home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky gnu.misc.discuss:2965 rec.photo:15835
- Newsgroups: gnu.misc.discuss,rec.photo
- Path: sparky!uunet!uunet.ca!cognos!mcgregoa
- From: mcgregoa@cognos.com (Alayne McGregor)
- Subject: Re: Time to boycott Photo CD
- Message-ID: <1992Sep9.164601.735@cognos.com>
- Organization: Cognos Incorporated, Ottawa CANADA
- References: <glcnc4@dixie.com> <JIMB.92Sep4013256@occs.cs.oberlin.edu> <BURLEY.92Sep5160246@geech.gnu.ai.mit.edu>
- Date: Wed, 9 Sep 1992 16:46:01 GMT
- Lines: 40
-
- In article <BURLEY.92Sep5160246@geech.gnu.ai.mit.edu> burley@geech.gnu.ai.mit.edu (Craig Burley) writes:
- >Also, the more publicly understood the concept being distributed, the more
- >able it is to flexibly adapt to varying needs, and the less likely it
- >is to contain dangerous bugs. (Think of that Xray machine or whatever that
- >turned out to have a software bug that killed several people -- if that
- >code had been published as free software several months _before_ the
- >machine was released, along with an offer of a total of, say, $100K reward
- >money to all those who find any serious bugs in the code prior to the machine's
- >release, would those victims be alive today? I think so! Would the machine
- >have ever been made under such circumstances? It's remotely possible it
- >wouldn't have, but then those people would be alive just the same. It's
- >very reasonable to consider legislating that any software that controls or
- >monitors any equipment that is actively capable of killing or seriously
- >injuring people must be published as PD, GPL, or similar software along with
- >announcements asking the public to examine the software for life-threatening
- >bugs. That oughtn't to hurt companies, since they're primarily making
- >sophisticated hardware with simple controlling software, and unfortunately
- >hiring simple software programmers to do the latter task. :-)
-
- Faulty analogy.
-
- The problem with the Therac-25 radiation (not X-ray) machine was, from what I've
- read, both a hardware and software problem. The software malfunctioned in
- not correctly calculating the intensity of the radiation in certain unusual cases;
- the hardware interlocks that should have prevented this level of radiation
- from reaching the patient did not react fast enough.
-
- Furthermore, the problem only showed up if an operator entered an incorrect value
- and corrected that value within a few seconds. This kind of timing problem is
- *not* easy to find just by reading code.
-
- Furthermore, while I certainly might be interested in the quality of
- software controlling my local nuclear power plant (for example), I doubt
- that I or many other people have the 6 months free to audit that code.
- Be realistic!
-
- Alayne McGregor
- mcgregoa@cognos.com
- alayne@ve3pak.ocunix.on.ca
-
-