{"schema_version":1,"research_id":"eoa_inverse_innovation_exp04_external_evaluation_20260802","source_assessment_id":"invariant_mode_decomposition_design__political_science:RETRIEVAL_FIRST:v0","cell_id":"invariant_mode_decomposition_design__political_science","search_queries":["site:oversight.gov Commerce OIG H referrals processes managing referrals report","site:oig.doc.gov audits evaluations FAQ annual risk assessment audit selection","network interventions public administration policy Siciliano Whetsell pdf","Borgatti key player problem network node removal paper","site:gao.gov Yellow Book audit planning risk assessment sufficient appropriate evidence data reliability 2024","site:ignet.gov quality standards inspection evaluation evidence data privacy 2020","site:gao.gov inspector general audit planning risk assessment data analytics","site:oecd.org public sector audit collaboration Brazil risk matrix audit selection information network","\"The Network Organisation of Consumer Complaints\" resolution complaint network","site:arxiv.org complaint network resolution Department of Justice Brazil 1008.1815","government ombudsman repeated referrals complaint delay offices report","hotline complaints referrals delays inspector general audit report status tracking","site:uscode.house.gov Inspector General Act access records audit authority 5 USC 406","site:justice.gov Privacy Act internal agency records routine use data analysis official guidance","site:whitehouse.gov OMB federal AI governance 2025 high impact decisions audits investigations","site:nist.gov AI Risk Management Framework human oversight validity reliability official","site:opm.gov 2026 general schedule pay table data scientist auditor Washington DC","site:bls.gov occupational employment wages data scientists 2025 median annual wage","AWS pricing EC2 official on demand 2026 compute S3 pricing","Azure machine learning pricing official compute storage","complaint management software referral analytics government oversight official product","case management software inspector general hotline referrals tracking analytics","government complaint referral case management workflow analytics product","audit planning software risk scoring public sector official product"],"sources":[{"source_id":"S1","title":"The Department Must Improve Processes for Addressing and Managing ‘H’ Referrals","publisher":"U.S. Department of Commerce Office of Inspector General","url":"https://www.oig.doc.gov/OIGPublications/OIG-22-014-I.pdf","source_class":"GOVERNMENT_OR_REGULATOR","publication_date":"2021-12-16","accessed_at":"2026-08-02","claims_supported":["Commerce lacked effective policies and internal controls for hotline referrals.","Approximately 29 percent of reviewed referrals were outstanding for more than one year despite a general 60-day response requirement.","The OIG recommended department-wide procedures, quality assurance, status tracking, and deadline enforcement."]},{"source_id":"S2","title":"Ombudsman’s approach to official information delay complaints","publisher":"Office of the Ombudsman, New Zealand","url":"https://www.ombudsman.parliament.nz/resources/ombudsmans-approach-official-information-delay-complaints","source_class":"OFFICIAL_GUIDANCE","publication_date":"2025-07-31","accessed_at":"2026-08-02","claims_supported":["Delay complaints have consistently comprised roughly 20–25 percent of official-information complaints.","The Ombudsman explicitly logs and tracks complaints to identify repeated noncompliance and systemic issues.","The office concluded that formal investigation of every case alone was not producing meaningful improvement."]},{"source_id":"S3","title":"Ombudsman's office publishes OIA and LGOIMA complaints information for July to December 2025","publisher":"Office of the Ombudsman, New Zealand","url":"https://www.ombudsman.parliament.nz/news/ombudsmans-office-publishes-oia-and-lgoima-complaints-information-july-december-2025","source_class":"OFFICIAL_ORGANIZATION_DATA","publication_date":"2026-03-12","accessed_at":"2026-08-02","claims_supported":["The office received 1,198 OIA complaints in the half-year, including 280 delay complaints.","Complaint volume alone does not permit direct comparison among agencies because exposure and context differ."]},{"source_id":"S4","title":"Audits and Evaluations FAQs","publisher":"U.S. Department of Commerce Office of Inspector General","url":"https://www.oig.doc.gov/about/faqs/faqs-audits-evaluations/","source_class":"OFFICIAL_GUIDANCE","publication_date":"","accessed_at":"2026-08-02","claims_supported":["Commerce OIG uses annual risk assessment and management-approved planning to select audits and evaluations.","Selection also considers legal mandates, requests, priorities, prior findings, and program risk.","Prospective analysis is within the stated scope of OIG reviews."]},{"source_id":"S5","title":"Quality Standards for Inspection and Evaluation","publisher":"Council of the Inspectors General on Integrity and Efficiency","url":"https://www.ignet.gov/sites/default/files/files/QualityStandardsforInspectionandEvaluation-2020.pdf","source_class":"STANDARD","publication_date":"2020-12","accessed_at":"2026-08-02","claims_supported":["Inspection methods must obtain sufficient and appropriate evidence and document limitations.","Sensitive and confidential information must be protected during planning and execution.","Recommendations must be directed to officials with authority to act."]},{"source_id":"S6","title":"Government Auditing Standards: 2024 Revision","publisher":"U.S. Government Accountability Office","url":"https://www.gao.gov/products/gao-24-106786","source_class":"STANDARD","publication_date":"2024-02-01","accessed_at":"2026-08-02","claims_supported":["Government auditing standards emphasize reliable, objective, high-quality engagements.","The revision requires risk-based quality management, monitoring, and leadership responsibility."]},{"source_id":"S7","title":"Network Interventions: Applying Network Science for Pragmatic Action in Public Administration and Policy","publisher":"arXiv; subsequently Perspectives on Public Management and Governance","url":"https://arxiv.org/abs/2109.08197","source_class":"PRIMARY_RESEARCH","publication_date":"2021-09-16","accessed_at":"2026-08-02","claims_supported":["Network interventions purposefully use network data to identify strategies for improving performance or outcomes.","The framework explicitly extends network intervention to public-sector interorganizational and governance networks."]},{"source_id":"S8","title":"The Key Player Problem","publisher":"SSRN","url":"https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1149843","source_class":"PRIMARY_RESEARCH","publication_date":"2008-06-23","accessed_at":"2026-08-02","claims_supported":["The key-player problem selects nodes according to their counterfactual effect on network communication or reach.","It is a close precedent for choosing intervention targets by network effect rather than ordinary centrality."]},{"source_id":"S9","title":"The network organisation of consumer complaints","publisher":"arXiv","url":"https://arxiv.org/abs/1008.1815","source_class":"PRIMARY_RESEARCH","publication_date":"2010-08-10","accessed_at":"2026-08-02","claims_supported":["Administrative complaint records have previously been represented as weighted networks.","The study related complaint-network features to solved and unsolved complaints and used PCA for dimensional reduction.","Its network connects companies to complaint categories, not responsibility transfers among offices."]},{"source_id":"S10","title":"Auditing Decentralised Policies in Brazil: Collaborative and Evidence-Based Approaches for Better Outcomes","publisher":"OECD","url":"https://www.oecd.org/en/publications/auditing-decentralised-policies-in-brazil_30023307-en/full-report/towards-evidence-based-audit-selection-in-brazil-s-decentralised-policy-areas_6634b1dc.html","source_class":"OFFICIAL_GUIDANCE","publication_date":"2020-11-26","accessed_at":"2026-08-02","claims_supported":["Risk-based audit selection is established public-sector practice used to allocate limited audit capacity.","Evidence, local indicators, collaboration, and professional judgment are used to select consequential audits.","Audit-selection methods should remain objective, transparent, mandate-consistent, and auditable."]},{"source_id":"S11","title":"Privacy Act of 1974","publisher":"U.S. Department of Justice, Office of Privacy and Civil Liberties","url":"https://www.justice.gov/opcl/privacy-act-1974?os=f","source_class":"OFFICIAL_GUIDANCE","publication_date":"2026-06-25","accessed_at":"2026-08-02","claims_supported":["The Privacy Act governs federal collection, maintenance, use, and dissemination of personally identifiable records in systems of records.","Disclosure generally requires consent or a statutory exception, and agencies must publish system-of-records notices."]},{"source_id":"S12","title":"Artificial Intelligence Risk Management Framework (AI RMF 1.0)","publisher":"National Institute of Standards and Technology","url":"https://www.nist.gov/publications/artificial-intelligence-risk-management-framework-ai-rmf-10","source_class":"STANDARD","publication_date":"2023-01-26","accessed_at":"2026-08-02","claims_supported":["The voluntary framework calls for governance, validation, monitoring, transparency, privacy protection, and documented human oversight of data-driven systems.","Intended use and limits should be defined across design, testing, deployment, and monitoring."]},{"source_id":"S13","title":"New York State Offices of the Inspector General Case Management System Request for Proposals","publisher":"New York State Offices of the Inspector General","url":"https://ig.ny.gov/system/files/documents/2025/03/nys-offices-of-the-inspector-general_cms_rfp_03-24-2025.pdf","source_class":"OFFICIAL_ORGANIZATION_DATA","publication_date":"2025-03-24","accessed_at":"2026-08-02","claims_supported":["An identifiable OIG sought a centralized case-management system with workflow-stage tracking, bottleneck visibility, analytics, and predictive capabilities.","The requirements include historical field versioning, data export, customizable workflows, role-based access, encryption, audit trails, dashboards, outcome forecasting, and secure integration.","This procurement substantially overlaps the proposed data, workflow, analytics, security, and adopter setting, although it does not specify spectral-mode perturbation ranking."]},{"source_id":"S14","title":"eCASE Investigations","publisher":"Casepoint","url":"https://www.casepoint.com/products/government/ecase-investigations/","source_class":"COMMERCIAL_FIRST_PARTY","publication_date":"","accessed_at":"2026-08-02","claims_supported":["The vendor offers OIG-specific complaint and investigation management from hotline intake through referral and closure.","The product states that it provides workflow tracking, automated assignment, triage, access controls, audit trails, reporting, and referral/outcome fields.","The page does not claim responsibility-transfer eigenmode analysis or prospective audit-target validation."]},{"source_id":"S15","title":"Salary Table 2026-DCB","publisher":"U.S. Office of Personnel Management","url":"https://www.opm.gov/policy-data-oversight/pay-leave/salaries-wages/salary-tables/26Tables/html/DCB.aspx","source_class":"OFFICIAL_ORGANIZATION_DATA","publication_date":"2026-01","accessed_at":"2026-08-02","claims_supported":["The 2026 Washington-area annual salary range is $102,415–$133,142 for GS-12, $121,785–$158,322 for GS-13, and $143,913–$187,093 for GS-14.","These rates anchor resource-equivalent labor estimates before benefits, overhead, contracting premiums, and security costs."]},{"source_id":"S16","title":"Amazon EC2 On-Demand Instance Pricing","publisher":"Amazon Web Services","url":"https://aws.amazon.com/ec2/pricing/on-demand/","source_class":"COMMERCIAL_FIRST_PARTY","publication_date":"","accessed_at":"2026-08-02","claims_supported":["General-purpose compute can be purchased by the second or hour without long-term commitments.","Storage, monitoring, data transfer, operating-system licensing, and government-compliance environments can add separate costs.","Compute expense is unlikely to dominate a bounded sparse-network feasibility study relative to labor and data governance."]}],"problem_evidence":{"support":"MODERATE","rationale":"Official evidence strongly verifies delayed, decentralized, inconsistently controlled complaint referrals and their accountability consequences: Commerce reported that about 29% of H referrals were over one year old, while New Zealand reports persistent delay complaints and explicitly tracks repeated noncompliance. However, no opened source verifies the candidate's more specific premise that audit planners rely excessively on complaint volume, that cases repeatedly circulate among mutually reinforcing groups of offices, or that a stable dominant transfer mode is being missed. The general problem exists and matters; the proposed modal diagnosis remains hypothetical.","source_ids":["S1","S2","S3"]},"stakeholder_evidence":{"support":"STRONG","rationale":"Commerce OIG is an identifiable audit authorizer using annual risk assessment and called for referral tracking and deadline enforcement. New York State OIG issued an RFP explicitly seeking bottleneck visibility, workflow analytics, predictive outcomes, security, and data-driven decision support. These are concrete expressions of need and purchasing/authorizing capacity, though neither organization has endorsed this spectral method.","source_ids":["S1","S4","S13"]},"prior_art":{"proximity":"SUBSTANTIAL_COLLISION","closest_analogues":[{"name":"OIG case-management modernization and predictive analytics","similarity":"New York State OIG requires lifecycle tracking, bottleneck analysis, predictive outcomes, dashboards, historical data, access control, and workflow optimization in the exact institutional setting.","remaining_difference":"The RFP does not specify an interoffice referral operator, leading eigenmode, marginal spectral perturbation ranking, or prospective comparison with complaint volume and existing risk scores.","source_ids":["S13"]},{"name":"eCASE Investigations","similarity":"An available OIG product already manages hotline intake, referrals, outcomes, assignments, triage, audit trails, and reports in an access-controlled system.","remaining_difference":"The first-party materials do not claim modal network analysis or holdout-validated audit-target ranking.","source_ids":["S14"]},{"name":"Key Player Problem","similarity":"It already selects nodes by their counterfactual network effect rather than relying on ordinary centrality, closely anticipating the proposed sensitivity sweep.","remaining_difference":"Its canonical objectives concern network disruption or reach, not a bounded change in a responsibility-transfer eigenmode predicting unresolved cases.","source_ids":["S8"]},{"name":"Public-sector network interventions","similarity":"The literature already proposes purposeful manipulation or targeting of nodes and structures in public-sector governance networks to improve outcomes.","remaining_difference":"It does not provide the candidate's particular referral-record construction, spectral outcome, audit workflow, or temporal validation design.","source_ids":["S7"]},{"name":"Complaint-network analysis and evidence-based audit selection","similarity":"Prior research already turns administrative complaints into networks and reduced-dimensional representations, while official audit practice already uses evidence-based risk indicators to allocate scarce audit capacity.","remaining_difference":"The opened work does not combine interoffice transfer edges, leading-mode sensitivity, realistic edge perturbations, and prospective unresolved-case prediction.","source_ids":["S9","S10"]}],"distinctive_claim_remaining":"At fixed audit capacity and using a preregistered longitudinal interoffice-referral operator that passes gap, drift, and residual checks, the top-k offices ranked by the marginal reduction in the leading transfer mode under the same feasible perturbation will predict more next-period unresolved or repeatedly transferred cases than top-k rankings from complaint volume, weighted degree, generic eigenvector/PageRank centrality, the incumbent risk score, and a conventional supervised tabular model. This is a predictive, not causal or misconduct, claim.","confidence":"HIGH"},"implementation_evidence":{"support":"MODERATE","rationale":"Existing OIG systems and procurements demonstrate technical feasibility for structured intake, referral, workflow-stage, time, outcome, historical, export, and access-control data. Sparse spectral computation and temporal evaluation are modest compared with those systems. The major unresolved issue is data fitness: public sources do not show that any candidate partner preserves every interoffice transfer as a directed event across multiple periods, rather than only an initial OIG-to-bureau referral and final status. A star-like or heavily censored workflow would not support the proposed modal interpretation. Privacy review, records controls, realistic perturbation definitions, and human-only use are necessary but appear manageable for a nondeployment study.","source_ids":["S1","S5","S6","S11","S12","S13","S14","S16"]},"scores":{"meaningful_impact":{"score":3,"rationale":"Referral delay and weak control can impair accountability, but the shortlist's incremental effect on resolution or audit impact is unmeasured.","source_ids":["S1","S2"]},"stakeholder_pull":{"score":4,"rationale":"Official recommendations and an active OIG procurement show strong demand for tracking, bottleneck analytics, and predictive workflow support, though not specifically for spectral ranking.","source_ids":["S1","S4","S13"]},"incremental_advantage":{"score":2,"rationale":"No performance evidence shows improvement over volume, conventional risk scoring, generic network measures, key-player objectives, or supervised predictive analytics already contemplated by OIG purchasers.","source_ids":["S8","S10","S13"]},"distinctiveness_plausibility":{"score":2,"rationale":"The exact integration was not found, but nearly every component is established and the remaining difference is a narrow model-and-validation claim.","source_ids":["S7","S8","S9","S10","S13","S14"]},"technical_implementability":{"score":3,"rationale":"The computation is feasible and relevant systems can retain structured workflow data, but a usable multi-step referral graph, stable mode, and defensible perturbation map have not been demonstrated.","source_ids":["S1","S13","S14","S16"]},"adoption_authority_feasibility":{"score":4,"rationale":"OIG management already controls risk assessment, audit planning, and case-system acquisition, so an advisory internal analysis has a credible authority path.","source_ids":["S4","S5","S13"]},"evidence_readiness":{"score":2,"rationale":"Public evidence supports the surrounding problem and workflow but cannot answer the incremental claim; access-controlled longitudinal records and partner participation are essential.","source_ids":["S1","S13"]},"safety_net_benefit":{"score":3,"rationale":"Human-only advice, withheld identifiable rankings, validation gates, and rollback to incumbent planning substantially limit harm, but ranking-induced scrutiny and privacy risks remain.","source_ids":["S5","S11","S12"]},"scalability":{"score":3,"rationale":"Sparse-network computation is inexpensive and case systems can export structured data, but coding conventions, mandates, referral topology, and model meaning are likely to vary materially across oversight systems.","source_ids":["S13","S14","S16"]}},"score_confidence":"MODERATE","costs":{"first_evidence":{"band_2026_usd":"50K_TO_250K","scope":"A 12–16 week, single-system retrospective feasibility and temporal-holdout study, including data extraction, privacy/security review, network construction, preregistration, baseline models, sensitivity checks, and an internal report.","confidence":"MODERATE","assumptions":["Approximately 0.5–0.8 FTE data scientist, 0.2–0.4 FTE audit-domain analyst, 0.1–0.2 FTE data engineer, and limited privacy/security counsel.","The partner already has exportable structured records; this excludes digitizing paper files or procuring a replacement case-management system.","CPU-based sparse computation is sufficient and no live operational integration occurs.","Resource equivalents include benefits/overhead or contractor premiums beyond published base salaries."],"source_ids":["S13","S15","S16"]},"initial_deployment_startup":{"band_2026_usd":"250K_TO_1M","scope":"Hardened internal pipeline integrated with one case-management system, repeatable data-quality tests, access controls, model/version registry, documentation, security assessment, audit-planning interface, governance approval, and staff training.","confidence":"LOW","assumptions":["One agency and one existing case-management platform are in scope.","No wholesale CMS replacement is required.","The model remains advisory and does not automatically initiate audits.","Security authorization, integration complexity, and legacy-data remediation dominate uncertainty."],"source_ids":["S11","S12","S13","S14","S15"]},"operational_launch":{"band_2026_usd":"50K_TO_250K","scope":"One shadow-mode annual or quarterly planning cycle followed by controlled human-review rollout, including parallel incumbent scoring, analyst training, review meetings, incident procedures, and independent validation sign-off.","confidence":"LOW","assumptions":["Startup integration is complete.","Launch covers one oversight system and does not alter live referral routing.","Existing audit-planning staff absorb part of the review workload.","No public identifiable rankings are released."],"source_ids":["S4","S5","S6","S13","S15"]},"annual_recurring":{"band_2026_usd":"50K_TO_250K","scope":"Periodic refreshes, drift/gap/residual monitoring, data-quality remediation, privacy and quality review, annual revalidation, documentation, and analyst support for one system.","confidence":"MODERATE","assumptions":["Roughly 0.4–1.0 combined FTE across analytics, audit, data stewardship, and governance.","Compute remains CPU-based and pay-as-you-go.","Major schema changes, new agencies, litigation, and case-management replacement are excluded."],"source_ids":["S5","S6","S12","S15","S16"]}},"verified_pipeline_gates":{"externally_supported_problem":{"status":"YES","reason":"Official investigations and complaint data verify consequential delays, weak referral controls, and persistent systemic complaint-processing issues, although not the hypothesized dominant mode.","source_ids":["S1","S2","S3"]},"externally_credible_adopter_or_authorizer":{"status":"YES","reason":"Commerce OIG controls audit planning, and New York State OIG has formally sought case-workflow bottleneck and predictive analytics capabilities.","source_ids":["S4","S13"]},"distinct_testable_incremental_claim":{"status":"YES","reason":"The claim specifies a top-k modal-sensitivity ranking, future outcomes, temporal holdout, and named volume, centrality, incumbent-risk, and supervised-model comparators.","source_ids":["S7","S8","S10","S13"]},"bounded_next_evidence_step":{"status":"YES","reason":"A single-system, retrospective, nondeployment temporal-holdout study can be preregistered with explicit data-fitness, performance, stability, and safety falsifiers.","source_ids":["S5","S6","S13"]},"no_unresolved_safety_or_authority_stop":{"status":"UNCERTAIN","reason":"No categorical stop was identified, and OIG management can authorize an internal study, but actual data access, Privacy Act/system-of-records compatibility, records-security approval, and affected-office governance have not been obtained.","source_ids":["S4","S5","S11","S12","S13"]},"credible_cost_scope_and_range":{"status":"YES","reason":"The four ranges are scoped to a one-system advisory implementation and anchored to 2026 federal labor rates, existing OIG system requirements, and metered compute; integration and authorization uncertainty is explicitly reflected.","source_ids":["S13","S15","S16"]}},"next_evidence_step":"Secure one consenting oversight partner and run a 12–16 week retrospective study without producing an operational shortlist. First, use a two-week data-fitness gate: verify event-level sender, receiver, timestamp, status, closure, repeat-transfer, and stable office identifiers across at least six reporting periods; report missingness, coding changes, graph density, strongly connected structure, and repeated-transfer prevalence. Stop if multi-step transfers are rare, largely unrecorded, or yield only a star-like referral graph. If data pass, preregister the operator, closure treatment, perturbation size, top-k audit capacity, temporal split, spectral-gap/drift/residual thresholds, and outcomes of unresolved status at 60/180 days and repeat transfer. Compare the modal-sensitivity ranking with complaint volume, weighted in/out degree, PageRank or ordinary eigenvector centrality, the incumbent risk score, a conventional supervised tabular/time-to-event model, and a key-player-style perturbation objective. Primary evidence is the next untouched period's precision@k or captured adverse outcomes at the same k, with bootstrap confidence intervals; secondary evidence covers calibration, rank stability across reasonable edge definitions, subgroup/office-type error patterns, structured residuals, and analyst usability. Falsify the claim if the lower confidence bound on improvement over the strongest comparator is not positive, if rankings are unstable under reasonable specifications, if the leading mode lacks preregistered separation or drifts materially, if residuals retain consequential pathways, or if legal/privacy review cannot authorize the data use. Advancement would require replication in a second institution before live planning use.","blocking_evidence":["No opened source shows that oversight bodies generally over-prioritize complaint volume or that incumbent risk scores systematically miss responsibility-transfer structure.","No public data dictionary confirms that a candidate oversight system records every interoffice transfer, repeat transfer, closure, and timestamp needed for a longitudinal directed operator.","The Commerce example may be primarily an OIG-to-bureau referral process rather than a recirculating network; this could eliminate the proposed modal mechanism.","No retrospective or prospective result demonstrates incremental prediction over existing risk scores, ordinary network measures, key-player methods, or conventional supervised analytics.","No partner has authorized access, confirmed Privacy Act/system-of-records compatibility, or approved office-level ranking and retention controls.","The policy-plausible perturbation is not yet operationally defined; changing closure or onward-transfer rates in a matrix may not correspond to an achievable audit or reporting-rule action.","Possible disparate scrutiny of specialized, high-workload, or legally mandated routing offices has not been measured.","Existing OIG case-management procurements already demand bottleneck and predictive analytics, so differentiation from configurable incumbent BI and forecasting remains unresolved."],"research_disposition":"PARTNERED_RESEARCH_PROGRAM","world_novelty_boundary":"A bounded ordinary-web search through 2026-08-02 found established complaint-network analysis, public-sector network interventions, key-player perturbation methods, evidence-based audit selection, OIG case-management products, and an official OIG procurement requiring bottleneck and predictive analytics. It did not find an opened source implementing the complete longitudinal interoffice-referral eigenmode, feasible marginal perturbation, and prospective top-k audit-target validation pipeline. Any distinctiveness is therefore limited to that narrow integration and its untested incremental performance claim. World novelty, patentability, freedom to operate, market size, and realized impact were not measured.","arm":"RETRIEVAL_FIRST","candidate_version":0,"controller_recommendation":{"action":"STOP_EMPIRICAL_RESEARCH_NEEDED","repairable":true,"material_progress_observed":true,"progress_targets":["Obtain a partner-approved data dictionary and quantify whether multi-step, repeated, and recirculating interoffice transfers are sufficiently prevalent and persistent for a modal operator.","Preregister operator construction, realistic perturbations, top-k capacity, outcomes, temporal split, spectral validity thresholds, residual tests, and exact falsification criteria.","Demonstrate positive temporal-holdout improvement over complaint volume, degree, PageRank/eigenvector centrality, incumbent risk scoring, a conventional supervised model, and a key-player-style objective.","Secure written privacy, records, security, and decision-authority approvals for the internal nondeployment study and document human-review and rollback controls.","Differentiate the proposed ranking from predictive analytics and bottleneck functions already available or sought in OIG case-management systems.","Measure ranking stability and disparate scrutiny across office types, lawful specialization, workload, and mandatory-routing roles before any operational use."],"reason":"Web evidence verifies the surrounding referral-delay problem, credible OIG adopters, substantial adjacent practice, and a falsifiable residual claim, but the decisive evidence requires proprietary longitudinal referral events and live partner validation. Under the evaluation rule, that requires STOP_EMPIRICAL_RESEARCH_NEEDED rather than further bounded web research."}}