AECL never publicly released the source code, but several experts including [Nancy Leveson] did obtain access for the investigation. What they found was shocking. The software appeared to have been written by a programmer with little experience coding for real-time systems. There were few comments, and no proof that any timing analysis had been performed. According to AECL, a single programmer had written the software based upon the Therac-6 and 20 code. However, this programmer no longer worked for the company, and could not be found.
The problem wasn't the programmer, it was the ones who hired them and allowed them to work without any safety policy being applied for both the realtime and radiation aspects.
Comments
Poorly designed checkbox that killed 10 sailors and cost >200 million USD:
https://news.ycombinator.com/item?id=41297563
Also see: "Safeware: System Safety and Computers"
and the work of the author, Nancy Leveson:
http://sunnyday.mit.edu/
Yet another article on the Therac-25 radiation therapy machine dose bug scandal from the 1980s. 62 previous posts:
https://hn.algolia.com/?q=therac-25
The problem wasn't the programmer, it was the ones who hired them and allowed them to work without any safety policy being applied for both the realtime and radiation aspects.