Errors and Failures: Towards a New Safety Paradigm

In France studies on technological risks began to question errors, failures and vulnerabilities at the end of the 1970s, focusing mostly on analyzing major accidents as consequences of the increasing complexity of socio-technical systems. During the 1980s and 1990s, research studies carried out in different fields (industrial risks, natural risks, health risks) underlined the importance of organizational factors in system vulnerabilities. Still, the bases of safety policies and safety management remained unchanged, with a strong reliance on rules and procedures. Building on an interdisciplinary reflection carried out at the beginning of the 2000s, this paper calls into question the prevailing approach as regards safety. Identifying the basic assumptions behind safety policies, it is argued that, in light of research advances in various fields of safety studies -- and more specifically in cognitive ergonomics -- they appear to be basically flawed. In a quite radical manner, a recognition of errors and failures as a part of the usual functioning of socio-technical systems, which are "naturally" unstable systems, is called for. As for risk control, it appears to result mainly from the capacity of operators, working groups and organisations for dynamically "making up" for errors and failures. These analyses open very stimulating prospects of research. However, the question of their social and political acceptability must be seriously considered.