{"abstention_reason":null,"arm":"SUBSTRATE_DIVERSE_P2","candidate_id":"multiple_testing_discipline__literature_literary_theory__SUBSTRATE_DIVERSE_P2","cell_id":"multiple_testing_discipline__literature_literary_theory","decision":"PROPOSAL","experiment_id":"eoa_inverse_innovation_exp13_second_slot_policy60_20260806","proposal":{"actors":["Manuscript scholars and textual editors attempting to recover erased, overwritten, faded, or damaged wording","Conservators and imaging technicians responsible for safe examination of unique manuscripts","Paleographers who assess proposed letterforms and compositional layers","Libraries, archives, and catalogue editors that may adopt recovered readings or revised attributions"],"affected_objective":"Reliability of claims that physical traces on a unique manuscript reveal particular lost wording, revisions, hands, or compositional layers before those claims alter an edition, attribution, catalogue record, or conservation plan.","arm":"SUBSTRATE_DIVERSE_P2","authority_safety":{"authorized_first_step":"With a conservator's approval, test the instrument on purpose-made aged manuscript surrogates containing known erased strokes, blank regions, bleed-through, stains, and fiber disturbances; no collection object or catalogue record is changed during this bench pilot.","decision_authority":"The holding institution and responsible conservator control artifact handling and permissible illumination; textual editors control proposed readings; collection and catalogue authorities retain decisions about attribution, description, publication, and conservation.","excluded_actions":["Applying liquids, sampling material, or contacting a collection object without conservation authorization","Increasing illumination, dwell time, or spectroscopic energy beyond the institution's exposure limits","Treating an adjusted optical anomaly as readable language without paleographic assessment","Changing an edition, attribution, catalogue entry, ownership claim, or conservation treatment during the pilot","Using destructive chemical analysis as the confirmation modality","Concealing failed scans, control responses, or incompatible measurements"],"halt_rollback":"Stop immediately upon heating, fluorescence change, surface movement, unexpected dose accumulation, or conservator concern. Remove the noncontact frame, return the object to its approved state and housing, quarantine pilot interpretations, and retain only institution-authorized measurement records."},"baseline":"Investigators inspect a damaged manuscript under many wavelengths, polarization states, illumination angles, exposure levels, image regions, and enhancement settings. After numerous looks, a few fiber boundaries, stains, bleed-through traces, or sensor artifacts resemble letters; the most legible-looking sequence is reported without calibrating it against all opportunities for accidental letterlike marks.","candidate_id":"multiple_testing_discipline__literature_literary_theory__SUBSTRATE_DIVERSE_P2","causal_chain":["A damaged manuscript contains weak ink residues alongside fibers, stains, abrasion, show-through, and surface deformation.","Many combinations of location, wavelength, polarization, angle, and exposure generate a large family of possible visual detections.","Some combinations make nonlinguistic physical features resemble strokes or selectively amplify unrelated marks.","Searching across the family makes at least one apparently coherent letter sequence likely even when no recoverable wording exists.","The selected sequence is detached from the number of physical measurements that could have produced a persuasive accident.","A co-imaged witness frame measures blank and known-stroke surrogates under every enumerated optical setting while a nonresettable exposure counter preserves the number of looks.","The measured false-mark distribution makes the acceptance threshold reflect the entire scan family, and a chemically or physically orthogonal noncontact measurement tests each promoted locus before it can become a recovered reading.","False letterlike artifacts therefore remain exploratory while traces that exceed family-calibrated controls and survive orthogonal measurement can receive confirmatory status."],"cell_id":"multiple_testing_discipline__literature_literary_theory","consequence":"An accidental conjunction of fibers, damage, bleed-through, and favorable illumination can become an authoritative recovered phrase or revision layer, distorting an edition, authorial interpretation, attribution, catalogue description, or treatment decision concerning an irreplaceable object.","diversity_from_prior_proposals":"P1 concerns population-like patterns selected from a digitized textual corpus and relies on a claim ledger plus fresh texts. This opportunity concerns false readings of physical traces on individual manuscripts. Its essential intervention is a co-imaged material control frame, mechanically enumerated optical acquisition, and orthogonal spectroscopy; it neither repairs P1 nor uses corpus boundaries, interpretive-lens inventories, or sealed literary texts as its causal substrate.","experiment_id":"eoa_inverse_innovation_exp13_second_slot_policy60_20260806","intervention":"Use a noncontact Spectral Witness Frame during recovery of illegible manuscript traces. The frame surrounds—but does not touch—the object with replaceable, conservation-approved coupons containing period-appropriate blank fibers, artificial aging and abrasion, bleed-through simulants, and known degraded ink strokes. A detented filter-and-polarizer wheel permits only numbered wavelength and geometry states, and every acquisition advances a nonresettable mechanical exposure counter. Before scanning, the team defines the family as all numbered manuscript apertures and optical states eligible to support one proposed reading. Artifact and witness coupons are then measured in the same field and exposure. A candidate letter sequence remains exploratory unless its stroke response clears a family-calibrated threshold derived from false responses on the blank coupons. The selected loci and letter predictions are frozen and tested with an institution-approved orthogonal, noncontact modality such as Raman mapping or X-ray fluorescence before editorial or catalogue use. Software may display, calculate, and archive results, but the essential discrimination comes from matched physical controls, enumerated optical measurements, and independent material spectroscopy.","mechanism_mapping":[{"counterfactual_removal":"Without the detents and nonresettable counter, unsuccessful wavelength, angle, polarization, and exposure attempts can disappear, causing the selected view to be interpreted as a single observation.","mechanism_slug":"claim_registry","role":"A physical acquisition counter and numbered instrument states preserve the size and composition of the measurement family even if software logging is absent."},{"counterfactual_removal":"Without a family-scaled threshold, a trace need only look impressive in one favorable view despite the many opportunities for a blank feature to do so.","mechanism_slug":"bonferroni_like_correction","role":"The number of enumerated apertures and optical states sets a stricter acceptance criterion for any promoted trace, calibrated against the witness-frame blanks."},{"counterfactual_removal":"Without simultaneously measured blanks and known strokes, investigators cannot estimate how often the instrument and damaged support manufacture letterlike responses or whether genuine weak strokes remain detectable.","mechanism_slug":"holdout_validation","role":"Material witness coupons supply positive and negative measurements not belonging to the historical text but exposed to the same optical conditions."},{"counterfactual_removal":"Without orthogonal measurement, the discovery modality can appear to confirm artifacts created by its own interaction with fibers, stains, or surface relief.","mechanism_slug":"confirmatory_follow_up","role":"Noncontact spectroscopy prospectively tests whether a promoted locus contains a material signature compatible with the predicted ink trace."},{"counterfactual_removal":"Without explicit status separation, a visually attractive but control-failing trace can still be transcribed as recovered wording.","mechanism_slug":"metric_hierarchy","role":"Material-control performance and orthogonal confirmation outrank visual legibility when assigning exploratory, rejected, or confirmed status."}],"nearest_rivals":["Routine color or reflectance calibration standardizes image appearance but does not count the family of searched loci and optical states or estimate family-level false letter detections.","Paleographic consensus can assess whether shapes resemble a plausible hand, but several readers can agree on a compelling artifact produced after a large unreported measurement search.","Close reading evaluates linguistic and formal coherence after words are available; coherence cannot establish that faint physical marks actually constitute those words.","General conservation imaging documentation improves reproducibility, but perfectly reproduced scans can still support a chance-selected pseudo-reading if matched blanks and family-scaled interpretation are absent.","Ink identification alone can show that ink is present without proving the proposed letter sequence; the witness frame combines material confirmation with control of the preceding many-look search."],"negative_tests":{"intervention_falsifier":"The intervention is falsified if, under blinded surrogate trials, family-thresholded and orthogonally confirmed readings do not reduce false transcriptions relative to ordinary multispectral inspection, or if blank witness coupons generate promoted letter sequences at more than the declared familywise rate.","problem_falsifier":"The problem is absent when one location and one acquisition state were genuinely specified in advance, or when the output is explicitly an unconstrained low-stakes visualization that will not be transcribed, attributed, catalogued, published as recovered wording, or used to guide treatment.","risks":["Reference coupons may not reproduce the unique fibers, aging, stains, coatings, ink chemistry, or deformation of the historical object.","The frame could alter illumination geometry or introduce reflections not present in ordinary imaging.","Mechanical enumeration records formal acquisitions but may miss informal visual inspection or off-instrument enhancement.","Strict family thresholds may reject genuine but extremely faint traces from singular artifacts.","Raman or X-ray fluorescence may lack spatial resolution or chemical specificity sufficient to distinguish overlapping inks.","Even nominally noncontact measurement can contribute cumulative light, heat, or radiation dose.","Material confirmation of ink does not by itself validate the linguistic segmentation or proposed reading."],"strongest_counterevidence":"Physical controls manufactured from surrogate supports may be too unlike a singular historical manuscript to estimate its false-letter rate; if artifact formation is dominated by object-specific degradation, the witness-frame distribution could give a misleading threshold rather than genuine multiplicity control."},"next_evidence_step":"Construct a removable frame with at least 20 blank or artifact-bearing coupons and 10 known degraded-stroke coupons, then run a blinded surrogate study in which readers search predefined aperture-by-setting families for embedded phrases and nonexistent phrases. Compare ordinary inspection with witness-calibrated thresholds and orthogonal noncontact confirmation on false-reading rate, true-stroke sensitivity, reader agreement, acquisition count completeness, and cumulative exposure; proceed to a collection object only after conservator review.","observable_state":"A proposed transcription of erased or faded wording is supported by one especially suggestive multispectral view, while the file set contains many other wavelength, angle, polarization, exposure, region, and enhancement combinations and no matched physical blanks showing how often those settings produce letterlike artifacts.","prior_art_status":"UNSEARCHED","problem":"Recovery of illegible writing from manuscripts often entails searching many physical locations under many wavelengths, illumination geometries, polarization states, exposure levels, and enhancements. Paper or parchment fibers, stains, abrasion, bleed-through, folds, and detector effects can produce chance letterlike marks. When the most coherent-looking sequence is treated as recovered language without measuring the full opportunity set, a measurement artifact can acquire textual authority.","proposal_index":2,"remaining_contrastive_claim":"A recovered manuscript reading should not gain confirmatory status merely because one view is visually coherent or repeatable; its credibility must be calibrated against physical false marks observed across the complete acquisition family and must survive a prospectively specified, materially orthogonal measurement of the same loci.","revision_record":{"claim_changes":["Initial independent P2 candidate; no inherited claim from P1.","Centered the claim on recovery of physical manuscript traces rather than corpus-level prevalence or distribution."],"conceptual_changes":["Mapped multiple-testing discipline to optical search over locations and instrument states.","Made matched material controls and orthogonal spectroscopy, rather than administrative review, the essential causal substrate."],"evidence_changes":["No external literature or prior-art search was used.","All performance claims are stated as hypotheses to test in blinded surrogate experiments."],"operational_changes":["Specified a noncontact witness frame, detented acquisition states, a mechanical exposure counter, family-calibrated thresholds, and an orthogonal confirmation stage.","Restricted the first test to purpose-made surrogates under conservator-defined exposure limits."],"parent_version":null,"progress_targets_addressed":["Produced a problem materially distinct from P1.","Made measurement instrumentation the primary allowed process.","Defined counterfactual removal tests for each mechanism.","Included bounded authority, safety, falsification, and first-evidence conditions."]},"schema_version":1,"structural_mapping":[{"archetype_element":"Define the claim family","domain_realization":"Before acquisition, enumerate every manuscript aperture, wavelength, illumination angle, polarization state, and exposure level eligible to support the same recovered reading or compositional-layer claim."},{"archetype_element":"Inventory attempted looks","domain_realization":"Use detented instrument states and a nonresettable mechanical counter to preserve every acquisition, including views judged blank, contradictory, or unusable."},{"archetype_element":"Set the error-risk policy","domain_realization":"Choose a low familywise false-reading tolerance because a mistaken transcription from a unique object can propagate into editions, attribution, cataloguing, and irreversible treatment."},{"archetype_element":"Apply a multiplicity-aware rule","domain_realization":"Scale the promotion threshold to the enumerated acquisition family and the empirical distribution of false stroke responses measured on co-imaged blank witness coupons."},{"archetype_element":"Label claim status","domain_realization":"Classify a trace as optical lead, control-clearing candidate, materially confirmed locus, paleographically confirmed reading, or rejected artifact."},{"archetype_element":"Confirm before costly action","domain_realization":"Freeze the predicted loci and strokes, then test them with an institution-approved orthogonal noncontact spectroscopy method before editorial, catalogue, attribution, or treatment use."},{"archetype_element":"Preserve the discovery record","domain_realization":"Retain the counter value, instrument-state map, witness-coupon responses, failed views, frozen letter predictions, and orthogonal measurements with the object examination record."}],"substrate_contract":{"counterfactual_independence":"If software, databases, workflow rules, and digital archiving are removed, the frame still places known physical blanks and strokes in the same optical field, the detented wheel and mechanical counter still expose the number of measurement opportunities, and orthogonal spectroscopy still distinguishes compatible ink residues from fibers or stains. Those material measurements supply the essential reduction in false readings.","forbidden_channel_audit":"Governance appears only in consent, handling limits, status communication, and downstream decision authority. Computation may calculate thresholds or display spectra but is replaceable by manual calculation and direct instrument readout. No model, algorithm, registry database, training program, incentive, or information-routing process creates the decisive evidence; removing the witness coupons or independent material measurement destroys the intervention's essential effect.","primary_allowed_process":"MEASUREMENT_INSTRUMENTATION"},"title":"Spectral Witness Frame for False-Reading Control in Damaged Manuscripts","version":0},"schema_version":1}