Killed by a Machine: The Therac-25

Re:Killed by a Machine: The Therac-25 by jes on Monday August 01, 2016 @02:05PM (#12204540)

I don't remember the details at this point, but somewhere in the IEC 62304 regulation, there is language that suggests that you have to assume 100% failure for software components.

I think it's in regard to risk controls, but I'm not certain of that.

Edit: Just looked it up: "If the HAZARD could arise from the failure of the SOFTWARE SYSTEM to behave as specified, the probability of such failure shall be assumed to be 100%"

[ Parent | on HN ]

Re:Killed by a Machine: The Therac-25 by HeyLaughingBoy on Monday August 01, 2016 @02:58PM (#12204966)

Yes, this is to avoid the problem many an FMEA (Failure Modes Effects Analysis) meeting agonizes over: "it's possible for the software to fail in this way, but how do we quantify the probability of it happening so we can calculate the risk factor?" By assuming that if it can happen, then it will" you remove any handwaving from the equation.

[ Parent | on HN ]

Re:Killed by a Machine: The Therac-25 by jes on Wednesday August 03, 2016 @12:53PM (#12219414)

I think the language has the side-effect that you describe, and I think that it is beneficial, but I don't know if that is the reason that the language was added to the IEC-62304 regulation.

Thank you for a helpful observation. It had not occurred to me.

[ Parent | on HN ]