NetNews Usenet Archive 1992 #31/NN_1992_31.iso/spool/sci/med/22969

Internet Message Format | 1992-12-22 | 4.7 KB

Path: sparky!uunet!cs.utexas.edu!news-is-not-mail
From: [e-mail] (Russell Turpin)
Newsgroups: sci.med
Subject: Reliability in medicine and engineering  (was: ... incomes)
Date: 22 Dec 1992 12:51:21 -0600
Organization: CS Dept, University of Texas at Austin
Lines: 84
Message-ID: <[e-mail]>
References: <[e-mail]>
NNTP-Posting-Host: im4u.cs.utexas.edu
Summary: And some comments about capitalism and medicine.

-*----
In article <[e-mail]>, [e-mail] (Carl Wuebker) writes:
>> ... When I make a mistake, the consequence is usually time
>> (if I'm lucky, it doesn't cost much & takes a few hours to
>> fix).  If, on the other hand, a physician makes a mistake,
>> it can cost a life (or perhaps the quality of life) for
>> somone. ...

In article <[e-mail]> [e-mail] (Maurice LaBiche) writes:
> Maybe your not as detailed a programmer as many others.  Depending
> on the type of system you work on/develop you have to handle real
> world system faults which could shut down the computer systems
> that control the electrical power grids or the 911/telephone service
> in your town.  These are live systems relied on by the masses in
> normal as well as emergency situations.  If these systems go out many
> lives can be lost.

Maurice LaBiche propagates a common, but wrong, perception.
Software for which extremely high reliability is a requirement,
such as in the control systems for an airliner, is *not*
developed by normal programming techniques made better by better
people, i.e., by hiring someone who is more "detailed a
programmer." Such reliability is achieved instead by a
qualitatively different engineering process in which the focus is
on redundancy in both the process and the design.  High
reliability often requires redundancy between hardware and
software, so that the hardware contains checks against software
errors and vice versa.

In this kind of engineering, it is always assumed that individual
programmers will make mistakes, and that a program will contain
bugs even after the programmer extensively tests it.  A process
and design is used that achieves reliability despite this fact.
Consider, for example, the softare on the space shuttle, which
was developed five ways by five different teams, using five
different compilers, which runs on five different computers,
whose decisions are arbitrated by a concensus mechanism.

No life should be put at risk because a programmer makes a
programming mistake.  Any company that develops software in such
a manner commits gross negligence.  (And, yes, it will be easy to
fill the courtroom with experts who will testify to this.)  The
programmer is likely not at fault, because most programmers are
not learned in reliability engineering.  Rather, the engineering
managers and the company officers are at fault for deploying a
inadequate engineering process to build a system in which high
reliability is required.

Carl Wuebker is correct that a programming mistake (or even
programming incompetence) costs only time and money, even when
the system under construction must function at a high level of
reliability. Checks are built into the group engineering process
to counter individual flubs.

Medicine, unfortunately, involves some processes where individual
human error can result in irrecoverable and tragic loss, and
where checks against this are only indirect.  The two examples
that come immediately to mind are surgical error and emergency
treatment error.  There are indirect checks, in the sense of
rigorous training, strict licensing, and professional accounting,
but these do not help at the time an error is made nor do they
provide a way to recover from the results.

There are more direct checks against individual error in other
parts of medical practice.  For example, an administering nurse
or pharmacist can stop a wildly wrong prescription before the
patient receives it.  I have the sense that tradition has impeded
the development of more checks that would be useful, but I would
rather have medical professionals comment on this.

> Capitilism and Medicine.  Live the reality.  I do.

Capitalism, in the sense of laissez faire markets, has next to
nothing to do with medicine here in America.  The medical
professions are tightly regulated, the income tax code stupidly
links medical financing to employment, insurance policies are
dictated by state boards controlled by large insurers, hospital
and clinic construction are state controlled, etc. ad nauseum.
Anyone who points to American medical care as an example of
free enterprise is living in fantasy land.  We already have a
state controlled medical industry.  Other countries merely have
*different* state controlled medical industries.

Russell