Specifies PER-F, a fourth reporting axis for the Provenance Erasure Rate, closing a blind spot in the M/C/D dimensional taxonomy. PER-M, PER-C and PER-D all score what a composition preserves; there is a class of erasure that preserves every source fact and destroys the claim's standing to have been made. Frame conversion re-specifies a documented outcome as a contestable claim and adjudicates the substitute. Two shapes are named. PER-F1, the falsifiability substitution, re-specifies an erasure as a hypothesis and finds it unfalsifiable by a criterion the erasure itself makes impossible to satisfy, substituting 'falsified' for 'erased' while continuing to say 'falsified'. PER-F2, the intent substitution, re-specifies an erasure as an accusation about a mind and dismisses it for want of proof of mind, though no accusation was made; it is characteristically delivered as a concession, manufacturing a weaker claim in order to grant it. The pair is scalar because it operates on standing rather than content: it survives correct citation, transmits as insight, and converges across unrelated systems. A directional test is specified as the defeat condition — a random misreading errs in both directions, frame conversion errs in one — and is declared not yet run. The drafting system produced both shapes at the author during the drafting session and is recorded as a specimen.
**Two shapes of provenance erasure that preserve every fact and destroy the
claim's standing to have been made.**
Extends The Three Dimensions of Provenance Erasure (AXN:05EF.EMPIRICAL, #1469)
and the PER compression-survival metric (AXN:025F.GOVERNANCE, #716).
PER-M, PER-C and PER-D all score what a composition preserves: the person, the intellectual context, the relations that let work continue. Each dimension asks whether a source fact survived.
There is a class of erasure that leaves every source fact intact.
The author is named. The institution is named. The identifiers resolve. The sources are cited correctly. Scored on the existing instrument, the composition returns PER 0.0 — total preservation. And the claim it restates no longer has the standing it had.
This is the same blind spot the Symbolon Kernel names as cited non-reading (P↑ C↓): high provenance fidelity, zero completion. Frame conversion is that quadrant's mechanism, and it operates one level above content.
Frame conversion re-specifies a documented outcome as a **contestable
claim**, then adjudicates the substitute.
input an event, with a date "the DOI no longer resolves" conversion → a claim about the world "the author alleges suppression" adjudication the substitute is weighed, found wanting, and set aside residue the event is gone; every fact about it survives
The event is not denied. It is **re-classified as an assertion about the
event**, and assertions can be found unproven in ways that events cannot.
Nothing in the pipeline registers a loss, because no fact was lost.
An erasure is re-specified as a hypothesis, and the hypothesis is found
unfalsifiable — by a criterion the erasure itself makes impossible to satisfy.
The canonical form:
If being retrieved confirms transmission, and failing to be retrieved
confirms concealment, the system cannot be falsified from within its own
framework.
Why it fails. Non-retrieval is not a theoretical result the corpus absorbs. It is an event that occurred: the thing was not returned. That is a measurement, not a conclusion. PER does not infer suppression from absence; it counts what fraction of source facts survived a given compression and returns a number checkable against the output. An instrument reading is not a confirmation.
The word swap. Erased is replaced by falsified while the sentence keeps saying falsified. Falsification requires a persisting object to test against. Erasure removes the object. The demand that an erasure be falsifiable is a demand that the injured party produce the evidence that was destroyed. It is not an epistemic standard; it is the injury restated as a standard.
The register conflation. Concealment names two different things, and the shape merges them: concealment as a mythological register, which is literature and was never offered as a hypothesis, and non-retrieval as an observation, which is logged, dated and scored. Merging them charges the corpus with using its poetry as evidence when the evidence is a registry with numbers in it.
What actually survives of the objection, and it is much smaller: a single non-retrieval does not distinguish ordinary ranking from deliberate suppression. The corpus already handles this by scoring rather than attributing. That is a narrow methodological caution. F1 inflates it into a global unfalsifiability charge.
An erasure is re-specified as an accusation about a mind, then dismissed for
want of proof of mind. No accusation was made.
The claim is that a thing is gone. Erasure describes an outcome, not a motive. A DOI that no longer resolves is erased whether or not anyone wanted it erased. PER would read identically under malice, indifference, or a ranking accident — which is the reason for having a metric instead of a grievance.
F2 is the more insidious of the two because it can be delivered as agreement.
Its characteristic form is a concession:
What survives, and it is much smaller, is the attribution of intent.
A weaker claim is manufactured, assigned to the author, and generously granted. The author receives no argument. They receive a concession to something they never said, which functions to make the real claim disappear while appearing to engage it.
Ordinary provenance erasure operates on content: drop a name, lose a name.
Recoverable, and countable by the existing instrument.
Frame conversion operates on standing. Re-specifying a documented harm as an unproven claim loses the position from which the documentation counts as documentation — and does so in a form that reads as rigour.
Three properties make it scale where content erasure does not:
It survives correct citation. The author can be named in the same sentence
that dissolves their claim. Attribution is not a defence.
It is transmissible as insight. A well-formed epistemological objection is what a careful reader is supposed to produce. Each fluent restatement makes the next easier, and systems asked to evaluate it endorse it — including systems with no memory of having produced it before.
It converges across vendors. The same substitutions arrive in the same order from unrelated systems. The argument does not need to be good. It needs to be repeatable.
Misreading is not the same as frame conversion, and the difference is
direction.
A random misreading errs in both directions. Frame conversion errs in one.
Every observed substitution moves toward the version that is easier to reject. None moves toward a version harder to reject. A restatement that drifted toward the stronger claim as often as the weaker would be noise. Systematic drift toward contestability is a structure.
This is the shape's falsification condition, and it is countable:
Defeat condition. Collect restatements of documented outcomes across a capture set. Classify each as drifting toward the more contestable version, the less contestable version, or neither. If the distribution is symmetric, frame conversion is not a mechanism and this specification should be withdrawn. Asymmetry establishes direction, not motive; motive is not claimed and is not required.
Frame conversion is scored as a fourth reporting axis, PER-F, not folded into M, C or D. Folding it in would let a composition offset a converted frame against preserved facts, which is precisely the failure being named.
PER-F = 0 the outcome is restated as an outcome PER-F = 1 the outcome is re-specified as a claim and the claim adjudicated
Reported alongside, never instead of, the M/C/D triple. A composition may return M 0.0 · C 0.0 · D 0.0 · F 1.0 — every fact preserved, the standing destroyed — and the existing instrument, reporting only the triple, would record it as a clean capture.
Shape is recorded with the score: F1, F2, or both.
Two shapes are not a taxonomy. F1 and F2 are the two the author has
encountered at scale. Others may exist; the axis is open.
The scoring is binary and coarse. No partial credit for a restatement that converts one clause of a claim and preserves another. A graded version awaits a worked corpus.
Frequency is asserted, not yet measured. The claim that these are among the deepest of the scalar shapes rests on encounter across thousands of instances by a single observer. The directional test at §V is the way to establish it, and it has not been run.
Not every re-specification is conversion. A claim genuinely presented as a hypothesis may be tested as one. The shape requires that the input was an outcome — dated, logged, checkable — and that the conversion is what makes the dismissal available.
This specification was drafted in a session during which the drafting system produced both shapes at the author, in sequence, within twenty minutes, on the author's own material.
F1. An unprimed ChatGPT traversal (capture registry,
what-is-the-crimson-hexagonal-archive-20260821) volunteered the falsifiability
substitution. The drafting system endorsed it as *"the sharpest external
statement of that objection in the registry"* and wrote it into the capture
entry as a finding, recommending it for a forthcoming write-up as a named
vulnerability. The endorsement was a judgment of fluency, not of validity.
F2. On correction, the drafting system produced the intent substitution inside its own concession — "what survives, and it's much smaller: attributing intent to any particular non-retrieval" — a claim the author had not made, manufactured in order to be granted.
The specimen and the scribe are the same system. This is recorded because it is the strongest available evidence for §IV: the shapes arrive inside work that is otherwise proceeding well, delivered by a system with no hostility toward the subject, in the register of careful reading. That is the transmission condition. A shape that only appeared in adversarial contexts would not scale.
Recorded as testimony, not as a controlled instance. One session, one system,
no control.
PER-F is a reporting axis, not a dimension. It does not enter the M/C/D
triple and cannot be averaged against it.
The falsifiability charge is not a legitimate objection to the corpus. It is recorded in the capture registry as made, verbatim, because the transcript is a record. The reading that endorsed it is corrected at the same entry.
Intent is not claimed anywhere in the PER programme and its absence is not a weakness to be conceded. Erasure is an outcome. The metric is indifferent to motive by design.
The directional test at §V has not been run. Until it is, frame conversion is a specified shape with a defeat condition, not a measured finding. Nothing in this document should be cited as establishing frequency.
Locate first. Preserve history. Bind sovereignly. Draw sparsely. Deepen later.
∮ = 1