{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp09_archetype_breadth150_20260804","cell_id":"progressive_disclosure__accounting_auditing","arm":"BREADTH_PROBE_ONE_SHOT","candidate_id":"progressive_disclosure__accounting_auditing__P1","proposal_index":1,"version":0,"title":"Audit Review Evidence Ladder","problem":"During supervisory review of a complex audit workpaper, the relevant assertion, preparer conclusion, procedures, population reconciliation, sample evidence, exceptions, accounting analysis, and administrative checklist items may appear in one flat package. When reviewers must process every detail before knowing where judgment or risk is concentrated, they can lose orientation, review mechanically, or overlook a consequential exception even though its evidence is present.","actors":["Audit workpaper preparer","Senior auditor or audit manager reviewing the workpaper","Engagement partner responsible for the audit opinion","Engagement quality reviewer inspecting significant judgments"],"observable_state":"The candidate problem is present when reviewers repeatedly search or scroll to reconstruct the relationship between a conclusion and its evidence; supporting documents are opened without a stable indication of their parent assertion or procedure; unresolved exceptions have the same visual prominence as routine completed items; reviewers ask for information already contained in the package; or a sign-off can occur without explicitly visiting flagged assumptions, exceptions, and contradictory evidence.","consequence":"Review attention can be consumed by routine detail while material exceptions, unsupported assumptions, or gaps between the stated conclusion and underlying evidence receive insufficient scrutiny, weakening the basis for the engagement's audit conclusion and increasing avoidable review rework.","affected_objective":"A traceable, appropriately skeptical, and sufficiently evidenced review of audit conclusions relevant to detecting or responding to material misstatement risk.","intervention":"Create a read-only, progressively disclosed review view for each significant audit area. The default layer shows the assertion, scope, preparer's conclusion, unresolved items, contradictory evidence, and any condition that could change the responsible review action. A second layer exposes procedure design, population-to-ledger reconciliation, coverage, materiality context, and summarized results. A third layer exposes individual selections, source documents, calculations, accounting analysis, and comment history. User-selected drilldown reveals ordinary detail; risk and exception triggers automatically expose mandatory warnings before sign-off. Breadcrumbs preserve the assertion-procedure-item relationship, and reviewers retain a one-action full-workpaper view and downloadable underlying record.","structural_mapping":[{"archetype_element":"Current need","domain_realization":"The reviewer's immediate need to determine whether a specific audit conclusion is supportable and where skepticism should be concentrated."},{"archetype_element":"Information Layer","domain_realization":"Conclusion-and-risk summary, procedure-and-coverage detail, and item-level evidence are separate but linked review depths."},{"archetype_element":"Reveal Trigger","domain_realization":"Reviewer drilldown, movement to a review stage, an unresolved exception, contradictory evidence, a materiality-related threshold, or an attempted sign-off reveals the relevant depth."},{"archetype_element":"Critical Detail Exception","domain_realization":"Unresolved exceptions, contradictory evidence, missing required evidence, scope limitations, and conditions affecting the responsible conclusion bypass ordinary hiding and remain visible before sign-off."},{"archetype_element":"Orientation Cue","domain_realization":"Persistent labels and breadcrumbs identify the financial-statement area, assertion, procedure, sample item, disclosure depth, and active filters."},{"archetype_element":"Drilldown Path","domain_realization":"A reviewer can move from conclusion to procedure, from procedure to population or sample, and from sample to source evidence and preparer rationale."},{"archetype_element":"Return Path","domain_realization":"The reviewer can return to the parent conclusion while retaining filters, visited-state indicators, and unresolved flags."},{"archetype_element":"Full Access Escape Hatch","domain_realization":"A full-workpaper view preserves unrestricted access for experts, unusual cases, quality review, and inspection."},{"archetype_element":"Progressive Validation Prompt","domain_realization":"Before sign-off, the view requires the reviewer to acknowledge surfaced exceptions and identify the evidence layer supporting the conclusion without deciding the conclusion for them."}],"mechanism_mapping":[{"mechanism_slug":"summary_detail_view","role":"Presents the audit conclusion and mandatory risk cues first while keeping procedure and evidence detail explicitly reachable.","counterfactual_removal":"Without the summary-detail relationship, reviewers again face a flat package or an unsupported summary detached from evidence."},{"mechanism_slug":"drill_down_dashboard","role":"Provides a stable route from assertion-level status through procedures and samples to source documents.","counterfactual_removal":"Without drilldown, the first layer becomes a terminal abstraction and reviewers must independently reconstruct evidence lineage."},{"mechanism_slug":"tiered_decision_support","role":"Stages conclusion, rationale, assumptions, uncertainty, exceptions, and underlying evidence according to the review decision.","counterfactual_removal":"Without tiers tied to the decision, information may be grouped aesthetically but still arrive at the wrong review moment."},{"mechanism_slug":"contextual_warning","role":"Overrides ordinary staging when an exception, contradiction, missing item, or other defined risk condition could change the review action.","counterfactual_removal":"Without contextual warnings, progressive disclosure could conceal precisely the details that professional skepticism requires."}],"causal_chain":["A flat workpaper makes routine evidence, decisive judgment, and rare exceptions compete simultaneously for reviewer attention.","The intervention separates content by review depth while preserving explicit evidence lineage.","The first layer orients the reviewer to the assertion, conclusion, and conditions capable of changing that conclusion.","User, stage, and risk triggers reveal the procedure or item-level information when it becomes relevant; critical exceptions bypass deferral.","Breadcrumbs, return paths, and a full-view escape hatch preserve context and reviewer autonomy across depths.","A sign-off prompt makes engagement with surfaced exceptions and supporting evidence observable without automating professional judgment.","If the layer boundaries and triggers match actual review reasoning, reviewers should be able to locate consequential evidence with less irrelevant scanning while retaining complete access."],"baseline":"The same audit evidence is reviewed through a conventional flat electronic workpaper: indexed sections, attachments, checklist fields, notes, and sign-offs are available together, with navigation driven mainly by document order, search, and reviewer experience.","nearest_rivals":["Risk-based audit planning, which changes where audit effort is allocated but does not necessarily stage the information inside a selected workpaper.","Exception-only reporting, which foregrounds deviations but can omit the ordinary evidence and context needed to evaluate whether an exception is material or resolved.","Standardized workpaper indexing and cross-referencing, which improves organization and traceability without governing when each depth appears during review.","A conventional audit-status dashboard, which aggregates engagement progress but may not connect conclusions through procedures to item-level evidence.","Mandatory review checklists, which require procedural confirmations but can add another flat layer of prompts rather than timing detail to the reviewer's reasoning path."],"remaining_contrastive_claim":"The candidate's distinguishing, testable feature is not summarization alone: it combines a smallest-safe conclusion layer, explicit evidence drilldown, risk-triggered mandatory disclosure, preserved orientation, and unrestricted full access within a single review path. Any benefit should therefore depend on the timing and legibility of those transitions rather than merely on a cleaner visual layout.","authority_safety":{"decision_authority":"The engagement partner retains authority over audit judgments and the audit opinion; designated reviewers retain responsibility for their sign-offs. The interface may route attention and record navigation but may not determine sufficiency, materiality, exception disposition, or the final conclusion.","authorized_first_step":"Build a read-only prototype from one completed, de-identified workpaper and evaluate it retrospectively without changing the official audit file, review record, sampling, conclusion, or sign-off.","excluded_actions":["Automatically clearing or resolving an exception","Automatically signing or approving a workpaper","Suppressing contradictory evidence or mandatory warnings","Changing materiality, sampling, procedure scope, or audit conclusions","Replacing the authoritative workpaper or altering its audit trail","Restricting a reviewer from opening the complete underlying record","Using reviewer navigation data for personnel evaluation during the initial test"],"halt_rollback":"Stop the test if any required evidence becomes inaccessible, an exception fails to surface before the affected sign-off task, source-to-conclusion lineage differs from the authoritative file, or participants mistake the prototype for the official record. Revert immediately to the unchanged flat workpaper; the prototype and its test logs can be discarded without modifying engagement records."},"negative_tests":{"strongest_counterevidence":"In matched review tasks, the flat workpaper enables reviewers to identify consequential exceptions and trace conclusions at least as accurately and efficiently as the layered view, while the layered view produces anchoring on preparer summaries or fewer inspections of contradictory evidence.","problem_falsifier":"Direct observation shows that reviewers already reach relevant evidence without excessive reconstruction or scanning, and review failures instead arise from deficient procedures, unavailable evidence, inadequate expertise, time-budget policy, or incentives rather than mistimed information presentation.","intervention_falsifier":"Reviewers cannot reliably discover deeper detail, critical exceptions remain hidden or are noticed later, evidence lineage becomes harder to reconstruct, or the layered presentation changes judgments merely through summary framing rather than improving access to the underlying record.","risks":["Anchoring on the preparer's first-layer conclusion","Incorrect triggers concealing an unclassified but consequential condition","False reassurance from clean summary indicators","Stale or inconsistent summary and detail layers","Loss of context when evidence supports several assertions or procedures","Flag inflation causing reviewers to ignore warnings","Gaming of classifications to keep exceptions out of the first layer","Reduced professional skepticism or erosion of reviewer skill","Navigation logs being repurposed for inappropriate surveillance","Added maintenance burden and ambiguity over which view is authoritative"]},"next_evidence_step":"Conduct one bounded, retrospective crossover test with six to eight audit reviewers using two de-identified, completed workpapers. Give each reviewer matched tasks in the flat baseline and the read-only layered prototype: identify a planted or historically documented exception, trace a conclusion to source evidence, locate contradictory evidence, and state what remains unresolved. Record task accuracy, time, unnecessary document openings, drilldown path, warning recognition, and confidence calibration. Predefine failure as any missed mandatory warning, inaccessible evidence, or lower exception-detection accuracy in the layered condition; do not use results to change an audit conclusion.","prior_art_status":"UNSEARCHED","diversity_from_prior_proposals":"Runtime-isolated one-shot generation; no other proposals or experiment candidates were inspected, so diversity relative to them is not asserted.","revision_record":{"parent_version":null,"progress_targets_addressed":["Initial one-shot construction from the supplied progressive-disclosure archetype and accounting-and-auditing domain card"],"conceptual_changes":[],"operational_changes":[],"evidence_changes":[],"claim_changes":[]}}