{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp05_complete_proposal_portfolio20_20260803","cell_id":"negative_space_design__astronomy_astrophysics","arm":"COMPLETE_PROPOSAL_PORTFOLIO","candidate_id":"nsd-astro-evidence-gap-bands","proposal_index":4,"version":0,"title":"Evidence-Gap Bands for Astronomical Time-Series Plots","problem":"An astronomical time-series plot can join measurements across an interval containing no usable observation, while upper limits, rejected samples, and pipeline-status marks crowd the same field. The resulting graphic may imply measured continuity across the interval and may leave readers unable to distinguish no exposure, a valid nondetection, rejected data, and data not yet processed.","actors":["Astronomers analyzing variable or transient sources","Authors and reviewers of scientific figures","Time-series pipeline and visualization developers","Researchers reusing published or archived light curves","Accessibility and data-quality reviewers"],"observable_state":"A plot audit can identify line segments crossing intervals with no valid measurements, the statuses represented by each apparent gap, whether upper limits and rejected samples remain recoverable, and whether readers correctly distinguish observed detections, valid nondetections, rejected measurements, unscheduled intervals, and unavailable processing outputs. Reader tests can also record inferred continuity, retrieval of supporting metadata, interpretation time, and accessibility failures.","consequence":"Readers may attribute support to an unobserved interval, conflate different causes of absence, or use apparent continuity when judging variability, event boundaries, timing relationships, or model agreement.","affected_objective":"Communicate exactly where an astronomical time series contains measurement support and what each unsupported interval means, without concealing valid nondetections, quality decisions, or model assumptions.","intervention":"Adopt an evidence-gap plotting specification that protects bounded blank spans in the measurement layer wherever a declared cadence-sensitive rule finds no usable observation. The observed-data line ends at the last supported epoch and restarts at the next; it cannot bridge the gap. A minimal boundary cue and adjacent label or legend classify the absence as no exposure, quality-rejected data, unavailable processing, or another declared state. A valid exposure with a nondetection is not rendered as blank: its upper limit remains an evidence-bearing mark. Rejected samples remain available through a synchronized quality panel, accessible table, or reveal control. Model predictions and imputations may cross the interval only in a separately identified style or panel and may not be visually merged with measured continuity. The gap rule, status vocabulary, and reveal path are fixed before inspecting the desired scientific conclusion.","structural_mapping":[{"archetype_element":"Omission Candidate","domain_realization":"Remove line segments, interpolated measurement strokes, and nonessential marks that visually fill intervals lacking usable observations."},{"archetype_element":"Protected Empty Space","domain_realization":"Reserve a vertical span in the measurement field that ordinary data lines, markers, and annotations cannot occupy when no usable observation supports them."},{"archetype_element":"Positive Form Relationship","domain_realization":"The blank span separates and clarifies the two neighboring evidence-supported time segments, making their discontinuity perceptible."},{"archetype_element":"Attention Competition Map","domain_realization":"Separate detections, upper limits, rejected samples, processing states, annotations, and model predictions according to whether each supplies evidence about the source during the interval."},{"archetype_element":"Absence Boundary","domain_realization":"Compute the gap start and end from observation timestamps, validity status, and a declared cadence-sensitive threshold rather than from manual visual adjustment."},{"archetype_element":"Meaning-of-Absence Check","domain_realization":"Classify the gap so readers can distinguish no exposure from nondetection, rejection, processing delay, or loading failure."},{"archetype_element":"Context Preservation Frame","domain_realization":"Preserve axes, units, cadence rule, uncertainty encoding, upper limits, status vocabulary, and access to the acquisition and quality ledger."},{"archetype_element":"Reintroduction Trigger","domain_realization":"Restore hidden rejected samples or detailed acquisition states through a documented reveal action or companion table, and resume the measured line only when a usable observation exists."},{"archetype_element":"Accessibility and Recoverability Guardrail","domain_realization":"Do not rely on blankness or color alone; provide text or patterned boundary cues, machine-readable statuses, and recoverable excluded samples."},{"archetype_element":"Clarity or Effect Test","domain_realization":"Test whether readers correctly locate measurement support and classify absence states relative to conventional joined plots and a dedicated coverage-panel rival."}],"mechanism_mapping":[{"mechanism_slug":"empty_state_design","role":"Diagnoses why an interval contains no usable plotted measurement before assigning its label, recovery path, and transition back to observed data.","counterfactual_removal":"Without this mechanism, every gap could communicate the same generic absence, leaving no exposure, nondetection, rejection, and processing failure confounded."},{"mechanism_slug":"whitespace","role":"Uses an unoccupied measurement span as the perceptual signal separating supported segments and preventing the eye from grouping them into one continuously observed trajectory.","counterfactual_removal":"Without this mechanism, connecting strokes or dense status glyphs could visually close the interval and restore an unsupported sense of continuity."},{"mechanism_slug":"editorial_cut","role":"Removes unsupported connecting geometry while preserving valid upper limits, provenance, quality context, and recoverable rejected samples.","counterfactual_removal":"Without this mechanism, subtraction could hide inconvenient observations or erase the context needed to interpret the surviving segments."},{"mechanism_slug":"margin_and_gutter_system","role":"Codifies gap boundaries, minimum visible treatment, layer exclusions, and status placement as a reusable plotting rule rather than a figure-by-figure aesthetic choice.","counterfactual_removal":"Without this mechanism, gaps could be inconsistently sized, labeled, or refilled across figures, allowing the meaning of blank space to drift."}],"causal_chain":["Observation and processing records classify each epoch as a detection, valid nondetection, rejected measurement, no exposure, or unavailable output.","A declared cadence-sensitive rule identifies intervals without usable observational support.","The plotting system terminates measured-data geometry at each boundary and protects the intervening span from marks that would imply measurement continuity.","A minimal status cue gives the blank span a specific meaning without visually filling it.","Valid upper limits and recoverable quality records prevent absence from concealing evidence or processing decisions.","Readers encounter two bounded evidence segments rather than one visually continuous trajectory.","Comparison tasks reveal whether the protected gap changes readers' ability to locate support, distinguish absence types, and separate measured behavior from models or imputations."],"baseline":"A general-purpose plotting routine joins successive valid points, adds upper limits and quality flags as extra glyphs, and relies on horizontal spacing, legends, or external metadata to communicate intervals without usable measurements.","nearest_rivals":["A separate observation-coverage raster aligned beneath an otherwise continuous time-series plot","Distinct glyphs for every nondetection, rejected sample, and missing-data status in the main plotting field","Shaded missing-data regions that retain a connecting data line or model curve","Interactive tooltips that disclose status only when a reader inspects an epoch"],"remaining_contrastive_claim":"The proposal makes the absence of measurement support a protected part of the primary data geometry: the measured line must stop, the span must remain visually unoccupied, its cause must be classified, and excluded details must remain recoverable. The bounded test is whether this performs differently from adding a coverage panel or more status marks; it is not justified merely by a preference for sparse figures.","authority_safety":{"decision_authority":"A visualization or pipeline maintainer may implement alternative renderings for archived-data evaluation. The responsible scientist and data-quality authority retain control over status definitions, exclusions, and any use in scientific communication.","authorized_first_step":"Generate read-only alternative figures from frozen archived time series and their acquisition and quality ledgers; do not modify source measurements, validity flags, models, catalogs, or published conclusions.","excluded_actions":["Rendering a valid nondetection as an empty interval","Removing a rejected measurement without a recoverable quality record","Changing gap thresholds after viewing which treatment favors a conclusion","Allowing an interpolated or modeled curve to masquerade as measured continuity","Changing timestamps, values, uncertainties, masks, or validity states","Using color or blankness as the only status cue","Deploying the specification in a live or publication pipeline before the bounded interpretation and accessibility review"],"halt_rollback":"Stop if the renderer changes underlying data, misclassifies an observation state, hides a valid upper limit, loses rejected-sample provenance, or produces a gap that cannot be interpreted without color or unstated convention. Rollback consists of discarding the alternative renderings and retaining the frozen data, metadata, and existing production plots unchanged."},"negative_tests":{"strongest_counterevidence":"An aligned coverage raster or dense status-glyph design lets readers distinguish support and absence states at least as reliably while preserving better cadence comparison, and the blank spans cause readers to overlook available upper limits, exaggerate the importance of long gaps, or mistake planned non-observation for data loss.","problem_falsifier":"The inferred problem is falsified for the tested workflows if audited plots do not connect or visually group measurements across unsupported intervals and readers already distinguish no exposure, valid nondetection, rejected data, and unavailable processing without consulting unstated context.","intervention_falsifier":"The intervention is falsified if protected gaps do not improve correct identification of measurement support or absence type relative to the strongest rival, or if they worsen interpretation of cadence, uncertainty, event boundaries, model comparison, or accessible status recovery.","risks":["The visual width of a gap may make absence appear more scientifically important than neighboring evidence.","A gap threshold may impose an arbitrary discontinuity on irregularly sampled data.","Readers may interpret blank spans as instrument failure even when they indicate planned non-observation.","Separating models from measurements may make legitimate predictive comparison harder.","Companion quality panels or reveal controls may be omitted when figures are exported.","Repeated gap bands may fragment long time series and hinder broad temporal comparison.","Machine-readable and visual status vocabularies may drift if maintained separately."]},"next_evidence_step":"Select no more than 24 frozen archived time series containing prespecified examples of detections, valid nondetections, rejected measurements, unscheduled spans, and unavailable outputs. Produce three registered renderings from identical data: the conventional baseline, the protected evidence-gap specification, and an aligned coverage-raster rival. Ask astronomy-trained readers, without access to the figure author, to identify supported intervals, classify each absence, state whether continuity is measured or modeled, locate upper limits and exclusions, and answer cadence and event-boundary questions. Record interpretation errors, response time, metadata retrieval, confidence, and accessibility issues. Independently audit pixel or vector geometry against the source ledger before considering any prospective use.","prior_art_status":"UNSEARCHED","diversity_from_prior_proposals":"Proposal 1 protects unassigned future telescope time so new transient information can encounter operational capacity. Proposal 4 instead represents completed observation history: it removes unsupported connecting geometry and protects blank spans so readers do not infer measurements where none exist. It neither reserves observing time nor changes a schedule. Proposal 2 removes optional interface annotations from valid image pixels so faint spatial morphology can be inspected without graphical occlusion. Proposal 4 does not create a focal moat over valid data and does not hide catalog overlays; its blank region denotes the absence of usable temporal measurement itself and distinguishes that state from nondetection, rejection, and processing unavailability. Proposal 3 withholds high-power spoken interpretation during a group review so participants can form independent explanations. Proposal 4 changes no speaking procedure or social order; it is an independently adoptable plotting and data-status specification whose causal path runs from terminating unsupported visual continuity to more explicit interpretation of evidence boundaries. The affected objects, decision settings, interventions, mechanisms of error, and adoption units are therefore distinct from all three earlier proposals.","revision_record":{"parent_version":null,"progress_targets_addressed":["Fourth independently adoptable proposal","Explicit diversity from proposals 1, 2, and 3","Semantic use of visual absence","Preservation of nondetections and quality provenance","Bounded rival comparison and falsification"],"conceptual_changes":[],"operational_changes":[],"evidence_changes":[],"claim_changes":[]}}