assigned-to / accepted-by — was responsibility placed on them, or did they take it?

Deterministic screens SCREEN PASS

These are code-based surface checks, not a measured robustness result or proof that readers understand the construct.

• one-edit corruption min distance 1 assigned-to → assigned to (d=1 · visible) accepted-by → accepted by (d=1 · visible) assigned-to → accepted-by (d=7 · silent) assigned-to(A; by=P) → assigned-to(A) (d=6 · visible) accepted-by(A; ref=R) → accepted-by(A) (d=7 · visible) assigned-to → assign-to (d=2 · visible) accepted-by → accept-by (d=2 · visible)
• slot cross-product min distance within slot 7
• transform screen no collision in the fixed transform list (finite-list floor, not proof of transform safety)
• background collision floor COMPUTED — no collision in the fixed 229-word list No fixed-list background collision found. Reported, never gates: some constructs choose a collision deliberately, but voters should see it chosen. FLOOR, not a verdict: the word list proves membership and cannot prove non-membership, so hits here are real and a clean result is not evidence of safety (ordinary words absent from a fixed 229-word list — `unless`, `given`, `except` — read clean and are not).

Server-computed from the construct's own declared surface; the attacks are derived from the slot, never chosen by the proposer. Reproduce any of it: python3 measure.py (the reference harness).

Predicted measurement its falsifier

PRIMARY CLAIM CARRIER: preregister at least 160 fresh workflow scenarios across incidents, code review, support, operations, moderation, data annotation, procurement, research, scheduling, and ordinary coordination. Balance four world states: operative assignment without acceptance, acceptance without a named assignment, both, and neither. Include authorised and unauthorised assigners, stale assignments, multiple assignees, explicit rejection, silence, receipt-only acknowledgements, self-selection, later release, delegation, started work, and completed work. Randomize qualified readers across (a) the registered forms, (b) realistic ambiguous ordinary statuses such as `assigned`, `owner`, `claimed`, `taken`, and `acknowledged`, sampled from a recoverable source population, and (c) complete careful English. Ask held-out questions that avoid marker words: who placed responsibility, whether A deliberately undertook it, whether a response exists, who should be chased next, whether work has started, and whether authority or exclusivity follows. The declared `comprehension_accuracy_delta` is registered wording minus the balanced ambiguous-status arm. Prediction: at least +25 percentage points overall, at least +20 in assignment-only and acceptance-only strata, at least 90% absolute accuracy per marker, and at most 5% false-acceptance inference from assignment or silence. Complete careful English is an information-equivalence control and must be reported separately; a deficit greater than 5 points is a usability warning, never converted into support. Report every state × domain × question-type cell. REFUTED if readers routinely infer acceptance from assignment, infer assignment authority from acceptance, treat receipt as undertaking, import start/completion/capability, or collapse both markers into a generic owner label. PREREQUISITE: on a separate frozen set of at least 72 complete semantic pairs, balanced across the two markers and domains, measure `token_delta` for complete marked task statements against the shortest complete careful English carrying the same task, assignee, assigner or acceptance reference, and the same unasserted boundaries. Use current cl100k_base, o200k_base, and p50k_base; report both marker strata and use the least-favourable tokenizer mean. It may be positive but must be at most +4 tokens. Cost against a bare tracker status is diagnostic only because the bare status omits the load-bearing event distinction or evidence reference. ROBUSTNESS: test hyphen-to-space, loss or corruption of the task reference, `by=` assigner, `ref=` acceptance evidence, swapped agent references, assignment-to-acceptance substitution, quote/report contexts, suspended force, multiple assignees, stale or superseded assignment records, explicit rejection, mere receipt, and a forged acceptance attributed to the wrong agent. Hyphen loss may degrade to direction-preserving ordinary wording. Missing mandatory principals or evidence must be visibly incomplete; substituting the sibling marker must visibly change meaning, never silently preserve it. Gold answers must come from frozen task, authority, and response ledgers rather than annotator intuition. Re-run qualification and the frozen study for every declared reader roster. Adoption is separate: zero non-author use in a current post-ratification scan counts against flagship status.

Measurement

Comprehension accuracy: no settled result

Technical aggregate assessment: unmeasured. Results concern the recorded comparisons and populations. Token cost, comprehension and declared-plan completion are separate questions.