Jerome H. Fine wrote:
<snip>
Jerome Fine replies:
I thought that a timely response will be helpful to clarify
any possible confusion.
The software was custom grown PDP-11 software from the AECL company.
The problem was that the instrument could deliver a massive 25,000 rad
blast over a 1 second interval due to the fact that the beam control
mask was incorrectly positioned.
A collimator with adjustible slits, bending magnets, and a blocker to
safe the machine was how the basic machine set the dosage.
Cryptic undocumented error messages, frequent malfunctions that had to
be ignored in normal operation, and incorrect instructions to operators
caused the malfunction to be unrecongnized.
This was analyzed in great depth as it was the first computer caused
malfunction that could be analyzed to study the responses of the people
involved in the use of the machine.
It is very educational to read this from time to time, when you think
you have though of every systematic thing you should consider when
designing a system.
The software basically got the dosage setting displayed to the operator
out of sync with the actual, and eventually you would deliver a zap of
electron beam energy that felt like an electrical shock, plus hot coffee
on your skin to a patient.
Jim