3 ms·
No news here. This 21-year-old college report [0]from 1998 (Porrello, cites include 1993 Leveson, Nancy G., and Clark S. Turner.) has -a lot- more facts without
by 8bitsrule 7y ago
No news here. This 21-year-old college report [0]from 1998 (Porrello, cites include 1993 Leveson, Nancy G., and Clark S. Turner.) has -a lot- more facts without the hysteria. It states:
"<b>One</b> programmer, over several years, revised the Therac-6 software into the Therac-25 software (AECL has not released any information about the programmer or his credentials)."
[0] https://web.archive.org/web/19980201101244/http://cobra.csc.calpoly.edu/~dbutler/papers/THERAC25.html https://web.archive.org/web/19980201101244/http://cobra.csc.... - "Death and Denial: The Failure of the THERAC-25, A Medical Linear Accelerator"
Previous Therac models had hardware interlocks to prevent some modes; they were removed in favor of software for the 25. No doubt there were some engineers who knew more about this.
- C1sc0cat 7y agoThis is amazing - I would have thought that any medical equipment would go through exhaustive testing and an order of magnitude more for systems involving radiation and a machine capable of lethal doses. At my first job we even had a separate safety officer for our low powered sources used for tracing waterflow.
- eigenvalue 7y agoPart of the problem is likely that the AECL is a quasi-public entity, and governments tends not to be nearly as vigilant when regulating or policing other branches of the same government as they are when regulating a 100% privately owned company.
- C1sc0cat 7y agoYes I saw that relying of crown immunity
- cameldrv 7y agoA lot of the regulations around safety critical software came about because of the Therac-25 accidents.