Proof-of-Mechanism Simulation Package for Implementing Affective Sovereignty

Proof-of-Mechanism Simulation Package for Implementing Affective Sovereignty This repository accompanies the paper: Ryan Sangbaek Kim (2025) Implementing Affective Sovereignty: Formal Foundations, Design Architecture, and Computational Governance DOI: 10.5281/zenodo.17154182 📊 Overview This package reproduces the proof-of-mechanism (PoC) simulation described in the paper. The goal is to demonstrate how sovereignty-aware design—via DRIFT protocols and Affective Sovereignty Contracts (ASC)—improves interpretability, accountability, and alignment in affective AI systems. The simulation compares three system conditions: Baseline (Static AI) — no abstention, no learning DRIFT (Dynamic) — uncertainty-aware selective prediction + on-the-fly personalization DRIFT + Policy — DRIFT + user policy (e.g., suppressing Neutral commentary) Metrics included: IOS (Interpretive Override Score) AMR (After-correction Misalignment Rate) AD (Affective Divergence, KL divergence) Abstention rate Key results: DRIFT + ASC reduces IOS, AMR, and AD significantly, while introducing a modest, tunable abstention cost—demonstrating the feasibility of sovereignty-by-design. 📂 Package Contents src/ — simulation scripts (simulation.py, metrics.py) data/ — synthetic interaction logs (CSV) results/ — aggregated metrics (metrics_summary.csv) figures/ — generated plots (frontier.png) notebooks/ — Jupyter notebook (reproduce_results.ipynb) requirements.txt — Python dependencies Dockerfile — reproducible container environment README.md — usage instructions LICENSE — MIT License for code 🚀 Reproducibility All experiments use fixed random seeds for deterministic output. Synthetic interaction logs and results can be regenerated via simulation.py. Full containerized reproducibility supported via Docker. Code licensed under MIT, data and documentation under CC-BY 4.0. 📌 Citation If you use this package, please cite both the paper and the dataset/code: Paper Kim, R. S. (2025). Implementing Affective Sovereignty: Formal Foundations, Design Architecture, and Computational Governance. Zenodo. https://doi.org/10.5281/zenodo.17154182 Dataset/Code Kim, R. S. (2025). Proof-of-Mechanism Simulation Package for Implementing Affective Sovereignty [Code & Data]. Zenodo. https://doi.org/10.5281/zenodo.17153891