Purpose and reading rule

This is the single orientation point for the project’s present state. It is a plain-language synthesis for contributors, reviewers, prospective users, and automated documentation readers. It does not create architecture, authority, a production authorization, or a claim about any participant. The linked canonical documents govern when they differ from this summary.

Machine-readable status

{
  "project": "OSI-PIA",
  "as_of": "2026-10-01",
  "overall_authority": "working",
  "overall_status": "proposed",
  "production_participant_processing": "not_authorized",
  "participant_data_in_repository": "prohibited",
  "graph_projection": {
    "synthetic_local_sandbox": "validated",
    "real_participant_projection": "not_authorized",
    "real_organizational_projection": "not_authorized"
  },
  "current_focus": "governed artifact metadata, capability infrastructure publication, semantic discoverability, literature positioning, standards interoperability, and machine participation conformance",
  "canonical_sources": [
    "README.md",
    "ROADMAP.md",
    "governance/GOVERNANCE_MODEL.md",
    "architecture/pia-intake/PIA_Intake_Subsystem_Framework.md",
    "architecture/graph_ontology/Graph_Architecture.md"
  ]
}

Executive state

Current Public Status

As of September 2026, OSI-PIA has an active public GitHub Pages site, a published public article series, documentation assurance records, registry and link validation, and an enabled public discussion channel. The project remains an active research and development system. It is not a production hiring, employment-screening, psychological-assessment, or personnel-decision tool.

The public article series now includes the machine-participation sequence The False Negative Stack, Access Is Not Authority, and When the Gate Is Right and Still Wrong, followed by Temporal Authority and Capability Infrastructure Should Belong to the People It Describes. These extend the series from access and source authority into chronology, portability, and the participant as a primary beneficiary. The new capstone You Can’t Find Work, and Companies Can’t Find Workers, Because the Labor Market Can’t Read People frames these concerns as a capability-infrastructure problem spanning evidence, translation, recognition, deployment, and governance. Publication remains a supporting explanation layer and does not promote exploratory research into canonical authority.

Artifact metadata is now governed through a versioned machine-readable contract and validator-backed controlled vocabularies. Publication creation follows a documented, reusable template path for individual articles and article series; metadata remains an interface for interpretation, not a substitute for human review or stewardship.

The project now maintains a living Project Bibliography with 41 article-cited source locators and seven research-candidate sources for PIA and OSI positioning. A working Literature-Positioning Matrix and accompanying PIA/OSI positioning notes separate established findings, recombination, and open propositions; they do not make novelty or causal claims. This checkpoint is recorded in the literature bibliography and positioning milestone.

A provisional PIA <-> Open Skills / LER Crosswalk now defines the working boundary between PIA’s richer evidence-and-inference layer, a future semantic translation socket, established portable standards, and OSI’s organizational receiving conditions. It is not canonical architecture or an authorization to export participant data.

The public Briefs and Project Updates section now preserves concise project-history context alongside the article series. Its first brief records the independent convergence with the Open Skills/LER field and the resulting human-centered division of labor. The Start Here orientation now links directly into the public publications, article series, briefs, and bibliography so repository and Pages navigation share one reading path.

OSI-PIA is a governed research-and-engineering project with two peer domains:

  • OSI examines organizational systems, conditions, relationships, and state change.
  • PIA helps a participant make their evidence, experience, preparation, capabilities, and development more visible without reducing them to a score or permanent label.

The project has a certified shared assurance foundation and a substantial PIA working implementation. It is not a production service and is not authorized for unsupervised real-participant processing. The central design commitment is that evidence, interpretation, review, and graph representation remain distinct and traceable.

What is operationally demonstrated

Area Present state Important boundary
Shared assurance Assurance Framework v1.0, CSV assurance engine, contracts, and regression checks are established Assurance does not itself authorize a downstream decision or graph write
Protected local intake Local encrypted participant-store candidate, roles, malware inspection, withdrawal, deletion, retention, recovery, and audit-chain validation are implemented and tested Working/proposed; not production security certification or public deployment
Evidence extraction and review Bounded text extraction, encrypted derived text, provenance, and participant/reviewer keep-correct-exclude decisions are implemented Unsupported or unreadable material is routed for review; extraction is not a capability conclusion
Credential meaning Participant-free credential catalog, resolution, definition review, and limited lookup routing are implemented Credential meaning does not prove completion, application, or performance
Capability mapping Source-grounded mapping proposals, separate reviewer decisions, supersession, and output-assurance holds are implemented A mapping is a bounded interpretation, not a permanent trait or score
Participant outputs Working participant preview, technical companion, correction path, and dry-run manifest are implemented No published report or production participant claim is authorized
Graph mechanics One embedded synthetic assertion was imported twice into local PIA-Sandbox, then read-only validation proved no duplicate nodes, relationship, or path Real participant projection, production target use, and durable import audit remain gated
OSI organizational evidence assurance A participant-free synthetic organizational package is validated from organization and provenance records through a bounded observation candidate No OSI diagnostic, Trust/Flow/Health construct, or organizational decision is authorized
OSI graph mechanics Three embedded synthetic organization -> source -> evidence -> observation paths were imported twice into local OSI-Sandbox; read-only validation proved three paths with no duplicate structure Real organizational projection, osi-reference import, production target use, and diagnostics remain gated

Current evidence flow

Source material
  -> protected intake and provenance
  -> extraction and evidence review
  -> bounded mapping proposal and separate review
  -> participant preview + technical companion + dry-run manifest
  -> synthetic-only sandbox projection assurance
  -> future authorized participant-minimized projection

At every arrow, the system is intended to preserve source identity, limits, review state, and correction history. A later stage cannot silently convert an earlier-stage claim into stronger evidence.

OSI now has a verified synthetic-only projection path through OSI-Sandbox covering three distinct bounded observations. That path deliberately precedes real organizational data, osi-reference imports, and any organizational diagnostic or analytic claim.

Explicitly not authorized or not yet complete

  • production, public, multi-user, or network-exposed participant intake;
  • unsupervised real-participant processing;
  • participant-data projection into Neo4j;
  • automatic consequential assessment, ranking, hiring, exclusion, or employability claims;
  • image and scanned-document OCR intake;
  • durable graph import audit, rollback, exception handling, and graph-side deletion/retention semantics for participant projection;
  • formal independent-review and exception governance for consequential use;
  • production threat modeling, incident response, monitoring, accessibility, backup/recovery operations, and external privacy/security review; and
  • promotion of the working PIA intake architecture, capability vocabulary, or Governance Model into canonical authority.
  • OSI graph import, diagnostic output, organizational scoring, or use of planned OSI constructs such as Trust, Flow, or Organizational Health, except for the one embedded synthetic-only OSI-Sandbox test path.

Current governance position

The proposed Governance Model has reached Congruence and has a completed ratification review, but remains working/proposed. The ratification review requires a scoped shared ADR before promotion.

The PIA Measurement Doctrine is the practical interpretive safeguard: the system evaluates evidence supporting a bounded claim, not a person’s worth or universal potential. The Clean Release Standard and repository validation enforce a participant-data-free committed tree.

The working Human-Centered Capability Infrastructure principle adds a shared design test: capability infrastructure should make a person’s capability understandable, portable, actionable, and developable for that person—not merely more useful to institutions or technical intermediaries. It remains proposed pending governance review.

Next governed decisions

  1. Define a participant-minimized projection contract: allowed fields, approval authority, target declaration, deletion/retention behavior, and acceptable assurance conditions.
  2. Define durable graph import audit, rollback, exception, and post-write validation requirements before any non-synthetic projection.
  3. Consolidate operational-readiness requirements for a controlled pilot: consent, privacy, threat model, incident response, key recovery, support, accessibility, and independent review.
  4. Decide whether to ratify or revise the Governance Model through the scoped shared ADR required by its ratification review.
  5. Review the seven bibliography candidates and update the literature-positioning matrix only after recording source-level limits and temporal scope.

Canonical detail