Re: AIA Position on Ada

[rogoff@sccm.Stanford.EDU (Brian Rogoff)]

"Marin David Condic, 407.796.8997, M/S 731-93" <condicma@PWFL.COM> writes:

       BTW: The argument "there's more C stuff out there..." is the only
       semi-plausable business/engineering reason to think that "C is
       better than Ada...", in my not-so-humble opinion. In every other
       respect Ada is as good as or better than C and I have never heard
       any convincing engineering reason to select C over Ada.

Well, that little chestnut ("There's more C stuff out there...") can be 
expanded quite a bit

(1) There is a larger existing base of C code than Ada code
(2) There are more tools for C than for Ada
(3) There are more programmers familiar with C than with Ada
etc.

No need to reply that a lot of this C code is unusable, that the tools exist 
to correct deficiencies absent in Ada 95, etc. I agree. IMO, the way to fight 
this battle is to write lots of Ada code that is better than the C, to build 
Ada tools that do more for Ada than Purify and co. do for C (how about 
Larch/Ada-95 and a free GC for Ada 95?), and to get more people looking at 
Ada as a C alternative.

       But then, this seems to be the way of things in a technological
       society. Beta was better than VHS - so VHS wins. VMS was better
       than UNIX, so UNIX wins. Macintosh was better than IBM-PC, so
       IBM-PC wins... etc. etc. etc.

I believe that what is going on is an example of positive feedback in a 
control system. Once a technology is a bit more popular than a competing 
technology, its popularity becomes the reason that people choose it over its 
competitors. Hence time to market is usually more important than quality, 
and the type of market "conquered" (say PC vs workstation) is also 
important. Also, arguments like "VMS was better than UNIX" are plain wrong. 
The Symbolics Lisp machine environment circa 1985 was arguably better than 
any OS of the time, yet it didn't run on any other hardware. UNIX is 
portable, VMS isn't. Too fucking bad for VMS and Symbolics. Similar arguments 
can be made for the Mac (expensive, closed, yet functional and elegant).

I first read the "positive feedback" argument in an old (late 1980s, early 
1990s) Scientific American article titled "Positive Feedback in Economic 
Systems" or something like that. I don't believe that this argument applies 
perfectly to programming language acceptance however. Note that COBOL, 
Fortran, Lisp, and PL/I :-) still have active user communities.

-- Brian