An Investigation of the Therac-25 Accidents
📜 Abstract
Between June 1985 and January 1987, six known accidents involved massive overdoses by the Therac-25 -- with resultant deaths and serious injuries. Analysis of the events indicated that the accidents were caused by a combination of factors, including programming errors not caught by typical software testing, lack of user training and understanding, incorrect system configurations, and inadequate communication and response actions among product users, manufacturers, and regulators. This paper identifies lessons learned for preventing similar accidents in future software-controlled systems, noting the need for greater diligence in software engineering, regulatory processes, and the operational environments of such systems.
✨ Summary
The paper “An Investigation of the Therac-25 Accidents” by Nancy Leveson and Clark S. Turner discusses the series of accidents involving the Therac-25 radiation therapy machine, which led to multiple fatalities and serious injuries during the mid-1980s. It identifies the causes of these incidents, primarily software errors, insufficient safety measures, and lack of effective communication among stakeholders. The paper is widely recognized in discussions about software safety and has had significant influence in the field of safety-critical systems, especially emphasizing the importance of thorough software testing, robust safety engineering practices, and reliable regulatory processes.
The insights from this paper have contributed towards improved safety standards and regulations in the medical device industry, shedding light on the crucial role of understanding and preventing software-related failures. According to an article on IEEE [https://ieeexplore.ieee.org/document/5991657], it stands as a seminal work influencing policy changes and awareness around managing the complexities of software interactions in life-critical systems. Additionally, it serves as a case study extensively referenced in software engineering curricula and safety engineering research, highlighting learnings for preventing similar catastrophic failures in the future.
Despite comprehensive safety testing procedures today, echoes of the Therac-25 case continue to resonate in discussions and analyses of both historical and current software failure incidents in safety-critical domains.