Hypothetical

Jeremy jepri at webone.com.au
Thu Oct 31 17:56:49 EST 2002


>I seem to recall a text book case for Software Engineering where a software

>failure in a micro-controller controlling an X-Ray machine caused the death

>of several patients and the engineers or similar responsible for the code 

>were found to be negligent.

Ironically, that exact case is covered on the same site

http://onlineethics.org/cases/therac25.html

I notice a common thread through most of these stories, which is that the ultimate
failure is blamed on either poor testing or the deliberate perversion of the
testing process.

But we do worse in the F/OSS world - we proudly declare we do little testing
for ourselves.  Our users are our testers (look ma, I'm on topic!).  When a
commercial software package goes wrong we think "What bonehead QA allowed this
code to be released?".  When F/OSS goes wrong, we just say "Mail the author"
or "What did you expect, using version 0.05?".

And yet, strangely, I'd probably feel happier being treated by a computerised
machine running F/OSS software.  Possibly because I know a few coders who go
to work bent fairly often.  They're good coders, but I don't know if their companies
have enough QA.




More information about the linux mailing list