Methodology
A result is only as useful as its inclusion rule.
This methodology governs the BlackLabel Public Build Evidence Registry. Future studies receive their own protocols, questionnaires, exclusions, uncertainty statements, and data dictionaries before results are released.
Protocol
Version 1.0 inclusion and validation rules.
- Unit of analysis
- One public BlackLabel case-study record with a stable slug and named subject.
- Inclusion
- The record must describe a real build and identify its current public status. Restricted or placeholder records are excluded from released datasets.
- Metric disclosure
- A metric counts only when the record publishes a non-placeholder value. Live sources are labeled separately rather than copied into the static count.
- Evidence control
- Numerical claims remain source-commented in the canonical registry and must resolve to internal delivery records or a live production surface before publication.
- Privacy
- Private customer records, credentials, personal data, and undisclosed commercial outcomes are excluded.
- Versioning
- Material row, method, or field changes increment the dataset version and update the reviewed date.
- Corrections
- Corrections replace the public record, retain a dated change note, and propagate to JSON, CSV, schema, and the associated evidence page.
Reproduce
The calculation is deliberately small.
- 1. Load the canonical case-study registry.
- 2. Keep every published, non-restricted record with a permanent slug.
- 3. Count non-editable, non-placeholder static metric disclosures.
- 4. Label records backed by a live metric source separately.
- 5. Export identical rows through JSON and CSV distributions.