{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp11_mechanism_context_external20_20260804","research_id":"eoa_inverse_innovation_exp11_external_scrutiny_20260804","cell_id":"invariant_mode_decomposition_design__organizational_management","opaque_id":"invariant_mode_decomposition_design__organizational_management__C","search_lanes":{"direct_problem":{"queries":["organizations delivery separate KPI thresholds miss correlated backlog rework staffing handoff delay incidents","project delivery leading indicators rework staffing handoffs incidents correlated metrics","software delivery metrics combinations predict performance DORA SPACE framework","multivariate statistical process control project management schedule cost performance early warning"],"source_ids":["SRC1","SRC2","SRC7"],"no_result_note":"The sources support correlated delivery measures and limitations of separate monitoring, but no retained source directly demonstrates the proposal's exact precursor: individually acceptable backlog, rework, load, handoff, incident, and throughput measures forming a repeatable growing mode before cross-functional failure."},"closest_prior_art":{"queries":["dynamic mode decomposition organizational performance management","dynamic principal component analysis process monitoring multivariate","predictive process monitoring business process remaining time outcome risk official paper","Hotelling T2 project performance monitoring multiple correlated indicators project management"],"source_ids":["SRC1","SRC3","SRC4","SRC5","SRC6","SRC7"],"no_result_note":"No direct organizational-delivery implementation combining transition eigenmodes, modal gains, drift and residual checks, and an intervention-effect map was found. Close pieces exist separately in project control, predictive process monitoring, dynamic statistical process monitoring, and DMD with control."},"historical_terminology":{"queries":["statistical project control tool engineering managers Shewhart 2001","project management early warning indicators statistical process control","dynamic principal component analysis statistical process monitoring 2000","project management rework cycle system dynamics delivery delay staffing workload"],"source_ids":["SRC2","SRC4","SRC6"],"no_result_note":null},"products_practices_standards":{"queries":["DORA software delivery performance metrics throughput instability rework","official workplace monitoring guidance automated decision making worker data","statistical project control tool engineering managers PMI","multivariate project control earned value duration cost"],"source_ids":["SRC1","SRC2","SRC7","SRC8"],"no_result_note":"DORA and project-control practice already reject purely isolated single-metric interpretation, weakening the proposal's simple baseline description, but neither retained source uses transition-mode gains to select interventions."},"non_english_regional":{"queries":["\"alerta temprana\" \"gestión de proyectos\" indicadores multivariantes retrasos retrabajo","\"Frühwarnsystem\" Projektmanagement Kennzahlen multivariat Verzögerung Nacharbeit","项目管理 多变量 预警 返工 延误 绩效 指标","基于动态主元分析 统计过程监视 多变量"],"source_ids":["SRC6"],"no_result_note":"Chinese searches recovered older dynamic-PCA process-monitoring terminology. Spanish, German, and Chinese project-management searches found early-warning and weighted-indicator approaches but no closer organizational transition-mode package among the eight retained sources."},"composition_subproblems":{"queries":["multivariate control chart project duration cost correlated indicators false alarms","predictive process monitoring remedial action resource shifting deadline","dynamic multivariate process control subspace identification state variables anomaly diagnosis","dynamic mode decomposition with control state snapshots intervention input matrix"],"source_ids":["SRC1","SRC3","SRC4","SRC5","SRC6"],"no_result_note":null}},"sources":[{"source_id":"SRC1","title":"Multivariate statistical control chart and process capability indices for simultaneous monitoring of project duration and cost","url":"https://doi.org/10.1016/j.cie.2019.03.021","publisher":"Elsevier, Computers & Industrial Engineering","date_or_year":"2019","source_type":"PRIMARY_RESEARCH","language":"English","claims_supported":["Project duration and cost indicators can be correlated, making separate univariate control charts unsatisfactory and increasing false alarms.","A Hotelling T-squared chart can continuously monitor correlated project-performance indicators as an integrated statistic.","The method is static multivariate project control rather than transition-mode estimation or mode-guided intervention."]},{"source_id":"SRC2","title":"A Statistical Project Control Tool for Engineering Managers","url":"https://www.pmi.org/learning/library/statistical-project-control-tool-engineering-managers-5327","publisher":"Project Management Institute / Project Management Journal","date_or_year":"2001","source_type":"PRIMARY_RESEARCH","language":"English","claims_supported":["Engineering project managers are identifiable adopters of statistical monitoring tools.","Modified Shewhart charts have been developed to monitor time, cost, and technical project parameters.","Statistical alerts are intended to support corrective action before project failure, establishing older adjacent practice."]},{"source_id":"SRC3","title":"Survey and Cross-benchmark Comparison of Remaining Time Prediction Methods in Business Process Monitoring","url":"https://doi.org/10.1145/3331449","publisher":"Association for Computing Machinery","date_or_year":"2019","source_type":"SECONDARY_RESEARCH","language":"English","claims_supported":["Predictive process monitoring uses historical execution logs to predict running-case outcomes, next activities, and remaining cycle time.","Predictions can support operational managers taking remedial actions, including resource shifts, before deadlines are missed.","A benchmark across 17 real-life datasets found material differences among methods and an accuracy-explainability tradeoff, supporting explicit rival comparison."]},{"source_id":"SRC4","title":"Dynamic multivariate statistical process control using subspace identification","url":"https://doi.org/10.1016/S0959-1524(03)00041-6","publisher":"Elsevier, Journal of Process Control","date_or_year":"2004","source_type":"PRIMARY_RESEARCH","language":"English","claims_supported":["Dynamic multivariate monitoring can model cross-correlated and autocorrelated process variables with a reduced state representation.","Static PCA can be inadequate when throughput changes, feedback, or disturbances create dynamic transients.","The method uses dynamic state-space identification, monitoring statistics, and contribution charts, closely overlapping the proposal's transition-model and residual-diagnostic components outside the organizational context."]},{"source_id":"SRC5","title":"Dynamic Mode Decomposition with Control","url":"https://arxiv.org/abs/1409.6358","publisher":"Joshua L. Proctor, Steven L. Brunton, and J. Nathan Kutz; arXiv","date_or_year":"2014","source_type":"PRIMARY_RESEARCH","language":"English","claims_supported":["DMD with control estimates low-order dynamic modes and separates underlying dynamics from actuation using state and control snapshots.","The method estimates both a state-transition map and an input-effect map, directly anticipating the proposal's generic mathematical lever.","The paper demonstrates the lever in dynamical and epidemiological examples, not organizational delivery management."]},{"source_id":"SRC6","title":"基于动态主元分析的统计过程监视 (Statistical Process Monitoring Based on Dynamic Principal Component Analysis)","url":"https://hgxb.cip.com.cn/CN/abstract/abstract9159.shtml","publisher":"化工学报 / Journal of Chemical Industry and Engineering (China)","date_or_year":"2000","source_type":"PRIMARY_RESEARCH","language":"Chinese with English abstract","claims_supported":["Older Chinese research explicitly treats serial dependence as a limitation of conventional multivariate PCA/PLS monitoring.","Dynamic PCA was proposed to extract disturbance-driving signals from time-correlated observations.","The application is simulated industrial-process monitoring, not organizational workflow or authorized management intervention."]},{"source_id":"SRC7","title":"DORA’s software delivery performance metrics","url":"https://dora.dev/guides/dora-metrics/","publisher":"Google Cloud DORA","date_or_year":"2026","source_type":"FIRST_PARTY_PRODUCT","language":"English","claims_supported":["Delivery teams already monitor multiple throughput and instability measures, including lead time, deployment frequency, recovery time, change failures, and deployment rework.","DORA reports that these delivery measures are correlated and can operate as leading and lagging indicators.","DORA warns against one-metric management, gaming, cross-context comparisons, and siloed ownership, so ordinary practice is not uniformly limited to independent threshold reviews."]},{"source_id":"SRC8","title":"Data protection and monitoring workers","url":"https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/employment/monitoring-workers/data-protection-and-monitoring-workers/","publisher":"UK Information Commissioner's Office","date_or_year":"2023 guidance, accessed 2026","source_type":"OFFICIAL_GUIDANCE","language":"English","claims_supported":["Systematic monitoring of worker groups and productivity analytics can fall within worker-monitoring data-protection duties.","Employers must identify a lawful basis, balance business interests against worker rights, use the least intrusive means, and consider a data-protection impact assessment.","Automated decision-making and third-party monitoring tools require additional scrutiny; employer responsibility is not displaced by using a vendor."]}],"problem_evidence":{"status":"PARTLY_SUPPORTED","finding":"The underlying problem is credible: project and delivery measures are correlated; univariate project-control methods can give unsatisfactory signals; delivery frameworks treat throughput, instability, rework, and delay jointly; and predictive process monitoring exists specifically to warn managers before adverse outcomes. However, the exact proposed empirical signature—six individually acceptable weekly aggregates forming a stable, growing coupled direction before cross-functional delivery failure—was not demonstrated by the retained evidence.","source_ids":["SRC1","SRC2","SRC3","SRC7"],"uncertainty":"Evidence spans construction/project control, business-process cases, and software delivery rather than one matched dataset. Correlation and multimetric usefulness do not establish a stable transition eigenmode, lead time, or causal precursor."},"adopter_evidence":{"status":"SUPPORTED","finding":"Engineering project managers, operational process managers, software-delivery leaders, and the accountable business-unit leader are identifiable adopters. The business-unit leader can authorize a shadow operational study, while data-protection, workforce-representation, and information-governance functions may be required co-authorizers.","source_ids":["SRC2","SRC3","SRC7","SRC8"],"uncertainty":"The precise required approvals vary by jurisdiction, collective agreement, data provenance, and whether nominally aggregated data can be re-identified."},"implementation_evidence":{"status":"PARTLY_SUPPORTED","finding":"All major technical pieces have precedents: correlated project indicators can be monitored jointly; predictive process models can be benchmarked on real logs; dynamic multivariate methods estimate reduced states; dynamic PCA handles serial dependence; and DMD with control estimates transition and intervention maps. No source validates their combined use on sparse weekly organizational aggregates or shows that mode-targeted management actions improve delivery without workload harm. Eight prospective weekly transitions alone would be inadequate to estimate and validate a six-variable transition model; fitting therefore requires a sufficiently long, regime-matched historical series or a lower-dimensional preregistered model before an eight-week shadow validation.","source_ids":["SRC1","SRC3","SRC4","SRC5","SRC6","SRC7"],"uncertainty":"Organizational dynamics are reflexive, interventions are confounded with managerial selection, weekly samples may be small and seasonal, and aggregate metrics can conceal unequal burdens."},"prior_art":{"disposition":"ADJACENT_PRIOR_ART","closest_analogues":[{"name":"Multivariate statistical project control","source_ids":["SRC1","SRC2"],"same_problem":true,"same_causal_lever":false,"overlap":"Monitors correlated project-performance measures over time, alerts managers to developing out-of-control conditions, and supports corrective action before failure.","remaining_difference":"Uses Shewhart or Hotelling statistics over project time/cost/technical indicators rather than estimating transition eigenmodes, modal gains, input effects, spectral separation, or mode drift."},{"name":"Predictive business-process monitoring","source_ids":["SRC3"],"same_problem":true,"same_causal_lever":false,"overlap":"Uses historical workflow data to predict delay or adverse outcomes during execution and enables operational managers to take remedial action.","remaining_difference":"Usually predicts case-level remaining time or outcomes with transition systems, regressors, or machine learning; it does not prioritize unit-level interventions by the gain of a coupled transition mode."},{"name":"Dynamic multivariate process monitoring and DMD with control","source_ids":["SRC4","SRC5","SRC6"],"same_problem":false,"same_causal_lever":true,"overlap":"Models serially and cross-correlated state transitions, extracts reduced dynamic directions, detects abnormal behavior, and can estimate how control inputs affect future state.","remaining_difference":"Validated mainly on physical or industrial processes, not reflexive organizations, and does not establish safe managerial use on aggregated backlog, rework, workload, handoff, incident, and throughput measures."},{"name":"DORA multimetric delivery management","source_ids":["SRC7"],"same_problem":true,"same_causal_lever":false,"overlap":"Treats delivery throughput and instability as correlated, shared team-level measures and warns against isolated or gameable metrics.","remaining_difference":"Tracks and interprets named delivery metrics rather than estimating locally invariant transition directions or selecting interventions from modal input effects."}],"contrastive_claim_remaining":"For one business unit with enough regime-matched historical weekly observations, a preregistered locally linear transition-mode model will identify at least one stable, separated, interpretable coupled precursor and produce shadow alerts with better out-of-sample lead time and precision than independent KPI thresholds, static PCA/Hotelling monitoring, and a conventional predictive-process baseline; authorized interventions projected to damp that mode will subsequently reduce its estimated gain without degrading delivery, workload, privacy, or subgroup outcomes.","contrastive_claim_falsifier":"The claim fails if no mode satisfies preregistered stability, spectral-separation, residual, and cross-window criteria; if independent KPIs, static multivariate monitoring, or conventional predictive models match or outperform alert lead time and precision; if estimated intervention effects are unstable or confounded; or if safe authorized controls cannot reduce the mode without workload or equity harm.","confidence":"MODERATE","search_limitations":"The search was deliberately capped at exactly eight retained direct sources. It covered English, Chinese, Spanish, and German terminology and older SPC, DPCA, project-control, process-mining, DORA, and worker-monitoring concepts. Two paywalled publisher pages were available primarily through detailed abstracts and bibliographic records. The search did not inspect proprietary organizational deployments, private datasets, every jurisdiction, patents comprehensively, or all possible synonyms."},"researchability_gates":{"externally_supported_problem":{"status":"PASS","rationale":"Independent project-control, process-monitoring, and software-delivery sources support correlated measures, limitations of isolated monitoring, and the value of anticipatory alerts. The exact modal precursor remains a hypothesis, but the underlying problem is externally grounded and falsifiable.","source_ids":["SRC1","SRC2","SRC3","SRC7"]},"identifiable_adopter_or_authorizer":{"status":"PASS","rationale":"Operational managers and engineering/project leaders are established users of related tools; an accountable business-unit leader is an identifiable authorizer, subject to data-governance and workforce/privacy review.","source_ids":["SRC2","SRC3","SRC7","SRC8"]},"distinct_testable_incremental_claim":{"status":"PASS","rationale":"No retained source combines organizational weekly aggregates with transition eigenmodes, modal-gain alerts, intervention-effect mapping, drift/residual controls, and workload safeguards. Comparative predictive and intervention claims can be preregistered and falsified.","source_ids":["SRC1","SRC3","SRC4","SRC5","SRC6","SRC7"]},"bounded_next_evidence_step":{"status":"PASS","rationale":"A bounded first step is a data-sufficiency and retrospective replay audit followed, only if estimability criteria pass, by an eight-week de-identified shadow comparison. This requires no staffing, evaluation, or automated management action.","source_ids":["SRC1","SRC3","SRC4","SRC5","SRC8"]},"no_unresolved_safety_or_authority_stop":{"status":"PASS","rationale":"Worker monitoring creates real privacy, proportionality, and governance obligations, but it is not an absolute stop. Team aggregation, purpose limitation, no personnel use, human review, a DPIA or equivalent assessment, worker consultation where applicable, and halt conditions bound the shadow study.","source_ids":["SRC7","SRC8"]},"adequate_search_evidence":{"status":"PASS","rationale":"All six required lanes were searched adversarially with direct, older, product/practice, regional non-English, and component-combination terminology. Exactly eight sources were retained, spanning seven publisher or author groups, with multiple primary, official, and first-party sources.","source_ids":["SRC1","SRC2","SRC3","SRC4","SRC5","SRC6","SRC7","SRC8"]}},"strict_success":true,"screen_survival":true,"remaining_research_value":"MODERATE","recommended_next_step":"Before the proposed shadow test, conduct a preregistered two-week feasibility audit of the available historical weekly series: verify aggregation and lawful provenance; require enough regime-matched observations for the chosen model dimension; check missingness, seasonality, structural breaks, re-identification risk, subgroup burden, and intervention timestamps; and freeze KPI, static PCA/Hotelling, and predictive-process baselines. If estimability and privacy criteria pass, fit only on historical data and run an eight-week prospective shadow evaluation of lead time, precision, calibration, mode stability, residuals, and workload signals. Do not use alerts for staffing or evaluation. A later randomized or staggered authorized process-adjustment study is needed for the intervention claim.","world_novelty_boundary":"This bounded search supports only an adjacent-prior-art assessment and a distinct comparative research hypothesis. It does not establish world novelty, patentability, freedom to operate, market size, routine feasibility across organizations, or realized impact."}