters are lenses, not bins.

FALSIFIABILITY

The chromatrait-cluster claim would be falsified if:

  • Activation-derived cluster membership fails to match independent practitioner cluster-tagging at meaningful agreement (κ < .5).
  • The 16 authored clusters dissolve under unsupervised clustering on the 649-career corpus (the corpus shows fundamentally different natural clusters).
  • Cluster signatures fail to generalize: the same cluster requires different signature weights to fit different occupational corpora.
  • Cluster-level patterns add no diagnostic value beyond the per-dimension scores (i.e., practitioners shown cluster activations make equivalent judgments to those shown only the 19-dimensional profile).

EDGE CASES / KNOWN LIMITS

  • Authored gold layer is small. Sixteen archetypes are not exhaustive. The framework treats this layer as opening canon, not closing canon — additional clusters can be added at lower tiers and curated upward.
  • Threshold sensitivity. Activation threshold of 0.6 is a tunable parameter; lower thresholds explode junction-row counts (most careers activate most clusters weakly), higher thresholds collapse to only the tightest fits. The framework treats threshold as a presentation choice, not a truth claim.
  • Cluster-vs-pillar ambiguity. Marketing pillars (Quiet Depletion, Manufactured Energy, Structural Absence, Mode Collapse, Personalization Error) are cluster-level patterns at varying granularity. The framework has not yet resolved which pillars are theoretical-cluster-grade vs. surfacing artifacts. The theory/clusters/ folder is the intended resolution site.
  • Translation discipline. Chromatrait archetypes are tri-lingual (en/fr/ru) by source; cluster names and essences must travel with the canonical translation table rather than being re-translated each surfacing.

DISCONFIRMING CASES TRACKED

  • Chromatrait archetypes whose signature-derived activations consistently mismatch practitioner judgment are flagged for signature revision in governance/. Persistent mismatch on an archetype indicates the signature is wrong, not that the activation formula is wrong.

REFLEXIVITY NOTE

The 16 chromatrait archetypes reflect the originator's pedagogical and clinical observation across two decades of MN practitioner work; they encode pattern-recognition that pre-dates the formal signature-weight schema. The shift from "authored archetype with descriptive essence" to "authored archetype with computable signature dimensions" is the move that makes the cluster layer falsifiable. A practitioner from a Jungian-archetype tradition may experience the framework's insistence on derivable activation as flattening; the framework's view is that non-derivable archetypes are unfalsifiable and therefore unaccountable.

RELATIONSHIP TO CURRENT CANON

  • Already integrated? No. THEORY-OF-TRAITS.md does not yet describe the cluster lineage; theory/clusters/README.md flags this as pending articulation.
  • Contradicts current canon? No.
  • Net-new? The chromatrait-as-gold-tier framing, the signature-dimension weight schema, the activation-strength formula, the derived-not-tagged commitment, and the cluster/pillar relationship are net-new to master canon.
  • Recommended action: Cherry-pick a Clusters & Lineage sub-section into THEORY-OF-TRAITS.md naming the gold-tier authored layer, the activation formula, and the derivability commitment. Pair with CLM-L031 and CLM-L032 to complete the Tagging & Clusters topic.

RESEARCH-BANK GAPS FLAGGED

For BACKLOG.md:

  1. Holland (1959, 1985) — RIASEC vocational typology and successor literature.
  2. Jung (1921) — Psychological Types; archetypal theory.
  3. Costa & McCrae — NEO-PI-R facet structure.
  4. U.S. Department of Education 16 Career Clusters — public-sector career taxonomy.
  5. **O*NET Job Family taxonomy** — operational occupational clustering.
  6. Unsupervised cluster discovery methods — for comparison with the authored gold layer.

NOTES

  • This claim closes the original Phase 2B plan: the framework's three-axis structure (Nature × Situation × Orientation) is now atomized at the claim level and the methodological infrastructure (Tagging & Clusters) is articulated.
  • Pairs with CLM-L031 (rubric — what predicts each dimension) and CLM-L032 (tagger architecture — how the layers fuse).
  • Worth elevating into theory/clusters/ as the seed file when the empty folder gets populated.
Citations · 0 research entries

No research entries linked yet. Gaps tracked in research/method/BACKLOG.md.

Related claims
← All claims