Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!usc!orion.oac.uci.edu!uci-ics!nancy
From: nancy@ics.uci.edu (Nancy Leveson)
Newsgroups: comp.software-eng
Subject: Re: Some Issues Lying Beneath a Recent
Message-ID: <25EC1A26.8067@paris.ics.uci.edu>
Date: 28 Feb 90 18:36:22 GMT
References: <3241@umn-d-ub.D.UMN.EDU> <102100003@p.cs.uiuc.edu>
Reply-To: nancy@ics.uci.edu (Nancy Leveson)
Organization: UC Irvine Department of ICS
Lines: 37

I have been an expert witness in two law suits involving the Therac
25 accident alluded to below.  Unfortunately, there is a lot of
misinformation about this accident that is circulating.  I have
figured out how to get around the proprietary information aspects
and am preparing a detailed paper that should clear up the misconceptions.
Until then, let me just clear up a couple of things to avoid further
circulation of misinformation:

In article <102100003@p.cs.uiuc.edu> johnson@p.cs.uiuc.edu writes:
>
>For example, a couple of years ago there were some people who were
>killed because an X-ray machine emitted much more power than it
>should have.

The correct radiation was released, but it was released without the
proper filter being in place to reduce the amount of radiation that
reached the patient.

>The problem was in supposedly in the user interface;

This is wrong (along with everything in the article that follows this
statement).  The problem (actually two problems found so far) was in
the code.  I think this wrong information originally appeared in an
article about this incident that appeared in Datamation, and it has been
repeated widely.  Although the user interface was far from desirable, it
did not cause the accident -- poor programming did.

My paper describing the incident details will be available in a month or
so, and I will try to get it published quickly -- I have not decided where
to send it yet.  In the meantime, please be careful when repeating rumors
about the Therac accident.  So far, everything I have seen written about it
has been completely wrong.

nancy

--
Nancy Leveson