Cross-model-authoring spec teaches wrong constructor-term keyword #28
Reference in New Issue
Block a user
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
The experiment-spec at experiments/2026-05-12-cross-model-authoring/master/spec.md (and the rendered ail.md) tells the LLM:
But the surface parser rejects
(ctor ...)in term position with [surface-parse-error] "unknown term head `ctor`". The canonical Form-A keyword for a constructor value is(term-ctor TYPE CTOR ARG*)—(ctor ...)is only valid inside(data ...)definitions. The canonical design/ docs (design/models/authoring-surface.md, design/models/rc-uniqueness.md, design/contracts/scope-boundaries.md) useterm-ctorconsistently; only the experiment spec drifted.Observed during a Qwen3-Coder A/B naming run (experiments/2026-05-21-naming-ab/runs/r1/) where every t4_count_zeros output produced
(ctor List Nil)etc. in term position, identically in all three cohorts — a constant 100% check-fail tax tied to spec drift, not naming. Fixing this should be a one-line patch to master/spec.md plus a re-render.