Re: Market pressures for more reliable software

["Ken Garlington" <Ken.Garlington@computer.org>]

"Shmuel (Seymour J.) Metz" <spamtrap@library.lspace.org.invalid> wrote in
message news:3b52d7f5$1$fuzhry$mr2ice@va.news.verio.net...
: In <eZq47.410$rt2.150095027@newssvr11.news.prodigy.com>, on 07/16/2001
:    at 12:56 AM, "Ken Garlington" <Ken.Garlington@computer.org> said:
:
: >Errr... hello? Note that I am saying that *your definition* of
: >decentralized programming would lead one to the exact ridiculous
: >conclusion that you take so much care in ridiculing.
:
: I noted that you said that; I also noted that your claim was bogus.

Unfortunately, your argument was a straw man, since you used a definition
different than the one I referenced in my premise.

: >I'm glad that we agree that defining "decentalized" in the manner
: >you appear to espouse is silly, since then there would then be no
: >possible chance to have a "centralized" environment.
:
: Really? No chance of having centralized control over programming, etc?
: It's not only possible, it's been done, quite often. My claim was that
: it was done only in specific contexts rather than as an industry norm.

Since your definition of "centralized programming" seems now to be based on
organizational structures, not technology (making your earlier discussion of
remote job entry somewhat puzzling), perhaps you could post some examples of
"centralized programming" from this list of historical precedents which
occured "quite often," yet not frequently enough to be an "industry norm",
and discuss what key concepts they shared to make them "centralized". Per
your previous post, such "centralization" should encompass analysis, design,
coding, documentation, configuration control, debugging and resource
allocation. (This would seem to be more "software engineering" than
programming, but that's a minor point.)

: However, now I'm quite puzzled at what position you are trying to
: defend. JUst what is it that you claim was universally centralized
: prior to the PC, and by what definition, since you deny that
: cetralization of the obvious is possible?

As I noted previously, one definition of "centralized" programming is the
business environment as described in Brooks (and other respondents to this
thread), with the key concepts of (a) the CPUs are physically located close
together, in a single room or suite and (b) where the task of programming
(coding) requires the programmers to either enter the program at a single
"master" console, or to submit the programming job to a centrally located
operator corps for entry.

By contrast, "decentralized" programming has automation routinely available
at the programmer's work station (desk, etc.) without need of an
intermediary. (Brooks describes this as "interactive programming," and
indicates that it was rarely used at the time even though the technology was
available - see pg. 136.) We can further refine this by distinguishing
between the use of time-sharing systems connected directed to the
centralized set of CPUs described above using "dumb" terminals, etc., vs.
the availability of individual CPUs (PCs, etc.) from each programmer's desk.
The former case would still be considered "centralized" by some, since a
failure in the main computing cluster took down all available services, the
availability of tools, etc. could still be centrally controlled, and so
forth.

My claim (supported by Brooks and other respondents to this thread) is that
this concept of "decentralized programming," certainly the more "permissive"
second type described above, was not typically used in business computing in
the early years of that field. The definition presented above notes the
relevance of the PC to this argument.

: >Actually, I made an A in my logic course.
:
: Then you must have forgotten quite a bit. Either that, or you are
: deliberate refraining from logic.

Or that the logic taught in computer science in mathematics is a subset of
that needed for logical argument (see below).

: >How did you do in your debate course?
:
: Ah, so you have decided that winning a debate is more important than
: being correct?

Based on this evasive answer (and the inability to see more than two
possibilities in the previous statement), I assume you've never taken a
debate course and are unfamiliar with its goals. There are several resources
on the web that describe constructing a logical argument, and the associated
fallacies often committed by the uneducated or devious. I recommend

http://faqs.jmas.co.jp/FAQs/atheism/logic

I suppose that you could establish "being correct" through some means other
than the logic steps taught in debate classes (faith, etc.), but it would be
hypocritical to use these methods and then castigate others for "refraining
from logic."

: >So, you experience on the customer side indicates that the Brooks
: >description is atypical?
:
: My experience on the customer side indicates that there is massive
: variation in management style. I don't recall Brooks denying that.
: Brooks was describing how things were at IBM, or at least how he
: perceived them.

Actually, if you haven't read "Mythical Math-Month" (or have forgotten it),
you might want to read it (again). Certainly, no one who reads it should
come away thinking he was limiting himself to IBM. His description of
organizational structures is on pg. 78-79; given the various forms of your
definition of "decentralized programming", I couldn't say whether he
describes a "decentralized" organizational structure or a "centralized"
structure.