{"caveats":["This is a RECOMMENDATION. It is not wired into probe_module.py or the live probe scheduler \u2014 nothing changes the probe cadence yet. It produces a priority list a scheduler or human can choose to consume.","Thompson sampling here treats each (country, domain) pair as an INDEPENDENT bandit arm. In reality DPI policy correlates domains within a country and censors flip many sites at once \u2014 a hierarchical / contextual model would be more correct. Treat the ranking as a heuristic.","A high-variance pair may just be intermittently REACHABLE (flaky resolver, congested transit) rather than censorship-uncertain. The recent-flip term partly separates genuinely-contested pairs from merely-noisy ones, but not perfectly.","Cold pairs with fewer than 2 observations are not ranked here; the existing fixed-list scheduler already covers them.","Recency weighting uses a 45-day half-life, so a block state that flipped months ago is heavily discounted relative to fresh observations."],"count":1,"filters":{"country":null,"domain":null,"limit":50,"min_observations":null,"recent_flip":true},"generated_at":"2026-09-21T04:50:01.585651+00:00","is_recommendation_only":true,"priority_formula":"posterior_variance * recency_weight * flip_weight * (1 + thompson_weight * (1 - |sample-0.5|*2))","results":[{"beta_alpha":1.522,"beta_beta":2.403,"country":"AZ","days_since_last_observation":43.2,"domain":"chat.openai.com","flip_weight":1.5,"last_blocked_at":"2026-08-09T00:00:00+00:00","last_observed_at":"2026-08-09T00:00:00+00:00","last_unblocked_at":"2026-08-08T00:00:00+00:00","n_blocked":6,"n_observations":23,"n_unblocked":17,"posterior_mean_block_prob":0.3879,"posterior_variance":0.048205,"priority_score":0.186731,"rank":125,"recency_weight":1.96,"recent_flip":true,"sources":["censoredplanet","ooni"],"thompson_sample":0.3176}],"schema":"voidly-atlas-probe-scheduling-priority/v1","total_pairs_scored":2966,"total_pairs_with_recent_flip":254,"wired_into_scheduler":false}