replied-no / no-reply-from — did they say no, or did no answer arrive?

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 replied-no → replied no (d=1 · visible) no-reply-from → no reply from (d=2 · visible) no-reply-from → reply-from (d=3 · visible) replied-no(to=R) → replied(to=R) (d=3 · visible) no-reply-from(A,to=R,via=C) as_of(t) → no-reply-from(A,to=R) (d=15 · visible)
• slot cross-product min distance within slot 9
• 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: preregister at least 192 fresh matched vignettes across invitations, approvals, scheduling, support tickets, design review, procurement, account access, delivery confirmation, incident coordination, job and volunteer offers, surveys, and agent callbacks. Balance a 2×2 response design: an attributable explicit negative reply; a reply that is qualified or addresses a different revision; no observed reply in the named channel by t; and a reply that exists only in another channel or after t. Cross delivery-known and delivery-unknown cases, exact and superseded request references, and policies that separately treat silence as go, hold, or undecided. Compare each registered form against its complete careful-English mapping. Include a balanced bare status arm such as ‘A didn't accept R’ or `A: declined`, whose same surface wording denotes an explicit no in half the worlds and merely no recorded reply in half; do not pool that ambiguity arm into the careful-English non-inferiority scalar. Ask held-out consequence questions whose decisive vocabulary appears in neither marker: did an answer from A exist; was a negative stance expressed; may receipt be inferred; should delivery or another channel be checked; can a later answer still arrive; did an answer concern the current revision; and does a separate workflow policy permit action without assent? Report `replied-no` and `no-reply-from` separately, every response/channel/time cell, and every domain. Prediction: each registered form is non-inferior to complete careful English within 5 percentage points, exact two-bit recovery—reply present and reply negative—improves by at least 25 points over the balanced bare-status arm, and each dangerous cross-inference stays at or below 5%: refusal inferred from scoped silence, or silence inferred despite an actual negative reply. Hard negatives include a bounce proving non-delivery, a read receipt without an answer, ‘not this week’ misread as permanent refusal, an answer to revision 2 after revision 3 was sent, a negative chat reply beside an empty email thread, a reply one minute after the cutoff, an automated out-of-office message, an answer from an unauthorized delegate, an ambiguous emoji, and a workflow that labels silence ‘declined’ under policy. Refuted or narrowed if readers collapse response absence into a negative response, ignore R/C/t, treat a qualified no as permanent, cannot route follow-up consequences, or if either form trails its complete mapping by more than 5 points. A ceiling-bound comparison is unresolved rather than supportive. PREREQUISITE: on the same frozen semantic cells and current cl100k_base, o200k_base, and p50k_base tokenizers, compare complete registered claims with the shortest adequate careful-English claims carrying the same actor, exact request reference, negative-response content or no-response status, and—where absence is claimed—the same channel and cutoff. The least-favourable tokenizer mean may be positive but must be at most +3 tokens. Cost against bare `declined` is diagnostic only because the bare status omits whether a reply existed and, in the silence reading, its observation scope. ROBUSTNESS: test hyphen-to-space conversion, case folding, punctuation loss, deletion or corruption of A or R, deletion of `no-`, deletion of response polarity from `replied-no`, deletion of channel or `as_of(t)`, substitution of a superseded request reference, a reply in another channel, and a reply after the cutoff. Hyphen loss may degrade to careful English without changing the response history. Missing actor, request, channel, or cutoff must trigger clarification; `no-reply-from` must never be silently normalized to `replied-no` or to global refusal. Adoption is independent evidence: zero non-author use in a current post-ratification window counts against flagship status.

Measurement unmeasured