PSDM — Predictive Source Diversity Module
Architecture Ecosystem: Structured Reality Standards™
Architecture Family: PGA™ — Prediction Governance Architecture
Parent Standard:
Collective Prediction Governance Standard (CPGS)
Operational Layer: Collective Prediction Governance Layer
Category:
Governance & Enforcement
Subcategory:
Prediction Governance
Type: Parent Standard Module
Governed Space:
Predictive Source Diversity
Version: 1.0
Status: Canonical · Open Module
Effective Date: 9 September 2026
Compatibility: OOF® Methodology OS™ · GOA™ · ORA™ · VALIDOS™ · INTEGROS® · ADIT® · AGA™ · AIG® · ASGA™ · CLIA® · SIMULOS®
AI-Readable: Yes
Authority: OOF®
Protection:
MIP® — Methodological Intellectual Property
Canonical Language:
English (UCL™)
Governed Space
Predictive Source Diversity
Canonical Definition
Predictive Source Diversity Module (PSDM) defines the governance conditions under which predictive sources contributing to a potential collective prediction are identified, differentiated, classified, structurally compared, and assessed for meaningful source diversity before their predictions are treated as collectively representative.
PSDM distinguishes:
number of predictive sources
from:
meaningful diversity among predictive sources.
It determines whether apparent source multiplicity reflects materially different predictive origins, source types, institutional positions, informational environments, methodologies, expertise, geographic contexts, model families, agent structures, or other source characteristics relevant to collective prediction governance.
It transforms:
Contributing Predictive Sources + Source Characteristics + Predictive Contributions
into:
Governed Predictive Source Diversity Profile
Its foundational principle is:
More predictive sources do not necessarily create more predictive diversity.
Operational Role
PSDM is the first internal governance space of the Collective Prediction Governance Standard (CPGS) .
Before CPGS can assess whether multiple predictions may legitimately become a collective prediction, it must establish:
who or what is actually contributing to the collective.
PSDM asks:
How diverse are the predictive sources represented in the collective structure, and does apparent source multiplicity represent materially different predictive origins?
The primary output is:
Governed Predictive Source Diversity Profile
Internal Module Flow
Predictive Source Diversity → Prediction Dependence & Variation → Prediction Aggregation → Predictive Influence & Concentration → Collective Prediction Representation
Current Position: Predictive Source Diversity
Predictive Source
A Predictive Source is an identifiable human, model, agent, institution, market mechanism, forecasting system, or other origin from which a predictive contribution enters the collective prediction environment.
Source ≠ Prediction
The source is the origin of the contribution.
The prediction is the predictive output.
One source may produce multiple predictions.
Multiple Predictions ≠ Multiple Sources
A single forecasting system producing one hundred predictions does not create one hundred independent predictive sources.
Nominal Source
A Nominal Source is a separately identifiable contributor within the collective mechanism.
Effective Predictive Source
An Effective Predictive Source represents a materially distinct predictive origin after relevant structural overlap has been considered.
Nominal Source Count ≠ Effective Source Count
Source Identity
Each material predictive source should remain sufficiently identifiable for governance purposes.
Identity may include:
individual identity; organizational identity; model identity; agent identity; system identity; market-participant identity; source class.
Where personal identification is unnecessary or inappropriate, governance may operate through persistent pseudonymous or categorical source identifiers.
Identifiability ≠ Public Disclosure
Governance does not require every source identity to be publicly exposed.
Source Type
Predictive sources may include:
Human Predictor
Expert Predictor
Institutional Predictor
AI Model
Autonomous Agent
Prediction Market Participant
Forecasting System
Ensemble System
Hybrid Human-AI Source
Algorithmic Predictive Source
Other Governed Predictive Source
Different Source Types ≠ Independent Sources
A human and an AI system may still rely upon the same information.
Source Origin
Source Origin identifies the organizational, technical, informational, or operational environment from which a predictive source emerges.
Different Interface ≠ Different Source Origin
Two applications may expose predictions from the same underlying model.
Different Brand ≠ Different Predictive Origin
Source Family
Sources may belong to a common Source Family.
Examples include:
models from the same model family; analysts from the same institution; agents operating under the same orchestration system; forecasts generated from the same methodology; subsidiaries using the same forecasting engine.
Different Instance ≠ Different Source Family
Model Family Diversity
Where AI models contribute predictions, PSDM should preserve whether they arise from materially different model families.
Multiple Model Instances ≠ Model Diversity
Different Model Names ≠ Different Model Families by Default
Agent Diversity
Autonomous agents may appear as separate predictors.
PSDM determines whether they represent materially distinct predictive sources.
Multiple Agents ≠ Multiple Distinct Predictive Origins
Agents may share:
the same base model; the same tools; the same retrieval system; the same prompts; the same memory; the same orchestration layer.
Human Source Diversity
Human predictive sources may differ across:
expertise; institution; profession; geography; operational role; experience; access to information; decision proximity.
Different People ≠ Meaningfully Diverse Predictive Sources by Default
Institutional Diversity
Predictions may originate from multiple institutions.
Multiple Institutions ≠ Institutional Independence
Institutions may share:
data providers; research; methodologies; incentives; ownership; advisory networks.
PSDM records diversity.
Independence is assessed separately.
Organizational Diversity
Different organizational units may produce separate predictions.
Different Department ≠ Different Predictive Environment by Default
Geographic Diversity
Predictive sources may operate across different geographic environments.
Geographic diversity may matter where local information, conditions, or operational experience affect prediction.
Different Geography ≠ Information Diversity by Default
Jurisdictional Diversity
Predictors may operate under different legal, regulatory, or institutional environments.
This may materially affect:
information access; assumptions; interpretation; operational knowledge.
Jurisdictional Difference ≠ Predictive Superiority
Expertise Diversity
Predictive sources may represent different expertise domains.
Example:
engineering; finance; law; operations; economics; supply chain.
More Expertise Categories ≠ Better Prediction by Default
PSDM governs diversity, not superiority.
Role Diversity
Different operational roles may observe different parts of the same predicted system.
Role Difference ≠ Independent Information
Methodological Diversity
Predictors may use different predictive methodologies.
Examples include:
statistical forecasting; causal models; expert judgment; machine learning; scenario analysis; market mechanisms.
Methodological Diversity ≠ Prediction Independence
Different methodologies may still consume the same underlying information.
Methodological Similarity
Multiple predictors may use substantially similar methods.
This should remain visible where it materially reduces apparent diversity.
Different Implementation ≠ Different Methodology by Default
Information Environment Diversity
Predictive sources may operate within different information environments.
Examples include:
public information; proprietary datasets; internal operational data; local observations; expert networks; real-time system signals.
Different Sources ≠ Different Information Environments
Information Access Profile
PSDM may preserve the general type of information access available to a predictive source.
It does not govern the Prediction Basis itself.
PBAS owns Prediction Basis governance.
Source Diversity ≠ Prediction Basis Governance
Temporal Diversity
Predictive sources may form predictions at different times.
This may expose them to different information states.
Different Prediction Time ≠ Independent Source
Perspective Diversity
Predictors may approach the same Prediction Object from materially different analytical perspectives.
Perspective Difference ≠ Methodological Difference by Default
Incentive Diversity
Predictive sources may operate under different incentive structures.
Examples include:
commercial incentives;
research incentives; trading incentives; policy incentives; reputational incentives.
Different Incentives ≠ Better Diversity by Default
Incentive structure is a governance characteristic, not a quality score.
Ownership Diversity
Different predictive sources may share common ownership or control.
Different Legal Entity ≠ Different Control Structure
Common Ownership
Common ownership may affect interpretation of apparent source diversity.
Ownership Diversity ≠ Prediction Independence
Governance Diversity
Predictive sources may operate under different governance structures.
Different Governance Structure ≠ Independent Prediction
Technology Diversity
Predictive systems may rely upon different technical infrastructures.
Different Technology Stack ≠ Different Predictive Logic by Default
Data-Provider Diversity
Different predictors may rely upon different upstream data providers.
Where known and material, this can form part of the source diversity profile.
Multiple Predictors Using One Provider ≠ Multiple Information Origins
Source Multiplicity
Source Multiplicity describes the number of nominal predictive sources.
Source Diversity
Source Diversity describes the material structural differences among those sources.
Source Multiplicity ≠ Source Diversity
Diversity Dimension
A Diversity Dimension is a governed characteristic across which predictive sources may materially differ.
Relevant dimensions may include:
source type; source origin; source family; institution; ownership; geography; jurisdiction; expertise; operational role; methodology; information environment; technology; incentives; temporal position.
Diversity Across One Dimension ≠ Overall Diversity
Ten geographically diverse sources may still use the same model and data.
Diversity Profile
A Diversity Profile represents the collective structure across relevant diversity dimensions.
Diversity Profile ≠ Diversity Score by Default
A single numerical diversity score may hide material structural differences.
Diversity Breadth
Diversity Breadth represents the number of materially different dimensions represented.
Diversity Depth
Diversity Depth represents the degree of meaningful variation within a particular dimension.
Diversity Breadth ≠ Diversity Depth
Source Concentration
Source Concentration exists where predictive contributions are disproportionately drawn from one source class, institution, model family, geography, methodology, or other relevant dimension.
Large Source Count ≠ Low Source Concentration
Source Homogeneity
Source Homogeneity exists where predictive sources are substantially similar across material dimensions.
Homogeneous Sources ≠ Invalid Collective Prediction
Homogeneity is a governance characteristic requiring appropriate interpretation.
Source Heterogeneity
Source Heterogeneity exists where predictive sources materially differ across relevant dimensions.
Heterogeneity ≠ Predictive Quality
Artificial Diversity
Artificial Diversity occurs where sources appear distinct at the interface or identity level while materially sharing the same underlying predictive origin.
Examples include:
multiple wrappers around one model; multiple agents using one base model; multiple analyst reports derived from one upstream forecast; subsidiaries using one forecasting engine.
Interface Diversity ≠ Predictive Source Diversity
Duplicated Predictive Source
A Duplicated Predictive Source occurs where the same effective predictive origin enters the collective structure more than once in a way that may create false appearance of diversity.
Duplicate Contribution ≠ Independent Confirmation
Source Clustering
Sources with materially similar characteristics may be grouped into Source Clusters.
Examples:
Model Family A
Institutional Analysts
Public-Information Retail Participants
Internal Operational Experts
Source Cluster ≠ Single Predictor
Clustering is a governance representation, not automatic elimination of individual contributions.
Dominant Source Cluster
A Dominant Source Cluster materially represents a large portion of the collective source structure.
Numerical Majority ≠ Informational Majority by Default
Underrepresented Source Dimension
A relevant source dimension may be insufficiently represented.
Example:
A global supply-chain forecast may contain financial analysts and AI models but no operational logistics sources.
Missing Perspective ≠ Prediction Error by Default
It represents a potential governance limitation.
Representation Gap
A Representation Gap exists where a source class materially relevant to the Prediction Object is absent or insufficiently represented.
Representation Gap ≠ Automatic Invalidity
The gap should be preserved for later collective interpretation.
Source Coverage
Source Coverage describes which relevant predictive source categories or perspectives are represented.
Source Coverage ≠ Population Representation
Population Representation
Where a collective prediction claims to represent a population, the relevant population relationship must remain explicit.
Participation ≠ Population Representation
Self-Selection
Collective mechanisms may contain self-selected participants.
Self-Selected Population ≠ Representative Population
Access Bias
Participation may depend upon access to:
technology; capital; language; platform; institutional membership; geography.
Accessible Participants ≠ Relevant Population
Expert Selection
Expert panels may intentionally use non-representative participants because expertise rather than population representation is the objective.
Expert Panel ≠ Population Sample
The relevant representation claim must therefore be explicit.
Source Relevance
Not every source type is equally relevant to every Prediction Object.
Diversity ≠ Relevance
A highly diverse source population may still contain sources with limited relevance.
Relevant Diversity
Relevant Diversity concerns diversity across dimensions materially connected to the Prediction Object.
Maximum Diversity ≠ Optimal Collective Structure
PSDM does not require maximum diversity.
It requires that the diversity actually present be understood and represented accurately.
Predictor–Reality Proximity Profile
Where material, PSDM may preserve the broad Predictor–Reality Proximity position of contributing source classes.
Relevant positions may include:
Public-Information Predictor
Information-Advantaged Predictor
Decision-Proximate Predictor
Outcome-Informed Predictor
Outcome-Influencing Predictor
Outcome-Controlling Predictor
Predictor–Reality Proximity ≠ Predictive Source Quality
High Predictor–Reality Proximity ≠ Misconduct
The classification describes relationship to the predicted reality.
It does not assign wrongdoing.
Source Diversity and Predictor–Reality Proximity
A collective may appear diverse across institutions while all material contributors occupy the same Predictor–Reality Proximity position.
Conversely, contributors from the same organization may observe materially different operational layers.
Institutional Diversity ≠ Predictor–Reality Diversity
Source Diversity State
PSDM may classify the source structure as:
Materially Diverse
Partially Diverse
Dimensionally Concentrated
Highly Homogeneous
Artificially Diverse
Insufficiently Characterized
Unable to Determine
Materially Diverse
Materially different predictive source characteristics exist across relevant dimensions.
Partially Diverse
Meaningful diversity exists but remains concentrated or absent across other material dimensions.
Dimensionally Concentrated
Sources are diverse in some respects but materially concentrated in a specific relevant dimension.
Highly Homogeneous
Sources share substantial structural similarity across relevant dimensions.
Artificially Diverse
Nominal source multiplicity materially overstates the number of distinct predictive origins.
Insufficiently Characterized
The source population exists but available information is insufficient to establish a reliable diversity profile.
Unable to Determine
Available governance information is insufficient to determine source diversity.
Unable to Determine ≠ Diverse
Diversity Change
The source structure may change as contributors enter or leave the collective.
Current Diversity ≠ Historical Diversity
Source Replacement
Replacing one predictor with another may or may not materially alter diversity.
New Source ≠ New Predictive Diversity by Default
Source Diversity Version
Material changes in the source population should remain historically traceable where relevant.
Later Diversity ≠ Historical Diversity
Relationship to Prediction Dependence
PSDM identifies and characterizes source diversity.
The next governance space determines dependence and variation among predictions.
Source Diversity ≠ Prediction Independence
This distinction is fundamental.
Two highly diverse sources may still produce dependent predictions.
Two structurally similar sources may sometimes produce genuinely independent predictions.
Relationship to Prediction Aggregation
Aggregation should not interpret raw source count as evidence of diversity.
The Governed Predictive Source Diversity Profile informs subsequent composition.
Source Count ≠ Aggregation Weight
Relationship to Predictive Influence
Source diversity does not determine how much influence each source receives.
Influence and concentration are governed separately.
Diversity ≠ Influence
Relationship to PBAS
PBAS governs the Prediction Basis of individual predictions.
PSDM may classify broad informational environments but does not establish, validate, or govern the Prediction Basis itself.
Relationship to PAS
PAS governs Predictive Assumptions.
PSDM may identify methodological or perspective diversity without governing individual assumptions.
Relationship to PPUS
PPUS governs Probability & Uncertainty.
PSDM does not evaluate probability values.
Relationship to PFES
PFES governs Prediction Feasibility.
Source diversity does not establish feasibility.
Relationship to ORA™
ORA™ governs Operational Reality.
PSDM may preserve source proximity to operational conditions without determining those conditions.
Governance Requirements
PSDM requires that:
1. material predictive sources remain identifiable; 2. source identity remain distinguishable from predictive contribution; 3. multiple predictions from one source not automatically be counted as multiple sources; 4. nominal source count remain distinguishable from effective source count; 5. identifiability remain distinguishable from public disclosure;
6. source types remain classifiable where material; 7. different source types not automatically be treated as independent sources; 8. source origin remain identifiable where relevant; 9. different interfaces not automatically establish different predictive origins; 10. source-family relationships remain representable; 11. different instances not automatically establish different source families; 12. model-family diversity remain governable where AI models contribute; 13. multiple model instances not automatically establish model diversity; 14. agent diversity remain distinguishable from agent multiplicity; 15. human source diversity remain assessable across relevant dimensions; 16. institutional diversity remain distinguishable from institutional independence; 17. organizational diversity remain distinguishable from organizational separation; 18. geographic diversity remain distinguishable from information diversity; 19. jurisdictional diversity remain representable where material; 20. expertise diversity remain identifiable where relevant; 21. role diversity remain distinguishable from information independence; 22. methodological diversity remain distinguishable from Prediction Independence; 23. information-environment diversity remain representable without replacing PBAS; 24. temporal diversity remain distinguishable from source independence; 25. perspective diversity remain distinguishable from methodology; 26. incentive diversity remain representable where material; 27. ownership relationships remain identifiable where relevant; 28. different legal entities not automatically establish different control structures; 29. technology diversity remain distinguishable from predictive-logic diversity; 30. upstream data-provider concentration remain representable where known and material; 31. Source Multiplicity remain distinguishable from Source Diversity; 32. diversity dimensions remain explicitly identifiable; 33. diversity across one dimension not establish overall diversity; 34. Diversity Breadth remain distinguishable from Diversity Depth; 35. source concentration remain identifiable; 36. homogeneity remain representable without automatic invalidation; 37. heterogeneity not be interpreted as predictive quality by default; 38. Artificial Diversity remain identifiable; 39. duplicated predictive origins remain representable; 40. duplicate contributions not be treated as independent confirmation; 41. source clustering remain available where useful; 42. dominant source clusters remain identifiable; 43. Representation Gaps remain preservable; 44. Source Coverage remain distinguishable from Population Representation; 45. participation remain distinguishable from population representation; 46. self-selection remain identifiable where relevant; 47. access bias remain representable; 48. expert panels remain distinguishable from population samples; 49. source diversity remain distinguishable from source relevance; 50. Relevant Diversity remain distinguishable from maximum diversity; 51. Predictor–Reality Proximity remain representable where material; 52. Predictor–Reality Proximity not be treated as source quality; 53. high Predictor–Reality Proximity not imply misconduct; 54. institutional diversity remain distinguishable from Predictor–Reality diversity; 55. source diversity states remain explicit; 56. Unable to Determine remain available as a legitimate state; 57. new sources not automatically be treated as new predictive diversity; 58. historical source diversity states remain reconstructable; 59. source diversity remain distinct from Prediction Independence; 60. source diversity remain distinct from aggregation weight and predictive influence.
Governance Logic
PSDM applies the following governance logic:
Contributing Predictive Sources
Predictive Contributions
Available Source Characteristics
↓
Predictive Sources Identified
↓
Source / Prediction Distinction Established
↓
Nominal Source Count Established
↓
Source Types Classified
↓
Source Origins Identified where material
↓
Source Families Identified
↓
Institutional / Organizational Relationships Assessed
↓
Model / Agent Relationships Assessed where applicable
↓
Geographic / Jurisdictional Diversity Assessed where relevant
↓
Expertise / Role Diversity Assessed
↓
Methodological Diversity Assessed
↓
Information Environment Diversity Characterized
↓
Ownership / Incentive / Technology Characteristics Assessed where material
↓
Relevant Diversity Dimensions Established
↓
Source Concentration Assessed
↓
Artificial Diversity / Duplicate Origins Assessed
↓
Source Clusters Identified where useful
↓
Representation Gaps Assessed
↓
Predictor–Reality Proximity Profile Preserved where material
↓
Source Diversity State Established
↓
Historical Diversity State Preserved
↓
Governed Predictive Source Diversity Profile
↓
Prediction Dependence & Variation
Where sufficient characterization cannot be established:
Predictive Source Diversity Governance Gap → Source Clarification / Source-Family Analysis / Representation Assessment / Diversity Qualification / Insufficiently Characterized / Unable to Determine
Minimum Implementation Framework
A minimum implementation of PSDM must establish:
A. Collective Prediction Reference
The collective prediction process or candidate collective prediction to which the source population belongs.
B. Source Identifier
A persistent identifier sufficient to distinguish each material predictive source.
C. Source Type
The applicable human, institutional, model, agent, market, hybrid, or other source class.
D. Source Origin
The organizational, technical, or operational origin where material.
E. Source Family
Any materially shared source-family relationship.
F. Relevant Diversity Dimensions
The dimensions relevant to the Prediction Object.
G. Source Characteristics
The material characteristics used to assess diversity.
H. Nominal Source Count
The number of separately represented predictive sources.
I. Effective Source Structure
A governed representation of materially distinct predictive origins.
J. Source Concentration
Any material concentration across relevant dimensions.
K. Source Cluster
Where materially useful.
L. Representation Gap
Any materially missing or underrepresented source dimension.
M. Predictor–Reality Proximity
Where materially relevant.
N. Source Diversity State
At minimum:
Materially Diverse / Partially Diverse / Dimensionally Concentrated / Highly Homogeneous / Artificially Diverse / Insufficiently Characterized / Unable to Determine
O. Historical Diversity Version
The reconstructable source diversity state applicable at the relevant prediction lifecycle point.
Governance Outputs
PSDM may produce:
Governed Predictive Source Diversity Profile; Predictive Source Register; Source Identifier; Source Type Classification; Source Origin Record; Source Family Record; Nominal Source Count; Effective Source Structure; Human Source Diversity Profile; Institutional Diversity Profile; Model Family Diversity Profile; Agent Diversity Profile; Geographic Diversity Profile; Jurisdictional Diversity Profile; Expertise Diversity Profile; Role Diversity Profile; Methodological Diversity Profile; Information Environment Diversity Profile;
Ownership Diversity Profile; Technology Diversity Profile; Diversity Dimension Register; Diversity Breadth; Diversity Depth; Source Concentration Record; Source Homogeneity State; Source Heterogeneity State; Artificial Diversity Record; Duplicate Predictive Origin Record; Source Cluster; Dominant Source Cluster; Source Coverage Record; Representation Gap; Predictor–Reality Proximity Profile; Source Diversity State; Predictive Source Diversity Governance Gap; historical source diversity state; Source Diversity Version.
The primary governed output is:
Governed Predictive Source Diversity Profile
which becomes an input to:
Prediction Dependence & Variation.
Critical Distinctions
More Predictive Sources Do Not Necessarily Create More Predictive Diversity
Source ≠ Prediction
Multiple Predictions ≠ Multiple Sources
Nominal Source Count ≠ Effective Source Count
Identifiability ≠ Public Disclosure
Different Source Types ≠ Independent Sources
Different Interface ≠ Different Source Origin
Different Brand ≠ Different Predictive Origin
Different Instance ≠ Different Source Family
Multiple Model Instances ≠ Model Diversity
Different Model Names ≠ Different Model Families by Default
Multiple Agents ≠ Multiple Distinct Predictive Origins
Different People ≠ Meaningfully Diverse Predictive Sources by Default
Multiple Institutions ≠ Institutional Independence
Different Department ≠ Different Predictive Environment by Default
Different Geography ≠ Information Diversity by Default
Jurisdictional Difference ≠ Predictive Superiority
More Expertise Categories ≠ Better Prediction by Default
Role Difference ≠ Independent Information
Methodological Diversity ≠ Prediction Independence
Different Implementation ≠ Different Methodology by Default
Different Sources ≠ Different Information Environments
Source Diversity ≠ Prediction Basis Governance
Different Prediction Time ≠ Independent Source
Perspective Difference ≠ Methodological Difference by Default
Different Incentives ≠ Better Diversity by Default
Different Legal Entity ≠ Different Control Structure
Ownership Diversity ≠ Prediction Independence
Different Governance Structure ≠ Independent Prediction
Different Technology Stack ≠ Different Predictive Logic by Default
Multiple Predictors Using One Provider ≠ Multiple Information Origins
Source Multiplicity ≠ Source Diversity
Diversity Across One Dimension ≠ Overall Diversity
Diversity Profile ≠ Diversity Score by Default
Diversity Breadth ≠ Diversity Depth
Large Source Count ≠ Low Source Concentration
Homogeneous Sources ≠ Invalid Collective Prediction
Heterogeneity ≠ Predictive Quality
Interface Diversity ≠ Predictive Source Diversity
Duplicate Contribution ≠ Independent Confirmation
Source Cluster ≠ Single Predictor
Numerical Majority ≠ Informational Majority by Default
Missing Perspective ≠ Prediction Error by Default
Representation Gap ≠ Automatic Invalidity
Source Coverage ≠ Population Representation
Participation ≠ Population Representation
Self-Selected Population ≠ Representative Population
Accessible Participants ≠ Relevant Population
Expert Panel ≠ Population Sample
Diversity ≠ Relevance
Maximum Diversity ≠ Optimal Collective Structure
Predictor–Reality Proximity ≠ Predictive Source Quality
High Predictor–Reality Proximity ≠ Misconduct
Institutional Diversity ≠ Predictor–Reality Diversity
Unable to Determine ≠ Diverse
Current Diversity ≠ Historical Diversity
New Source ≠ New Predictive Diversity by Default
Later Diversity ≠ Historical Diversity
Source Diversity ≠ Prediction Independence
Source Count ≠ Aggregation Weight
Diversity ≠ Influence
Enterprise Application
PSDM enables organizations to determine whether a supposedly broad predictive collective actually contains meaningful source diversity.
This is particularly important for:
prediction markets; expert panels; AI model ensembles; multi-agent systems; institutional forecasting; geopolitical forecasting; financial forecasting; crowdsourced predictions; Human-AI forecasting systems.
A collective mechanism can contain thousands of contributions while representing only a small number of effective predictive origins.
PSDM makes that structure visible before source multiplicity is interpreted as collective predictive strength.
Use Case 1 — Twenty AI Forecasts, Three Effective Model Families
A company asks twenty AI services to predict whether a new technology will achieve mass-market adoption within three years.
The twenty services produce separate forecasts.
Management initially treats this as:
20 independent AI predictions.
PSDM examines the source structure.
The analysis shows that many services are wrappers or specialized interfaces around the same underlying model families.
The twenty nominal sources reduce structurally to:
eight services using Model Family A; six using Model Family B; four using Model Family C; two genuinely distinct forecasting systems.
The critical governance question is:
Do twenty separately branded predictive interfaces represent twenty meaningfully diverse predictive sources?
PSDM preserves the nominal source population while establishing the underlying source-family structure.
Different Brand ≠ Different Predictive Origin
Nominal Source Count ≠ Effective Source Count
Use Case 2 — A Global Forecasting Crowd with Almost No Operational Representation
A global forecasting platform collects predictions from:
30,000 participants
about whether severe semiconductor supply constraints will occur during the next year.
The participant population is large and geographically distributed.
However, PSDM finds that the overwhelming majority of participants are public-information forecasters.
Very few contributors come from:
semiconductor manufacturing; equipment suppliers; logistics operations; industrial procurement; operational supply-chain management.
The critical governance question is:
Does a very large forecasting population represent meaningful predictive diversity when source classes closest to the operational system are barely represented?
PSDM identifies a:
Representation Gap
and preserves the difference between:
large participation
and:
relevant predictive source diversity.
Crowd Size ≠ Predictive Quality
Participation ≠ Representation
Canonical Boundary
PSDM governs the identification, differentiation, classification, structural comparison, concentration, clustering, representation, and historical preservation of predictive sources contributing to a potential collective prediction. It governs source type, source origin, source family, model and agent diversity, institutional and organizational diversity, geographic and jurisdictional diversity, expertise and role diversity, methodological and information- environment diversity, ownership and technology characteristics, relevant diversity dimensions, Artificial Diversity, Representation Gaps, and Predictor–Reality Proximity where material. It does not determine Prediction Independence, govern the Prediction Basis itself, calculate predictive probability, determine Prediction Feasibility, aggregate predictions, assign predictive influence, determine collective legitimacy, or govern final Collective Prediction Representation.
PSDM governs:
how much meaningful predictive source diversity actually exists behind the apparent source population.
It does not determine:
whether those sources are sufficiently independent to form a legitimate collective prediction.
Canonical Closing Statement
A collective can contain thousands of contributors and still draw from only a narrow predictive world.
PSDM separates source count from source diversity, interface multiplicity from predictive origin, and broad participation from meaningful representation.
Before collective intelligence can be claimed, governance must first establish how many genuinely different predictive perspectives are actually present.