Re: Ichbiah's letter to Anderson: Here it is

[sampson@cod.nosc.mil (Charles H. Sampson)]

In article <SRCTRAN.93Apr29102415@world.std.com> srctran@world.std.com (Gregory
 Aharonian) writes:

>    The whole point of all of my messages to date is that the Ada Mandate
>is a gross distortion of the marketplace with very significant economic
>implications that have never been fully examined by the DoD. ...

     I don't have to stretch too far to accept the claim that the Ada man-
date is a gross distortion of the marketplace.  However, I would rather
characterize it as simply ignoring the marketplace.  The mandate reaffirms
the basis of the entire effort that led to Ada: the DoD will save substan-
tially by focussing on a single language for its tactical systems.  Fur-
ther, it chooses Ada as that single language.  Usually it's not a good idea
for a legislature to decide technical issues, but in my prejudiced opinion
they lucked into the right decision this time.

>                                                       ...  At best, given
>what I have seen, over the life-cycle, using C/C++ is slightly more cost
>effective than Ada.  At worst, it's much better.

     [referenced later]

>                                 ...  The Ada Mandate forces the DoD to
>become businessmen - to have to deal with marketing, competition, free
>market supply and demand, cost/benefit analysis, investment, etc.

     To the contrary, the mandate doesn't require the DoD to be businessmen
at all, or at least not much.  There's generally no decision to be made;
use Ada.  They are only required to be businessmen to the degree that they
can evaluate a cost/benefit analysis in support of a waiver.

>           IF ADA IS AS COST-EFFECTIVE AS ITS PROPONENTS PROCLAIM,
>           THE MANDATE WOULD BE IRRELEVANT, SINCE FOR EVERY BIDDED
>           PROJECT, THE ADA BIDS WOULD ALWAYS BE LOWER.

     That's not the way military contracts are bid.  Usually implementa-
tion, IV&V, and a succession of maintenance contracts are bid independent-
ly.  It's well-known that implementation costs are a small part of the
life-cycle costs of military software.  If contractors were allowed to
choose the implementation languages, they might well say, "We can do this a
lot cheaper in B- than in Ada," provided they didn't have to consider the
costs of the follow-on maintenance contracts.  As a matter of fact, during
the process that led to Ada it was accepted that the costs of implementing
in this new language might be higher than in some other languages.  It was
decided that this was acceptable if total life-cycle costs were lowered.

     Finally, the Ada mandate has not precluded other languages.  The only
requirement is that use of another language on a particular project must be
shown to be cost-effective over the entire life of the software.  Based on
your earlier statements, this shouldn't be too hard to do.  I wonder why it
hasn't happened often.

                                   Charlie