Beagle ... alas

In article ,
Vincent D. DeSimone wrote:
(And it's not like slower/worse/costlier has a conspicuously better track
record, especially at Mars...)

I agree with your response, but there have been too many examples brought up
in this newsgroup, as well as the news feeds, that FBC is just plain flawed.

Which are those? Mars Pathfinder? Mars Global Surveyor? Clementine?
Lunar Prospector? Mars Odyssey? NEAR? Chipsat? MOST? Mars Express?

My belief that the opinion voiced earlier this year that you can get get two
of these options by only sacrificing the third, is the way to go. It was
called "FBC: Pick 2". I like to rhyme it by saying "FBC: 2 Out Of 3".

That's certainly the party line among the dinosaurs of the space business.
And for *them*, it's true: you cannot get a mammal by putting a dinosaur
on a starvation diet.

The correct statement is "faster, better, cheaper, same old management:
pick any three". Naturally, the same-old-management people like to shorten
that, on the assumption that there will never be a change in management.

To make FBC work, you have to do things *differently*. As rk perceptively
observed, "It's not the slogan, it's the execution." Proper execution is
almost impossible to do if the Same Old Management is in charge. You need
to build a new (sub-)organization, insulated from the failings of the old
one. Just chanting "faster, better, cheaper" every day, while doing the
same old things, is not enough.

(A big factor in the success of Mars Pathfinder and Sojourner was that the
Old Guard at JPL were convinced the mission would fail, so they stayed
away from it. A big factor in the failure of Mars Climate Observer, and
to a lesser extent in that of Mars Polar Lander, was that the Old Guard
enthusiastically climbed on board after the spectacular success of Mars
Pathfinder. NASA does FBC right only by accident.)

... _Complete_ testing of hardware and software should always be
considered an unavoidable overhead cost that is figured into the "C" portion
of the equation...

There is no such thing as "_complete_" testing. It is *always* necessary
to eventually call a halt to testing and fly the thing. Pretending
otherwise is dangerous self-delusion, which prevents discussions of the
tradeoffs and thus largely prevents rational decision-making about them.

There is always a balance between expenditure and risk, which will be
chosen differently for different missions. And never forget that there is
always some risk of surprises, of being blindsided from an area you judged
unimportant, so spending lots of money trying to drive risk to zero is
foolish. Once you have reduced known risks to a certain point, the
unknown risks dominate the problem, and further spending on reduction of
known risks buys almost no real improvement in mission reliability.

It's also important to remember that testing itself is not foolproof.
Galileo's atmosphere probe was tested in a centrifuge to ensure that its
G-switches worked. The test results were fine. But at Jupiter, the 20G
switch came on first, and the 5G switch second -- almost certainly, they
were wired backward *and so was the test harness*. The same thing
happened with ERS-1's magnetorquers. And these were cost-is-no-object
megaprojects.

Finally, schedule monthly reviews to ensure that the project is not
"wandering" away from the two goals that you have chosen.

Consider carefully just how much time is spent *preparing* for formal
reviews. A project which schedules them monthly almost certainly will
never get as far as flying anything, because its engineers will never have
time to do much real engineering.
--
MOST launched 30 June; science observations running | Henry Spencer
since Oct; first surprises seen; papers pending. |