{"id":"membership-detection-unreliable-in-accuracy-and-propagation","text":"The membership subsystem is unreliable at two independent levels that interact destructively: failure detection is both the single correctness bottleneck AND permanently miscalibrated (no arrival-rate history for adaptive thresholds), while membership and data convergence operate at fundamentally different rates (O(log N) membership gossip vs O(N) data propagation in ring topology), meaning incorrect liveness decisions propagate through membership faster than data can adjust to them.","truth_value":"IN","source":"","source_url":"","source_hash":"","justifications":[{"type":"SL","antecedents":["gossip-failure-detection-lacks-adaptive-accuracy","convergence-rate-asymmetry-membership-vs-data"],"outlist":[],"label":"Miscalibrated detection decisions propagate at O(log N) while data convergence lags at O(N), so the system acts on stale liveness information with compounding effect."}],"dependents":["distributed-cluster-lacks-reliable-infrastructure"],"metadata":{"last_reviewed":"2026-05-30T09:01:17","review_result":"pass"},"created_at":"","updated_at":"","reviewed_at":"","verified_at":"","retracted_at":"","explanation":{"steps":[{"node":"membership-detection-unreliable-in-accuracy-and-propagation","truth_value":"IN","reason":"SL justification valid","antecedents":["gossip-failure-detection-lacks-adaptive-accuracy","convergence-rate-asymmetry-membership-vs-data"],"label":"Miscalibrated detection decisions propagate at O(log N) while data convergence lags at O(N), so the system acts on stale liveness information with compounding effect."},{"node":"gossip-failure-detection-lacks-adaptive-accuracy","truth_value":"IN","reason":"SL justification valid","antecedents":["gossip-failure-detection-governs-cluster-correctness","gossip-no-arrival-history"],"label":"The correctness bottleneck permanently lacks the data needed for adaptive accuracy"},{"node":"gossip-failure-detection-governs-cluster-correctness","truth_value":"IN","reason":"SL justification valid","antecedents":["gossip-failure-detection-gates-replication","gossip-suspicion-is-timeout-based","gossip-cleanup-bounds-membership-growth"],"label":"All distributed data operations flow through the gossip liveness set, making failure detection accuracy a cluster-wide concern"},{"node":"gossip-failure-detection-gates-replication","truth_value":"IN","reason":"premise"},{"node":"gossip-suspicion-is-timeout-based","truth_value":"IN","reason":"premise"},{"node":"gossip-cleanup-bounds-membership-growth","truth_value":"IN","reason":"premise"},{"node":"gossip-no-arrival-history","truth_value":"IN","reason":"premise"},{"node":"convergence-rate-asymmetry-membership-vs-data","truth_value":"IN","reason":"SL justification valid","antecedents":["gossip-convergence-is-olog-n","ring-topology-convergence-is-linear-in-node-count"],"label":"O(log N) membership convergence vs O(N) data convergence creates a growing window where the cluster knows who's alive but hasn't finished replicating — the gap widens with cluster size"},{"node":"gossip-convergence-is-olog-n","truth_value":"IN","reason":"premise"},{"node":"ring-topology-convergence-is-linear-in-node-count","truth_value":"IN","reason":"SL justification valid","antecedents":["ring-requires-n-minus-one-rounds","ring-topology-needs-multiple-sync-rounds","ring-requeues-via-apply-remote-change"],"label":"Three observations pin convergence delay to the store-and-forward mechanism: apply-remote-change requeues to pending, and pending drains one hop per sync"},{"node":"ring-requires-n-minus-one-rounds","truth_value":"IN","reason":"premise"},{"node":"ring-topology-needs-multiple-sync-rounds","truth_value":"IN","reason":"premise"},{"node":"ring-requeues-via-apply-remote-change","truth_value":"IN","reason":"premise"}]}}