Reliability is a new science: we are on the right way

Reliability has long been treated as an engineering practice supported by testing, statistics and standards, yet its status as a scientific discipline remains unsettled. From the perspective of the philosophy of scientific truth, a science is characterized by a dual structure that unifies empirical truth and mathematical truth under an axiomatic system. The empirical truth is established through universally repeatable controlled experiments, and the mathematical truth is expressed through isomorphic symbolic representations. Building on this criterion, this paper examines whether reliability can be regarded as a science in the strict sense of scientific truth. We first clarify the scientific object of reliability by proposing a new definition: reliability is defined as the repeatability of system function across time and space. This definition shifts reliability from an experience-based engineering concern to a more explicitly defined theoretical object. On this basis, a coherent framework of reliability science is constructed. Reliability principles are organized as an axiomatic system centered on margin, degradation and uncertainty; reliability experiments are formulated as controlled and repeatable processes designed to verify the chance-causal relations implied by these axioms, with a hierarchical structure ranging from controlled observation to controlled experiment; and a corresponding mathematical framework is developed to express reliability laws through distance, relation and change. This scientific structure further reveals the shortcomings of current mathematical tools in dealing with time-evolving uncertainty, leading to the natural development of an axiomatic extension, called Biandong Statistics, which arises as a derivative outcome of the proposed framework. The validity of this framework is further illustrated by its applicability across engineering, living and social systems, demonstrating that reliability is a general system-level property wherever meaningful system functions exist. By integrating axioms, experiments, mathematical representation, and cross-domain applicability into a unified structure, this work provides a philosophical and structural basis for understanding reliability as a coherent scientific discipline, in which empirical and mathematical components jointly realize the dual structure of scientific truth and outlines the overall disciplinary landscape of reliability science.