|
Apollo Guidance Computer History Project
Second conference
September 14, 2001
Trust in Automatic Systems
JIM MILLER: You asked about trust in automatic systems. I remember just a short
anecdote: We were visited at the Laboratory one day by John Glenn and Alan Shepard and I
think a third astronaut. It was at the time when reporters were still chasing around after
the seven astronauts - it was a big thing when the astronauts were coming. The Laboratory
was showing those folks what we were doing.
And one of the things that, at the time, we had the arrogance to believe we should do
was to make sure that nothing could be done by the crew that would endanger them. So there
were a lot of tests in the software that said, if the delta V is in the wrong direction or
this or that, then they couldn't execute the proceed order to go ahead because the
software thought something was wrong. I remember two things that Shepard, with his
marvelous way with words, said. On this one he said, "You guys think you know more
about what is going to face us during flight than we will know during the flight. You're
wrong." He said, "Take all of those lockouts out. If we want to kill ourselves,
let us. We may be saving ourselves." Of course he was right. And I think everybody
sort of woke up to that. That made the software a lot easier too (laughter) because we did
not have to worry about all those checks.
DAN LICKLY: It did let him start the prelaunch program on the back from the moon.
JIM MILLER: They weren't capable of perfection. But they were pretty well trained,
and they certainly were motivated.
DAN LICKLY: They were very well trained. I will say that. But they also got very
tired. I had the re-entry which is at the end of five days out there. And they swore
through all the early years that they were going to fly it manually. As far as I know,
none of them ever touched a manual stick. They were so beat. And then all of a sudden you
hit 7 Gs. I'm not sure they could even see what was going on by then. They didn't get any
sleep out there for all that time. I'm surprised they were doing as well as they did after
all that time.
MARGARET HAMILTON: We had a different opinion of this subject matter than the
astronauts did. At least the software group did at the time this took place, the
prelaunch. We talked about that the last time, in our last meeting -- when the astronauts
selected prelaunch during flight. So I think there's sort of an in-between, you don't want
to overdo it but you also want to make the crucial areas safe so you won't have a
catastrophe.
JIM MILLER: Shepard, in that meeting, was responsible, I think, for the demise of
another thought that we had had, which was since the computer was not being perfectly
reliable - it was going to be necessary to do in-flight repairs. So there were a lot of
thoughts about diagnosing what had gone wrong, changing AGC modules and so forth. I don't
know how far along that got. But Shepard's comment on that was, "yeah, and we should
all train to be brain surgeons so we can operate on each other." I think that was the
end of in-flight repair.
Introduction of project management methods
site last updated 12-08-2002 by Alexander Brown |
|