research-document

Theory Traceability Matrix v0.1

Theory Traceability Matrix v0.1

Purpose: Make every major theory statement traceable to supporting evidence, competing explanations, falsification tests, and current epistemic status.

Theory Statement Supporting Evidence Confidence Competing Explanations Falsification Test Current Status
Identity-Capability separation improves characterization. FE-008 pilot work on artifact characterization and identity-capability separation. Moderate Reviewer consistency may improve because the instrument is more structured, not because identity and capability are objectively distinct. Independent reviewers using alternative classification instruments do not show improved agreement or traceability from identity-capability separation. Provisionally Supported
Structured redesign improves traceability. FE-011A internal blind pilot summaries across SWOT, Five Whys, and OODA. Low Generic scaffolding or longer structured prompts may explain the improvement rather than Framework Engineering specifically. Controlled redesign trials against non-FE structured baselines fail to improve traceability or evidence handling. Under Investigation
Procedural reasoning may be represented by a finite primitive vocabulary. FE-012A internal extraction and reconstruction run with partial vocabulary stabilization and recognizable reconstruction in most cases. Low Same-model convergence, prompt structure, or analyst contamination may explain the apparent stabilization. Independent extractors fail to converge and the primitive vocabulary keeps expanding across diverse artifacts. Provisionally Supported
Primitive vocabulary can synthesize coherent procedural structures. FE-012B internal synthesis run with coherent syntheses in 7 of 10 cases without missing primitive requests. Low Prompt bias, reviewer bias, or generic procedural writing habits may explain the coherence. Independent reviewers judge synthesized structures incoherent or highly incomplete across a larger novelty set. Provisionally Supported
Reasoning grammar and execution grammar may be distinct. FE-012B missing primitive pressure around synchronization, scheduling, and sequencing, combined with FE-012A stronger compression of reasoning-heavy artifacts. Low A single grammar with insufficient temporal primitives may explain the same results without requiring a separate coordination grammar. Coordination-heavy artifact extraction and synthesis succeed cleanly after only minor refinement of the existing reasoning grammar, without requiring a distinct grammar family. Under Investigation
Framework Engineering currently functions as a characterization and redesign methodology. FE-008 identity-capability work; FE-011A redesign pilot. Low General systems-analysis or structured evaluation techniques may explain the same observed gains without a distinct Framework Engineering methodology. Independent trials show no reliable advantage in characterization traceability or redesign quality beyond generic analytical scaffolding. Provisionally Supported

Notes

  • Confidence remains capped by internal-run limitations.
  • This matrix is explanatory and audit-oriented.
  • It should be revised when new independent evidence arrives.