Project Organization

[Rainer Blome]

Hi Tunesfolk,

I'd like to send you a few clippings I found about a year ago and
thought they might be interesting for this list.  They are from
`More Moon Probe for Your Money', an article by Bruce D. Berkovitz
published in the April 1995 issue of MIT's `Technology Review'.  It
is about DOD's very successful yet comparatively cheap `Clementine'
moon survey mission.  (No, I don't subscribe to the magazine, it's
too militarist for my taste.)

The essence of the article is that NASA needs a big reorganization
because its structures stifle innovation and are "Inefficient to
the Core".


The first clippings deal with establishing reasearch goals:


    One of NASA's institutional problems is that it lacks an
    effective process for establishing goals and identifying the
    best compromise among costs, risks and payoffs.

This sentence is what reminded me of Tunes.


    ...  A more effective approach would be to define a
    comprehensive list of research goals that are reasonably
    specific but that do not dictate a single solution.

The point `What is Tunes?' should get a more specific answer.  `No
Kernel' dictates a solution.


    ... develop programs that, while not ideal solutions, represent
    the most effective compromise within the agency's limited
    budget ...

Nobody will ever get it completely right in the first place.  We're
still so much in the infancy of information processing, one should
not try to actually build `The Ideal System'.


    ... trying to achieve very ambitious goals while using the
    limited ... technology that was available ... when the basic
    design was frozen.  The cost of the program could have been
    reduced significantly if NASA had seriously analyzed tradeoffs
    and deferred some research goals just a few years. ...  But
    there simply was no effective process for determining which
    goals could be achieved easily with mature technology and which
    might better be put off.

Morale: don't bother supporting old technology like systems w/o
MMUs or FPUs.


The last clipping suggests establishing multiple subprojects that
deal with a single problem.

    ... because it is both a funding agency and an operating
    agency, it unnecessarily restricts the sources of ideas for
    ... research.  `Clementine' was cost-effective because BMDO,
    which held the dollars and was the party that would use the
    data from the mission, had the prerogative to shop around for a
    spacecraft and an incentive to get the most for its money.  As
    the ``seller'' of the system, NRL, in turn, had an incentive to
    be as responsive to BMDO as possible, knowing that the latter
    could look for another supplier or forgo the mission
    completely.

    NASA usually does not enjoy such a market.

Neither does Tunes.  Maybe it should.  Although we don't have any
money to spend, `to be allowed to do the job' seems to be enough of
an incentive.


Just thinking aloud.

Greetings,    Rainer


Responses by email, please (I'll be glad if there are none).