1. Purpose

The Goodhart / Learning Drift Restorer diagnoses when a metric has become the target and selection has drifted away from coherence.

2. Core Question

Is the system learning the proxy instead of improving coherence?

3. Construct Class

Type: Diagnostic / Restoration Workflow

Source status: Canon-Ready

Primary module: Cybernetics

Related modules: AI, Security, Scaling, Coherence

4. Inputs

• target metric
• actual coherence outcome
• feedback integrity
• evaluator independence
• hidden debt
• proxy reward
• field outcome
• recurrence

5. Outputs

• Goodhart status
• evaluator capture risk
• proxy divergence
• restoration sequence
• metric redesign need
• field validation plan

6. Failure Modes Detected

Not specified in the registry index.

7. Restoration Links

Not specified in the registry index.

8. Tool Readiness

Workflow Ready

9. Tool Potential

This can become a tool for AI evaluations, institutional KPIs, security dashboards, education systems, and governance metrics.

10. Source Notes

This compact spec sheet was generated from the Constructs Registry index. It preserves the current source fields and should be expanded against the full construct template during editorial review.

11. Machine-Readable Summary

construct_id: "CONSTRUCT-014"
title: "Goodhart / Learning Drift Restorer"
abbreviation: ""
construct_class: "Diagnostic / Restoration Workflow"
tool_readiness: "Workflow Ready"
primary_module: "Cybernetics"
source_status: "Canon-Ready"
registry_source: "content/archive/constructs/registry/index.md"