Fri, 18 Dec 1998 07:29:11 -0500
> If you look at Jones' data, it becomes obvious that we should use either
> MathCAD OR Excel. My report on commercializing Ada comes to the conclusion,
> that Jones' has once again proven that domain specific libraries and tools
> increase programming efficiency. ....
Yes. As I have stated before, I believe that Jones' numbers are only valid
(if at all) in the domain the "language" is intended for. And he states
that some of the numbers are intuitive guesses based only on an examination
of the languages' descriptions.
> ...... If you wish a preprint, I will send you the
> .PDF file. This preprint will be posted to an appropriate Web page (CAUWG).
If the PDF is on the Web page, just tell me the URL. Otherwise, yes, I
prefer PDF to HTML for anything of substance. :-)
> ..... many computer science studies.... unacceptable in the scientific
> literature. To this day, the DoD does not know the comparable costs of using
> programming languages. Ada does NOT need a mandate; she needs hard data!
> have not seen any data proving that the CMU Capability Maturity Model levels
> have anything to do with quality or cost.
Correct again. I personally believe the CMM is valid. However, the CMM
is not derived from quantitative empirical measurements. It was
originally formulated mostly by "common sense" -- and "sense" is neither
common nor infallible. When the CMM was formulated, no company was level
four or five. The first one that made level five allegedly did it by
writing its directives from the CMM questionaires. However, I am aware of
data that does support the CMM in a way. Watts Humphrey developed a
"personal software process" based on CMM Level Five. He then designed a
course that introduces parts of that process over time. He and the SEI
have collected many data points showing that the introduction of his
process does correlate to improvements in quality, cost of quality. As
far as I know, there is no data on whether it improves overall cost. And
PSP data cannot really prove anything about the larger CMM.