Workshop: Theory–Methodology Derivation

Question

What is the general structure behind "a theory ships with a derived, action-shaped methodology" — and should derivation become a KB vocabulary term, possibly restructuring what we now call distillation?

Provenance

Started 2026-07-17 from a conversation that began as a terminology check: an external passage used "crystallization" / "progressive crystallization", which is not current vocabulary (source-passage.md preserves it with the mapping to current terms). Unpacking the passage surfaced several connected threads that are explored here in separate files so each can move at its own pace.

Status (2026-07-17)

The audit is done, the direction is decided (retire the term, no successor), and the first semantic wave — the obvious discovery/abstraction cluster, the original mistake — is applied and committed; see the execution status. The scope/collision reflection this workshop triggered has been promoted to the library as kb/notes/vocabulary-collisions-prevented-at-write-time-not-read-time.md. Wave 0 is complete: the two-layer structure note is promoted (kb/notes/theory-and-methodology-form-a-two-layer-execution-system.md, all ten hard-case rows classifying cleanly against it), the derived/abstracted lineage semantics are in the link grammar (kb/reference/link-vocabulary.md, with cp-skill-write updated), and the discovery lifecycle is defined (kb/notes/definitions/discovery-lifecycle.md). The draft's six pre-Wave-1 acceptance tests pass. The open front is Wave 1: the entangled semantic cases and legacy lineage-edge classification — see the execution status.

Threads

  • two-layer-theory-methodology.md — promoted to theory and methodology form a two-layer execution system; workshop draft removed, this note is now the canonical citable home the other threads point at
  • derivation-selection-vocabulary.md — closed as consumed (2026-07-17): the decomposition analysis, the bet doctrine, and the discovery-lifecycle refinement all fed the decided direction; the decision lives in migration-plan.md, the dominant-regime call in kb/reference/link-vocabulary.md, the lifecycle in kb/notes/definitions/discovery-lifecycle.md; thread deleted
  • learning-as-derivation-caching.md — closed as absorbed (2026-07-17): the amortization framing and the ACT-R/SOAR proceduralization analogue are stated in the structure note's promotion paragraph; the citation question resolved as one-line mention, no source ingest; thread deleted
  • methodology-as-inductive-bias.md — closed as absorbed (2026-07-17): the exact-spec/proxy-theory durability prediction became the candidate mixed-boundary signal in the structure note's open questions, and the corner-case-sampling question became a new open-question bullet there; neither cleared the bar for its own note; thread deleted
  • effective-theory-borrowing.md — closed as dropped (2026-07-17) per its own adoption bar: the four-point operational checklist (declared cutoff, documented fallback, matching procedure, correspondence re-check) never became a review criterion or type-spec section, so the borrowing stays cited-in-passing — the physics loans live in the structure note's "Physics vocabulary for the moving parts" section, and the Batterman singular-limits point survives as its mixed-artifact caveat; not handed to philosophy-borrowing; thread deleted, retrievable from git history if a methodology-artifact gate is ever wanted
  • polation-structure-of-generalization.md — closed as absorbed (2026-07-17): Ord's triple became the grading section of kb/notes/definitions/discovery-lifecycle.md (with the no-geometry caveat); its unabsorbed open questions (polation as a review-time gate signal, quote-fidelity mapping) are recorded in post-sweep-queue.md; thread deleted
  • distillation-control-trap.md — closed as operationalized (2026-07-17): the "what controls this artifact" diagnostic became the classifying question in kb/reference/link-vocabulary.md, and the trap's concrete case lives in kb/notes/vocabulary-collisions-prevented-at-write-time-not-read-time.md; thread deleted
  • obvious-distillation-cases.md — first staging pass over kb/notes/ and kb/reference/: obvious discovery/abstraction fixes, obvious derivation/selection rewrites, deferred META surfaces, and mixed notes to hold back
  • distillation-usage-audit.md — evidenced-by: 464 classified distill* instances across the KB; tallies, collisions, and migration mechanics
  • migration-plan.md — the decided direction (retire distillation with no successor term; theory in citable notes, boundary in link-label grammar) sequenced into waves: receiving surfaces, split entangled usages, mechanical rewording, infrastructure
  • receiving-vocabulary-draft.md — closed as consumed (2026-07-17): Wave 0 promoted all its surfaces — derivation/selection material into the structure note and link-vocabulary.md, the discovery amendment into definitions/discovery-lifecycle.md — and its hard-case table and six acceptance tests passed against them; thread deleted
  • post-sweep-queue.md — edits parked while the final Wave 1 review sweep runs: the structure note's two physics additions, the corpus-evidence restore decision, Wave 3 prep inventory, and open questions carried from closed threads
  • prose-has-no-scope.md (moved to the vocabulary-governance workshop; promoted 2026-07-17 to vocabulary collisions are prevented at write time, not resolved at read time) — the reflection thread: prose technical terms are unscoped globals and context assembly is concatenation without a linker, so sense collisions merge silently at composition; the remedy is prevention (one term, one sense, corpus-wide), not symbolic disambiguation — the mechanism behind this workshop's minimality constraint and no-successor-term end state

The threads connected through the first one, as planned: the structure note became the primary durable artifact, the three framings fed its argument and open questions, and the vocabulary thread decided the terms — by retiring the contested one. What remains open is execution bookkeeping: the migration plan, its evidence (audit, staging, worksheet), and the post-sweep queue.

What closes this workshop

  1. A structure note in kb/notes/ stating the two-layer theory–methodology pattern (generator, derived fast path, fallback, promotion-by-matching) in settled vocabulary.
  2. The vocabulary migration executed or handed off: the direction is decided (retire distillation with no successor term — "derive" stays ordinary English, the theory lives in the structure note, the DER/AMP boundary lives in link-label grammar and gates; ampliative traffic to the polation-graded discovery lifecycle, the compound being the technical term with bare "discovery" ordinary; mixed cases carry a dominance bet) — see migration-plan.md. Closure requires Wave 0 (structure note, label semantics, and discovery amendment promoted) and Wave 1 (entangled usages split) done, with the mechanical waves either done or turned into a tracked instruction.
  3. ~~The effective-theory borrowing either promoted into the philosophy-borrowing workshop's candidate list or dropped.~~ Done — dropped (2026-07-17): the adoption checklist never became operational, so the loans stay cited-in-passing in the structure note; see the thread entry above.
  4. ~~The learning and inductive-bias framings either absorbed into the structure note's argument and Relevant Notes or extracted as their own notes if they carry independent claims.~~ Done — both absorbed (2026-07-17): amortization/proceduralization was already in the structure note's promotion paragraph; the durability prediction and the sampling question landed in its open questions; see the thread entries above.

Bookkeeping

Working files are plain markdown, workshop register. Positions attributed to "current position" are where the originating conversation landed; open questions are genuinely open.

Complete file listing (generated at build time)