FE-012B primitive synthesis experiment: research frontier

Scope and source artifacts

• research/experiments/FE-012B-primitive-synthesis/hypothesis.md
• research/experiments/FE-012B-primitive-synthesis/protocol.md
• research/experiments/FE-012B-primitive-synthesis/methodological-review.md

These files are treated as one analysis unit because they share an evidence base or form one accepted/current package. Generated derivatives, templates, prompts, private working packets, and superseded runs are not counted as independent conclusions.

Knowledge extraction

• Primary objective: Test whether a frozen primitive vocabulary can synthesize coherent structures for novel problems.
• Primary claims: A small backbone appears useful. Coordination-heavy problems expose missing grammar. Generic coherence is not domain adequacy.
• Methodology: Novel problem synthesis followed by methodological review.
• Accepted hypotheses/findings: Bound, Observe, Evaluate, Decide, and Act appear indispensable within this run.
• Rejected or unsupported hypotheses: The current vocabulary is sufficient across execution-heavy domains.
• Assumptions: Novel problems do not trigger memorized framework reconstruction. Review criteria distinguish generic from domain-specific coherence.
• Limitations: Independent reviewers and larger novelty samples are absent.
• Known uncertainties: Coordination primitives; Memorization effects; Independent composition agreement.
• Confidence: bounded by the source package; no confidence is promoted by this frontier analysis.

Adversarial challenge

The finite-vocabulary hypothesis coexists with repeated missing-primitive pressure in coordination tasks. Evidence that would materially change the assessment includes independent replication, preregistered negative controls, matched causal comparison, or a demonstrated failure of the current measurement instruments. The analysis deliberately treats planned validation as a dependency, not as completed evidence.

Top five opportunities

Semantic duplicates were merged into the linked repository-wide records. The five selections maximize uncertainty reduction and repository reuse, rather than creating five unique labels for each source package.