Competency Taxonomy & Prerequisite Graph

The real problem

A lesson list and its display order are not a competency model. If “lesson A appears before lesson B” silently becomes a prerequisite, an editorial reorder can incorrectly block or unlock learning routes. This lesson builds Competency Taxonomy & Prerequisite Graph Contract v1 so competency identity, directed relationships, edge types and closure evidence are explicit and auditable.

The artifact must explain why a route is open or blocked and exactly what evidence can close an edge without using page order as authority.

Lesson boundary

MKN-003 owns the competency taxonomy and prerequisite graph. MKN-002 owns project/JIT learning design and MKN-004 owns longitudinal mastery state. This lesson does not absorb either adjacent outcome.

Three boundaries: content order ≠ competency dependency, edge ≠ mastery, and equivalence ≠ identity merge.

Mental model: node + typed edge + closure

Each competency has a stable identity inside a framework. Each dependency is a separate edge with an edge_id, direction prerequisite → child, edge_type, strength, evidence condition and version. Closure is derived from evidence; it is not a permanent property of the edge.

When evidence is missing or disputed, typed hold/review states remain visible instead of being coerced into satisfied.

The 16-dimension contract

The artifact binds framework, competency, statement, taxonomy level, edge, type, prerequisite, child, strength, evidence condition, closure, equivalent evidence, bridge, cycle, version and provenance.

competency_id and edge_id are stable identities. Statements may evolve, but identity must not be regenerated just because wording changes. version makes semantic changes traceable.

Why order and text similarity are not authority

Content order is useful for UX and pedagogy, but a dependency requires an explicit edge. Reordering lessons without changing the graph must not change prerequisites.

Similar statements across frameworks are not automatically the same competency. During import, distinct identities remain distinct and alignment/equivalence is recorded as a governed reference. Text similarity alone must leave closure on hold.

CASE supplies standardized competency-framework, item and association exchange mechanisms. MAKNOW separately owns the local hard/recommended/bridge and closure semantics used here.

Worked baseline

Assume CMP-FUNC is a hard prerequisite for CMP-TEST. Edge EDGE-101 points CMP-FUNC → CMP-TEST, and its closure condition requires independent evidence for the prerequisite competency. Valid evidence exists, so the route can reach ROUTE_AVAILABLE_HARD_CLOSED.

Now reorder the lessons in the UI. The edge did not change, so the dependency does not change. In transfer, external competency EXT-FUNC-A has similar wording but a different ID; a reviewed association may satisfy a recommended edge through equivalent evidence without merging the identities.

Common failure modes

1. Display-order inference: page order becomes the graph.
2. Direction blindness: A→B and B→A are treated as equivalent.
3. Hard-edge inflation: weak activity/completion closes a hard prerequisite.
4. Text-similarity merge: two framework items are merged because wording looks similar.
5. Cycle blindness: a three-node cycle is treated as routable.
6. Version amnesia: edge semantics change while version/provenance stay unchanged.

Hard, recommended, equivalent and bridge

A HARD_PREREQUISITE blocks until valid evidence closes it. A RECOMMENDED edge is advisory: the route may remain available while the gap stays visible. Equivalent evidence is usable only when the relationship and provenance are governed. A BRIDGE is a repair/preparation route, not fabricated closure.

These meanings are local policy. CASE provides a representation/interoperability mechanism rather than MAKNOW closure policy.

Production implementation bridge

Version nodes and edges, run self-loop/cycle checks on each graph release, and materialize learner closure separately from canonical graph definitions. A direction/type mutation should create a new edge fingerprint and mark prior closure for recomputation.

For imported frameworks, keep mapping records separate from canonical identities so a mapping can be revised or revoked without rewriting graph history.

Guided and independent practice

Build the baseline hard edge and populate all 16 dimensions. Reverse only display order: state must not change. Then reverse the actual edge direction: the old closure is no longer an oracle and recomputation is required.

For independent practice, introduce a three-node cycle, then change a hard edge to recommended without an equivalence policy. Each mutation must produce an explainable failure or review state.

Verification evidence

Ask five questions: Are identities stable and distinct? Is edge direction/type explicit? Does hard closure trace to valid evidence? Does cycle detection operate on the graph rather than display order? Can equivalence/bridge provenance be reconstructed?

The offline lab exposes all input records and rules and never touches a real tenant or learner record.

Assessment and transfer

Assessment covers boundary retrieval, baseline construction, reversed-edge and cycle diagnosis, and transfer to a second framework. Transfer cannot copy baseline closure: identities stay distinct, the new association must be traceable, and the recommended edge is recomputed from changed evidence.

If only wording is similar or equivalence lacks governed provenance, the result is HOLD_UNVERIFIED_EQUIVALENCE.

References and freshness

FACT: 1EdTech CASE 1.1 defines standardized exchange of standards, competencies, skills and learning outcomes and models digitally identifiable items plus associations. INFERENCE: Versioning relationship identity, direction, type and evidence condition makes a prerequisite graph more auditable than inferring dependencies from course order. DESIGN: Hard/recommended semantics, cycle policy, closure states and bridge/equivalence policy are MAKNOW local design authority.

Sources: 1EdTech CASE 1.1 specification portal and Introduction to CASE; freshness checked 2026-09-14, review due 2027-01-15 or earlier on material change.