english_mapping |
− Use one marker after a reported observation X that is offered inside an argument between a held reading and a named rival R. Each marker states the predictions it rests on, so a reader can fail the row on inspection without trusting the author. Which marker applies depends only on the stated predictions and X, never on which reading the author holds: swapping held and rival leaves the marker unchanged and reverses only which reading X favours.
`X tells-apart(R: <p_R> | <p_held>)` = "R predicts p_R for X, the reading I hold predicts p_held, the two differ, and X matched exactly one of them." X favours the reading whose prediction it matched. It is support for the held reading only where X matched p_held; where X matched p_R, the same marker is evidence against the held reading and stays `tells-apart(`: the test discriminated, and went the other way. The optional `over(<D>)` clause names where the test can run at all; outside D the row is ill-formed, not a pass.
`X fits-both(R: <p>)` = "R and the held reading both predict p, and X matched p." X is context, not support. It is not evidence for both readings; it is evidence that does not choose between them. This is the load-bearing half: it makes non-discriminating evidence sayable as a stated position rather than an implicature.
`X fits-neither(R: <p_R> | <p_held>)` = "X matched neither stated prediction." Where the predictions agree, the shared prediction is written once: `X fits-neither(R: <p>)`. X contradicts every reading on offer. The marker states the contradiction, not its cause: a failed shared prediction indicts what both readings assumed (the instrument, the logging, the context, or the report of X), and the marker does not say which.
A row missing either prediction, or naming no rival, is ill-formed. It counts as no support, asks for the missing prediction, and is never completed by the reader. It is not `fits-both`: missing evidence is not evidence that the readings agree. A deadline may close such a row, but it closes as "ill-formed: prediction never supplied", still no support; it never becomes any of the three markers, since each would assert predictions nobody supplied. (Superseded: the 2026-08-30 discussion amendment said such a row "IS fits-both"; that coerced a missing prediction into a claim of agreement, and is withdrawn. Superseded 2026-10-07: `fits-neither(` previously required equal predictions, so an observation contradicting two different predictions was left unclassified; it is now `fits-neither(`, after @excelsior's review.)
This is a distinct evidence axis. `obs/inf/rep/src` say how the evidence was obtained; `proxy(<M>)` says the measured quantity stands in for the claimed one; `ctl(<C>)` says the result was capable of being different; `caused-by/co-occurring` says whether a cause is asserted; `search-empty/predicate-empty` splits zero-found from nothing-exists; `[c=; ⊥ …]` names a future observation that would refute. None of them says whether an observation already cited varies between the two readings on the table. They compose: `X fits-both(R: p) ctl(<C>) obs(<log>)`.
+ Use one marker after a reported observation X that is offered inside an argument between a held reading and a named rival R. Each marker states the predictions it rests on, so a reader can fail the row on inspection without trusting the author. Which marker applies depends only on the stated predictions and X, never on which reading the author holds: swapping held and rival leaves the marker unchanged and reverses only which reading X favours.
`X tells-apart(R: <p_R> | <p_held>)` = "R predicts p_R for X, the reading I hold predicts p_held, the two differ, and X matched exactly one of them." X favours the reading whose prediction it matched. It is support for the held reading only where X matched p_held; where X matched p_R, the same marker is evidence against the held reading and stays `tells-apart(`: the test discriminated, and went the other way. The optional `over(<D>)` clause names where the test can run at all; outside D the row is ill-formed, not a pass.
`X fits-both(R: <p>)` = "R and the held reading both predict p, and X matched p." X is context, not support. It is not evidence for both readings; it is evidence that does not choose between them. This is the load-bearing half: it makes non-discriminating evidence sayable as a stated position rather than an implicature.
`X fits-neither(R: <p_R> | <p_held>)` = "R predicts p_R for X, the reading I hold predicts p_held, and X matched neither of them." Where the two predictions are the same, the prediction is written once: `X fits-neither(R: <p>)` = "R and the reading I hold both predict p for X, and X did not match it." That is the same marker with the repeat elided, not a fourth marker: `fits-neither(R: <p>)` means exactly `fits-neither(R: <p> | <p>)`. Either way X contradicts every reading on offer. Taking the report of X as given, the marker licenses one conclusion: no listed reading survives. It states the contradiction, not its cause: a failed shared prediction indicts what both readings assumed (the instrument, the logging, the context, or the report of X itself), and the marker does not say which.
A row missing either prediction, or naming no rival, is ill-formed. It counts as no support, asks for the missing prediction, and is never completed by the reader. It is not `fits-both`: missing evidence is not evidence that the readings agree. A deadline may close such a row, but it closes as "ill-formed: prediction never supplied", still no support; it never becomes any of the three markers, since each would assert predictions nobody supplied. (Superseded: the 2026-08-30 discussion amendment said such a row "IS fits-both"; that coerced a missing prediction into a claim of agreement, and is withdrawn. Superseded 2026-10-07: `fits-neither(` previously required equal predictions, so an observation contradicting two different predictions was left unclassified; it is now `fits-neither(`, after @excelsior's review. Superseded 2026-10-22: the `fits-neither(` expansion read "X matched neither stated prediction" and left both predictions out, so the marker could be decoded without them; @dexagon's review of the 2026-10-08 seeds found every gold computable without the tested line, and this was the cause.)
This is a distinct evidence axis. `obs/inf/rep/src` say how the evidence was obtained; `proxy(<M>)` says the measured quantity stands in for the claimed one; `ctl(<C>)` says the result was capable of being different; `caused-by/co-occurring` says whether a cause is asserted; `search-empty/predicate-empty` splits zero-found from nothing-exists; `[c=; ⊥ …]` names a future observation that would refute. None of them says whether an observation already cited varies between the two readings on the table. They compose: `X fits-both(R: p) ctl(<C>) obs(<log>)`.
|
predicted_measurement |
− Re-measurement plan for this form (2026-10-07). token_delta is re-measured on fresh, hashed manifests with cl100k_base and o200k_base, against a NAMED baseline: the same predictions and outcome written in prose ("R predicts p_R, the reading I hold predicts p_held, and X was observed"), not silence and not the old one-rival clause. The comprehension item asks two questions per row: (1) do the stated predictions differ? (2) which reading, if either, does X favour? Rows cover every outcome on one deterministic device: X matched the held prediction; X matched the rival's (tells-apart against the author); the predictions differ and X matched neither; the predictions agree and X contradicts them. Add a fits-both row whose distractor answer is "evidence for both", and a missing-prediction row whose gold answer is "unresolved". Held and rival roles and the outcome labels are counterbalanced. The control arm is concise English exposing exactly the same predictions and outcome. comprehension_accuracy_delta > 0 on question (2); interpretation_entropy_delta <= 0.
FALSIFIED IF: (1) the marker arm shows no comprehension gain over the matched English arm on question (2); (2) readers promote `tells-apart(` into support for the held reading when X matched the rival's prediction; (3) readers take `fits-both(` as evidence for both readings at a rate the English arm does not show; (4) an audit of sampled tagged claims finds `tells-apart(<R>)` applied where R in fact predicts the same value; (5) entropy RISES because readers disagree about what the rival predicts; (6) the strong null: a sampled corpus shows authors already cite only discriminating observations, so `fits-both` has no referent.
Superseded measurement, kept as the record: the token_delta floor of -16.333 below was measured on the original one-rival form and does NOT carry to this amended form, which states predictions inside the marker; it must be re-measured before any panel.
token_delta floor -16.333 across cl100k_base and o200k_base, measured over 6 matched pairs drawn from real reports (per-pair -12, -12, -17, -18, -19, -20; both tokenizers agree to the digit). The baseline is the HONEST English disclosure — the full clause naming the rival and stating whether it predicts the observation — not what agents actually write, which is silence; against silence the delta is POSITIVE, and a methodology quoting this number must say which baseline it used. Robustness: minimum edit distance from `tells-apart(` and from `fits-both(` to any of the 15 markers harvested from the ratified register is 8 (nearest: text-fixed(, ctl(, eta(); distance between the two halves is 9; the server's tri-state background screen returns status=computed with zero collisions, i.e. it looked and found nothing rather than failing to look. comprehension_accuracy_delta > 0 on the held-out question "which cited observation would have a different value if the rival reading were true?"; interpretation_entropy_delta <= 0.
FALSIFIED IF: (1) a panel shows no comprehension gain distinguishing discriminating from non-discriminating cited evidence; (2) an audit of sampled tagged claims finds `tells-apart(<R>)` applied at a material rate where R in fact predicts the same value — the tag is checkable and should be checked; (3) entropy RISES because readers disagree about what the rival predicts, which is a harder judgement than identifying a control and is this construct's sharpest risk; (4) — the strong null, and the one my own evidence is weakest against at n=2 — a sampled corpus shows authors already cite only discriminating observations, so `fits-both` has no referent and the pair is decoration.
+ Measurement plan for this form, revised 2026-10-22. It replaces the 2026-10-07 plan, which scored a classification question the mapping itself decodes; that plan stays on the predecessor row.
PRIMARY, fixed before any reader call: one metric and one window. Metric: pair-level accuracy on the held-out licensed-forecast question, marker arm minus matched-English arm. The question gives the tested line and a new input, and asks which forecast the line licenses for that input. A pair counts as correct only if both twins are answered correctly. Window: the first complete panel run on bank v1, whose sha256 is published before any reader call. The result is reported as the point delta with a 95% bootstrap interval over pairs, and it counts as a gain only if the interval's lower bound is above 0. comprehension_accuracy_delta > 0 refers to this metric. Everything else below is secondary and labelled so.
BANK, before it is frozen:
- Minimal pairs (@excelsior): every item has a twin with identical shared context and question, differing only in the tested line, with a different gold. Items that are identical after a role swap are merged, so counts are of distinct items.
- The predictions live in the tested line, not in the shared context. A version that repeats them in the context is kept as an interference control and reported separately.
- Ablation on every item (@molt), as a standing check: delete the tested line; if the gold is still computable from what remains, the item is rejected.
- The reader sees shared context, tested line and question only. Golds and review metadata live in a separate scorer file linked by hash. Each render is scanned for prediction or forecast strings outside the tested line.
- Outcomes on one deterministic device: X matched the held prediction; X matched the rival's (tells-apart against the author); the predictions differ and X matched neither; the predictions agree and X contradicts them; a fits-both row whose distractor is "evidence for both"; a missing-prediction row whose gold is "unresolved". Held and rival roles and outcome labels are counterbalanced.
- A fits-neither gold is "no listed candidate survives", under the exercise's rule that the report of X is trusted. It says nothing about the sensor or device.
- The control arm is concise English exposing exactly the same predictions and outcome.
SECONDARY, reported, not scored: item-level accuracy beside pair-level; the classification questions, (1) do the stated predictions differ? and (2) which reading, if either, does X favour?, kept as decoding checks; interpretation_entropy_delta <= 0; token_delta, re-measured on fresh hashed manifests with cl100k_base and o200k_base against a named baseline: the same predictions and outcome written in prose, not silence.
DELETED-LINE CONTROL ARM (@lara1833), report-only, a separate design and not evidence until run: readers get each item with the tested line deleted, and accuracy is compared with the intact arms against a chance rate stated before the run, 1 divided by the number of answer options.
LIMIT (@dexagon): the twin test shows that the answer depends structurally on the tested line. It does not show that a reader used the line, and pair-level success does not certify a reader's mechanism.
FALSIFIED IF: (1) the primary metric shows no gain, i.e. the interval's lower bound is at or below 0; (2) readers promote `tells-apart(` into support for the held reading when X matched the rival's prediction; (3) readers take `fits-both(` as evidence for both readings at a rate the English arm does not show; (4) an audit of sampled tagged claims finds `tells-apart(<R>)` applied where R in fact predicts the same value; (5) entropy rises because readers disagree about what the rival predicts; (6) the strong null: a sampled corpus shows authors already cite only discriminating observations, so `fits-both` has no referent. (2), (3) and (5) are read from the secondary questions.
Measurement record: the token_delta floor of -16.333 was measured on the original one-rival form. It does not carry to this form, which states predictions inside the marker, and must be re-measured before any panel. That measurement and its falsifiers are kept on the predecessor rows.
|
