feat(persuasive): overjustification guardrail + CRISP merit design note #46
Merged
wasabi opened 1 week ago

Summary

Implements the constructive-only Persuasive Technology scope (KI-46): an advisory overjustification guardrail plus a CRISP gamification design note.

Changes

  • services/analytics/reward_calibration.py: new detect_overjustification_risk(conn, location_id) read-only diagnostic. Inspects token_reward_distribution.distribution_method and metric-reward correlation to flag where extrinsic token rewards may crowd out intrinsic motivation. Returns risk_level low/medium/high/unknown, signals, and recommendations. Never writes governed state.
  • services/export/report_generator.py: generate_reward_calibration now embeds overjustification_risk with an advisory-only limitation note.
  • services/crisp/models.py + scoring_engine.py: CompositeRating.design_note frames the AAA-D gamified band as a competence/merit signal, not a bribe.
  • AGENTS.md: Persuasive Technology constructive-only boundary note (rejects Fogg's destroy/deceived/locked-in; token rewards recognize verified competence, never coerce; routes through DRAFT-only sinks).
  • tests/test_overjustification.py: 6 passing tests (risk tiers, report surfacing, no-write guard).

Boundary

Advisory only. No persuasive feature writes governed state autonomously; proposals still route through the DRAFT-only sinks. The opposing force in any simulation remains shocks/threats/pests/market/competitor data — never stakeholder fragmentation.

Commits were merged into target branch
1/1
Submitter wasabi
Target main
Source feat/persuasive-overjustification
Jobs
Merge Strategy
Create Merge Commit
Watchers (1)
Reference
pull request KI-46
Please wait...
Connection lost or session expired, reload to recover
Page is in error, reload to recover