This paper completes the retentive systems framework by analyzing failure dynamics in distributed systems. Building on the concepts of retentive regimes and threshold-based transitions, collapse is defined as the reversal of the retentive condition when disruption exceeds retention. The work identifies distinct failure modes, including critical overload, retention decay, delayed collapse, and false retention, and introduces the notion of pre-collapse indicators that allow early detection of instability. The contribution establishes collapse as a non-random, threshold-driven process and defines the full cycle of system behavior: emergence, stabilization, and breakdown.
Logacheva Yulia (Fri,) studied this question.