law-benefit-score
Copyright 2026 Flemming.AI

This product includes software developed by Dr. Stefan Flemming
and contributors (https://flemming.ai/).

------------------------------------------------------------------
Third-party acknowledgments

JSON (de)serialization uses `serde` + `serde_json`
(https://github.com/serde-rs), Apache-2.0 / MIT licensed.

Module SDK glue is provided by `chain-module-sdk`
(https://git.flemming.ai/fai/chain-module-sdk-rust), Apache-2.0.

------------------------------------------------------------------
Methodological lineage

The seven-dimensional benefit taxonomy (Schutz, Markt,
Rechtssicherheit, EU-Binnenmarkt, Sicherheit, Umwelt,
Einnahmen) is juristisch-politisch konventioniert. It draws
on the EU Better Regulation Guidelines + Toolbox 2023
("Multi-Criteria Decision Analysis"-Vokabular) and on the
methodological framing of Studie §3.2.

Dispersion = sample standard deviation across the seven
dimensions. Surfaces "high mean but split panel" cases where a
single dimension dominates and others disagree — the audit
log preserves the per-dimension input so the Juristen-/Beirats-
panel can reconstruct the disagreement.

Inter-rater reliability (Cohen's κ ≥ 0.7) is the methodological
freeze-threshold per Studie §3.3; v0.2 will expose a per-
dimension Juristen-vote panel that computes κ across reviewers
and adds it to the audit log.
