FDE BenchmarkPublic document reader
Available in sourcev0.1 · live public draft · official release pending

Canonical source view. This renders the bundled repository document as readable HTML. Source metadata remains in the repository, and relative document links remain inside this reader.

Synthetic reproducibility demonstration

Artifact class: executable synthetic software check
Release candidate: 0.1.0-draft
Human participants: none
Official score or credential: none

This package demonstrates one narrow property: the released experimental scorer produces the same frozen summary from the same conspicuously synthetic input in the supported source environment. It does not reproduce a human finding or establish construct validity, reliability, fairness, field performance, reviewer agreement, or employment utility.

Run

From the repository root:

node --experimental-strip-types releases/0.1.0-draft/reproducibility/run.mjs

The command:

  1. verifies every SHA-256 receipt in manifest.json against the current source bytes;
  2. reads SCHEMAS/examples/valid/scoring-input.json;
  3. executes packages/experimental-scorer/src/index.ts;
  4. projects a small stable summary rather than a human-facing score;
  5. compares that summary byte-for-value with expected-summary.json; and
  6. exits nonzero if a receipt or result changes.

The scorer is expected to return insufficient_evidence, suppress numeric output, and label the run as experimental and synthetic. Any change must be reviewed rather than silently updating the expected file.

Package contents

File Function
run.mjs Deterministic receipt verification, runner, and output assertion
expected-summary.json Frozen expected projection
manifest.json SHA-256 receipts for the demonstration and its direct inputs
SYNTHETIC_REPRODUCIBILITY_DEMONSTRATION.md Public interpretation, limits, and maintenance rules

Interpretation boundary

  • Every identifier in the input explicitly names a synthetic fixture, subject, reviewer, or policy.
  • The input states synthetic: true, non_inferability_attested: true, release_mode: calibration, and official_numeric_output_allowed: false.
  • The frozen summary contains no person, candidate, company, customer, or real field record.
  • A passing run establishes implementation consistency for this fixture only.
  • This package has not been independently reproduced or reviewed.
  • The draft release remains unauthorized for real-data intake, official scoring, ranking, certification, or hiring use.

The repository source manifest must be regenerated for the final candidate after all release files are frozen. This package manifest does not replace that release-wide receipt.