Rédiger un PREreview

This preprint presents a conceptual framework for understanding failure phenomena that recur across biological, social, and artificial systems.

The paper proposes Constraint-Induced Failure Theory, whose central proposition is that failure should be understood not as an intrinsic property of a particular system but as a structural phenomenon induced by common constraints acting on self-maintaining systems.

To develop this framework, the study introduces the concept of self-maintaining systems, identifies six fundamental self-maintaining functions, proposes a common constraint space shared across different implementation domains, formulates the Failure Mode Generation Principle, and presents the Life-History Observation Hypothesis as a comparative perspective on the accumulation of failure phenomena throughout evolutionary history.

Rather than claiming that biological, social, and artificial systems are identical, this study provides a comparative theoretical framework for describing recurrent failure structures across different domains. The objective is to establish a common conceptual language connecting cybernetics, general systems theory, complexity science, and AI safety research.

This work is a conceptual study rather than an empirical investigation.

This preprint is the public Version 2.0 of the manuscript. It is distinct from the anonymized manuscript submitted for peer review to an international academic journal.