Path: santra!tut!draken!kth!mcvax!uunet!lll-winken!ames!oliveb!apple!vsi1!v7fs1!mvp
From: mvp@v7fs1.UUCP (Mike Van Pelt)
Newsgroups: sci.space
Subject: Why NASA is in trouble
Message-ID: <289@v7fs1.UUCP>
Date: 23 Mar 89 00:33:55 GMT
Reply-To: mvp@v7fs1.UUCP (Mike Van Pelt)
Organization: Video7, Cupertino, CA
Lines: 155

After a message I left flaming NASA for, among all the usual things,
expecting to replace all its retiring expertise with an "Expert
System", I received the following Email.  I thought it should be
posted, but the sender didn't wish his name attached to it.  He did
give his permission for me to be post it without his name.
----------------------------------------

Mike -

I saw your message on Heavy Lift Launchers from March 10 (sent to the
space news feed) as a part of the Space News Digest on ARPAnet.  A
couple of minor corrections on the Saturn V.  The two left over S-Vs
are at Marshall and at JSC, but the one at JSC is not in front of the
visitor's center, but is at the parking lot.  (A minor nit to pick, I
know.)

Castigating NASA for the long lead time to develop a new launcher is
not wholly appropriate.  Several major differences exist between the
NASA of the Apollo days and the NASA of today.  First, NASA had a
stable budget that was not as subject to Congressional Micro-management
and program stretchouts.  If the money needed was allocated and was
promised to be adequate (barring a war or other disaster) for the
scheduled program, it would make things easier.  For examples of the
problems that unstable funding and program strechouts lead to, read the
histories of the Shuttle program in "Challenger: A Major Malfunction"
and "Prescription for Disaster".  In addition, 60% of this fiscal
year's Space Station budget will not be available until May or June.
This means that for 40% of the money, 58 to 66% of the year's work is
to be done.  Yes, some equipment can be purchased at year's end, but
right now the Station is in the design phase, which is engineering
labor intensive, not hardware intensive.  This is due to a political
ploy by the Democrats, which would have allowed Dukakis to kill the
Station and move the money to aid for the homeless.

Second, NASA was put under the DOD procurement regulations, making any
significant ($100,000+) purchase a nightmare of paperwork.  For
something as big in $$ as a heavy lift vehicle it would take 2-3 years
to: write a Request For Information & Request For Proposal (RFP), get
the RFP approved by purchasing and legal, issue the RFP, wait 3+ months
for questions from the bidders, write the responses to the questions
(this has to be done by taking all the questions and writing a document
which answers all the questions without giving away the questions),
wait 3-6 months to get proposals from the vendors, set up a place for
the Source Evaluation Board to review the proposals and read them, and
finally (3-6 months more) award the contract.  This assumes that there
won't be a preliminary weeding out of bidders to 2 or 3 finalists, as
was done with Shuttle.  {Shuttle had 4 bidders at the first cut - I
think they were Rockwell, Boeing, Grumman and McDonnell-Douglas, then
it was cut to 2 - Rockwell and ?Grumman.  The Boeing proposal was a
fully reusable liquid fueled system with a flyback first stage and
higher payload, but was deemed too expensive to develop.  The Grumman
candidate shuttle had a higher payload, but a lower crossrange landing
capability and DOD did not want to use a vehicle that could not
guarantee an emergency landing outside a Communist country.}

Then, if anyone protests to the GAO, add 6+ months for the protest to
be looked at and ruled on.  (The validity of the protest has nothing to
do with anything these days - recently the contract for the Advanced
SRB for the Shuttle was narrowed down to 2 competitors from 5 or so - a
small Canadian company with only a few employees and no track record in
large boosters has protested and is now delaying the bidding process!)
So, in order simply get the program rolling will take 2-3 years.  Even
small purchases take a lot of time - it takes 4-6 weeks to buy a PC
around here.  I've got a PO for a Mac IIx in the works and it has been
there for 30 days now.  It took 2 weeks to get it signed off.

Third, NASA doesn't have the money to do what they did in Apollo - try
several lines of attack to the problem so as to have a backup if one of
them is a dead end.  The LEM that was finally used in Apollo was the
secondary contractor design.  Fourth, NASA (and its support
contractors) are having a hard time in recruiting engineers.  The
average government engineer starts out at $26,500 (for FY 88) and the
average nationwide was $30,500.  That $4000/yr difference is harder to
overcome, as the benefits of civil service are not what they used to be
- the retirement, insurance, etc. are all now on a par with private
industry levels or are more expensive.  In addition, all new civil
servants are now part of Social Security.  The only significant
advantage of civil service is a more liberal vacation policy.  It is
hard to get someone to take pay that starts out 15% below average, work
in grungy buildings, in crowded offices, put up with the B.S. and
paperwork and expect the pay gap to get worse as time goes by.  The
glory of working on the space program doesn't sell as well as it did in
the 60s.  I am making 20% less than my former co-workers at my old job
- and I have more experience and a MS degree to boot.  Recruiting at
NASA Ames in the SF Bay Area is a nightmare - after 10 years experience
you might make $42000/yr or so.  That doesn't go far when 1 BR
apartments are $700/month and houses are $200k.  The government tends
to get two kinds of engineers to apply - those that want to work on the
projects the government does, and those that can't find jobs anywhere
else.  In the Apollo days, the first kind of applicant overwhelmed the
second at NASA.  NASA has seen a reversal of that since the early 70s.
Draw your own conclusions for the other agencies.  The pay issue goes
all the way to the top of NASA managers - the head of JSC (a 2.5
billion+ enterprise with 5-6000 employees and contractors) can only
make about $72000 a year with a small bonus added on.  The starting
salary for a lawyer in New York City at an established firm is $69,000
and for MBA type it is in the 50s, so why should a good manager stay
around and make 50% of what he could in private industry?  What seems
to happen is that many managers bail out to industry as soon as they
have 20-25 years in at NASA and have a good pension built up for
retirement and then go to industry to add a second pension and make
some bucks.

On the subject of the expert systems article in the New York Times -
the writer had it somewhat wrong.  The INCO expert system that was
shown in the picture is not intended to capture the expertise of
retiring NASA engineers.  The AI aspect of the project is a minor part
of its success.  It has been well-accepted here at JSC more due to its
fast, readable data displays and flexible human interface than any AI
content.  The expert system in it is relatively small and not very
sophisticated (which was intentional - in a project that is as
mission-critical as this, you walk before you run).  The current
displays in mission control are cryptic and use a lot of hexadecimal
codes to display fault info.  The controllers have to memorize them or
look them up in books, which is time consuming and unwieldy.  (If this
sounds very stone-age, remember that the shuttle had a mid-70s
technology freeze for the electronic systems, and mission control is
driven by a bunch of old IBM mainframes with 80x24 character displays
and simple vector graphic displays like the old Tektronics 4000 series.)

There is a department working on a project to do design knowledge
capture for the space station.  The station will be on orbit for 30+
years, so NASA realizes that the designers will be moving on and will
not be around to answer all the questions that will come up during
operations.  So, a project that will capture the reasons for design
decisions, the tradeoffs, the pitfalls and possible problems and
failure modes is being launched.  It is hoped to put most of this
information into a database that will be available for future reference
and possibly for use in building expert systems.  NASA is very
conservative when it comes to AI, so don't expect that the engineers
will be replaced by black boxes anytime soon.  In any case, if the
expertise of a retiring engineer can be captured, that is better than
not having it at all.  I don't see why it should upset you.

NASA is not solely the province of evil moneywasters that intend to
sacrifice the future of the space exploration to the USSR.  There are a
lot of good people here that are looking down the road and want to do a
good job.  However, when Congress says design to cost not specifications
as happened on Shuttle, delays, stretchouts and overruns will (and did)
happen.  It really chaps me to see some of the uninformed, polemical
drivel that is put out on the net as the "truth" about NASA.  NASA is
not perfect.  NASA is not the ultimate example of what government
should be.  NASA is a human institution, with all the problems that
entails.

(name deleted by request)

---------------------------------
He's right.  Let's not forget, there are still some good people
at NASA, and they deserve better than they're getting from the
twits in Congress.
-- 
"Ain't nothin' in the middle                  Mike Van Pelt
o' the road, 'cept a yellow                   Video 7
line and dead 'possums."                      ...ames!vsi1!v7fs1!mvp