Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!henry
From: henry@utzoo.UUCP (Henry Spencer)
Newsgroups: can.politics
Subject: Re: Some problems with Star Wars #2 (peripheral issues)
Message-ID: <5726@utzoo.UUCP>
Date: Mon, 24-Jun-85 17:00:27 EDT
Article-I.D.: utzoo.5726
Posted: Mon Jun 24 17:00:27 1985
Date-Received: Mon, 24-Jun-85 17:00:27 EDT
References: <1197@utcsri.UUCP>
Organization: U of Toronto Zoology
Lines: 42

I'm going to make a longer response to the main content of Ric Hehner's
paper, but I thought I'd get a couple of peripheral things out of the way
first, separately.

> ...Ric and Andrew feel ... that debating on Usenet accomplishes little.

I disagree, but not for the obvious reasons.  It is quite possible that
there is little chance of the debaters changing their views as a result
of debate.  What such discussion/debate can, and sometimes does, accomplish
is to inform and educate the spectators.  This is not trivial.

> ... When airplane navigation software fails,  a
> planeload  of  people  may die.  This has happened more than
> once; apparently the airlines consider it  tolerable. ...

I would like to see some references on this.  If Ric is thinking of the
Mount Erebus and KAL007 crashes, then he has weakened his argument with
a serious error of fact.  Neither of these crashes was a software failure
(modulo some small uncertainty about KAL007).

The Mount Erebus crash is definitely known to have been human error.  Some
navigation reference points were changed, and the change was not propagated
to all the people who should have been told.  As a result, correctly-working
navigation software flew the plane into a mountain that the software thought
was well off to one side.  Bad weather made it impossible for the pilots to
detect this in time.  There is no suspicion of hardware or software failure.

KAL007 is a bit less certain, because of lack of evidence, but the most
plausible theory is a subtly-incorrect order given by the crew to the
navigation software.  (Ironically, slightly "smarter" software might have
saved KAL007.  The navigation software had to be given explicit instructions
for each leg of a complex course, and had no overall knowledge of the course
or the constraints on it.  Hence the software was unable to apply any real
consistency checks to the orders given to it.)  As far as I know there is
no suspicion of hardware or software failure.

Finally I doubt that the airlines -- or the passengers, or the crews, or
the government regulatory agencies -- consider fatal crashes "tolerable"
regardless of their cause!
-- 
				Henry Spencer @ U of Toronto Zoology
				{allegra,ihnp4,linus,decvax}!utzoo!henry