microcode, compilers, and supercomputer architecture

From: cem14_at_cornell.edu <(cem14_at_cornell.edu)>
Date: Tue Apr 6 12:23:23 1999

Max Eskin <max82_at_surfree.com> wrote:
>
> I have heard of Airbuses crashing because the controls were poorly
> designed. I've never heard of a 777 crashing. I've also never heard of any
> medical machines failing, though this would tend to be underpublicized.

Oh yes, my brother-in-law flies A-300's and told me about this.
Apparently, this was the first all-electronic cockpit model. When
you are about to land, there is a moment in which you have to
decide whether you want to abort the landing or not (for whatever reason).
After that, you have to land no matter what. At this point,
the console in the plane in question asked the pilot whether
he wanted to continue the landing or abort it. The thing is,
the person who programmed this was a native french speaker,
and apparently the question contained a double negation. Most
pilots would choose the right answer, but this one was choosing
to abort the landing when he meant to continue the landing procedure;
the computer then would perform an overpass of the airport,
without the pilot knowing why. They repeated the attempt to land
several times, failing everytime because of the same reason.
Also, the procedure was designed so that you could not simply
take the computer offline and do everything manual.
Eventually, they ran out of fuel. A classical case of
bad operator interface design, with grave consequences.

Carlos.
Received on Tue Apr 06 1999 - 12:23:23 BST