"Simulation Data for Feedback-Controlled Intervention in a Model of Information Cocoons"
"This dataset contains synthetic simulation outputs supporting the manuscript \u201cFeedback-Controlled Intervention in a Model of Information Cocoons: Concentration, Affinity, and Diffusion Trade-offs\u201d by Yun Tao and Zongwei Luo. It covers two targeted studies: an original-model intervention comparison with 80 recipient runs, and a calibrated utility-frontier extension with 214 recipient runs, including 104 calibration runs and 110 held-out evaluation runs. Each run uses 500 synthetic agents, 18 topics and 2,000 update steps. The archive includes per-run CSV summaries, JSON trajectories, model parameters, calibration selections, four independent donor action arrays, a data dictionary and SHA-256 integrity records. Recorded diagnostics include preference concentration, expected and sampled affinity, distinct-topic utility, unique-topic coverage, modeled attention and diffusion dose. These are model-level diagnostics rather than empirical measurements of human wellbeing or item-ranking performance. The release supports inspection of feedback, static and open-loop intervention trade-offs; it does not establish general superiority of adaptive control. It excludes third-party MovieLens and Last.fm user records and is not a standalone simulator distribution or a complete archive of every historical experiment in the manuscript."