Evaluation Artifacts — How to Read This Folder

This folder contains the evaluation artifacts for RGDS (Regulated Gate Decision Support).

Together, these files define how decision quality and governance effectiveness are assessed in a regulated, human-governed decision-support system.

These artifacts are aligned to the v2.0.0 whitepaper-defined governance baseline, including mandatory options analysis, evidence completeness classification, residual risk capture, named human accountability, and bounded AI disclosure.

The artifacts are intentionally layered.
They should be read in sequence, not as isolated documents.

---

Recommended Reading Flow

1. Evaluation Note — Interpretive Context

File:
evaluation-note.md

Purpose:
Provides interpretive context for RGDS evaluation.

• Explains why specific evaluation measures exist
• Reinforces governance principles (e.g., AI is not evidence by default)
• Prevents misinterpretation of metrics as model benchmarking

This document is non-authoritative and conceptual.
It exists to orient readers before reviewing formal evaluation definitions.

---

2. Evaluation Plan — Authoritative Evaluation Contract

File:
evaluation-plan.md

Purpose:
Defines how RGDS is evaluated as a human-governed decision-support system.

The evaluation plan specifies:

• evaluation scope
• decision quality dimensions
• governance and AI-use checks
• roles and responsibilities
• evaluation cadence and success criteria

This is the authoritative evaluation definition.
All other evaluation artifacts align to this plan.

---

3. Evidence Quality Rubric — Scoring & Judgment Framework

File:
evidence-quality-rubric.md

Purpose:
Provides structured criteria for assessing whether evidence is fit for the decision being made.

The rubric supports:

• consistent reviewer scoring
• explicit confidence assignment
• transparent discussion of gaps and limitations

It operationalizes the “evidence quality” dimension defined in the Evaluation Plan.

---

4. Decision Gate Extract — Flattened, Review-Ready View

File:
decision-gate-extract.md

Purpose:
Defines a flattened decision representation derived from RGDS decision logs.

This extract is designed for:

• phase-gate forums
• executive reviews
• portfolio-level analysis

It preserves traceability to:

• external requirements
• observed gaps
• backlog items
• full decision log JSON

The decision log remains the source of truth.

The Decision Gate Extract is a read-only derivative and must never be used to approve, override, or reinterpret a decision.

---

5. Decision Gate Extract — Power BI Sample — Operational Example

File:
decision-gate-extract-powerbi-sample.md

Purpose:
Provides a concrete example of how the Decision Gate Extract can be operationalized in a BI tool (e.g., Power BI).

This sample illustrates:

• a BI-ready schema
• example transformations
• governance-focused dashboard views
• drill-through to auditable decision records

It is illustrative only and does not mandate tooling or implementation.

---

Supporting Traceability & Scoring Artifacts

The following artifacts support traceability, defensibility, and repeatability across all evaluations.

---

6. IND Requirements Gap Log — Why Work Exists

File:
ind-requirements-gap-log.md

Purpose:
Documents observed gaps between external IND expectations and common delivery practices.

Each gap:

• is assigned a stable IND-GAP-XXX identifier
• is linked to one or more backlog items
• provides evidence-backed justification for RGDS design decisions

This artifact explains why evaluation and backlog work exists.

---

7. Requirements Traceability Matrix — End-to-End Alignment

File:
requirements-traceability-matrix.md

Purpose:
Provides end-to-end traceability across:

External requirement
→ Observed gap
→ Backlog item
→ RGDS artifact

The RTM enables:

• governance review
• audit readiness
• confirmation that evaluation artifacts map back to real requirements

This is the primary traceability artifact for reviewers.

---

8. Scorecard Template — Per-Decision Capture

File:
scorecard-template.csv

Purpose:
Defines a lightweight, repeatable format for capturing per-decision evaluation signals, including:

• reviewer confidence
• evidence completeness
• governance execution indicators
• decision outcome context

The scorecard supports:

• per-decision evaluation
• phase-level aggregation
• retrospective trend analysis

Scorecards support evaluation only; they do not constitute approval, rejection, or risk acceptance.

---

How These Artifacts Fit Together

Evaluation Note
→ Evaluation Plan
→ Evidence Quality Rubric
→ Decision Gate Extract
→ Power BI Sample

Supported by:
→ IND Requirements Gap Log
→ Requirements Traceability Matrix
→ Scorecard Template

• The Note explains intent
• The Plan defines evaluation
• The Rubric enables scoring
• The Extract supports review
• The Sample shows operationalization
• The Gap Log & RTM ensure traceability
• The Scorecard captures execution evidence

Together, they ensure RGDS evaluation is:

• defensible
• interpretable
• auditable
• explicitly human-governed

---

Notes

• Evaluation artifacts are expected to evolve as RGDS matures.
• Changes should preserve traceability and backward compatibility.
• BI tooling and AI assistance must never become decision authorities.

RGDS evaluation exists to support better human decisions, not to replace them.