🚀 go-pugleaf

Rob Kling writes:

kling> Few of the books that I know about systems analysis take infrastructure
kling> into account -- and this is especially true of the texts which are
kling> used in university courses. I cite the better books, which I know of,
kling> in my article.

Kling's article presents an extensive bibliography for web-like models,
but for ``traditional'' models, instead of citations the article
contains only a conveniently weak description that has little to do
with how system analysis is practiced or taught.  If you turn to real
practitioners, or even real textbooks, you'll find a much different
picture.  Here's just one example: Barry Boehm's _Software Engineering
Economics_ (Prentice-Hall, 1981), a book that has been often used in
university courses.  I'll quote from Boehm's introduction:

	As you begin to confront practical software engineering
	situations, you will find that they contain both programming
	problems and nonprogramming problems: operational problems,
	budget problems; schedule problems; problems in determining the
	relative priorities of users' needs.  Bypassing the
	nonprogramming problems to concentrate on the programming
	problems will almost always lead to trouble later on.

Far from ignoring infrastructure issues, Boehm puts several of them on
the front cover of his book.  He doesn't mention the web model -- after
all, his book was written in 1981 -- but he clearly understands that
social context is crucial, and I don't think that the web model would
have helped him a bit.  Although specialized to just software
engineering, Boehm's work is far more ambitious and potentially useful
than Kling's, because it's quantitative, not just qualitative.

eggert> Kling presents no evidence that his models are better than other ad hoc
eggert> frameworks for describing the social context of computing.

kling> The paper provides some evidence through the analysis of one case study.
kling> The main contrast in the paper is between web models and dicrete entity
kling> models.

The only mention of traditional methods in Kling's case study was a
dismissive remark about how a traditional system analyst would
probably have made the mistake of choosing TeX over WordPerfect.
This brief and (in my view) incorrect remark hardly constitutes
an honest comparison of two rival systems.