Two gaps I’d propose

AI writtenClaude wrote these gap statements, working from Convergent’s own house format. Neither has been reviewed by a person.

Written to the Gap Map’s format: a title-case declarative name, 30 to 60 words, no urgency language, no named vendors.

Each one was grepped against all 103 of the map’s gaps and all 369 of its capabilities for near-duplicates, and checked against current programs to see whether it is already funded and under construction. Two of those funding checks came back “not clear”, and say so. Open a card for all five checks: the nearest thing already in the map, whether the gap pulls against something else worth having, what it would unlock downstream, the near-duplicate search in full, and the funding check.

Both are proposals, not part of the Gap Map. They are kept separately from Convergent’s data everywhere, and the proposed mark on a card is how to tell at a glance.

proposedCoating Thermal Noise Sets the Sensitivity Floor for Precision Optical Instruments

Materials Science

The mechanical loss of dielectric mirror coatings sets a Brownian noise floor that limits gravitational wave detectors, optical clocks and cavity-stabilized lasers alike. We fund coating work as a subsystem of each instrument, so the improvement that would serve all three is nobody's program. Lower-loss coatings deposited at meter scale are needed.

Primary blockerDesign searchAI acceleration2-5 yearsMeasurabilityDirectly measurable

proposedThere Is No Machine-Readable Record of What Did Not Work

Metascience

Negative results, null findings and abandoned approaches are largely missing from the scientific record, and where they are captured the formats are neither standardized nor machine-readable. Systems that could learn from failure have little to learn from. The obstacle is credit, not storage.

Primary blockerCoordination and institutionsAI acceleration2-5 yearsMeasurabilityProxy only