Re: Aspiring highest-order programmer

From: Edward G. Nilges (spinoza1111_at_yahoo.com)
Date: 06/11/04 

Date: 10 Jun 2004 15:49:55 -0700

blmblm@myrealbox.com wrote in message news:<2iqo8pFqegl2U1@uni-berlin.de>...
> In article <f5dda427.0406092312.43f8958c@posting.google.com>,
> Edward G. Nilges <spinoza1111@yahoo.com> wrote:
> >Willem <willem@stack.nl> wrote in message
> >news:<slrncb8ntm.25iv.willem@toad.stack.nl>...
> >> JKop wrote:
> >> ) My post was completely void of sarcasm. At first I genuniely thought that
> >> ) you were just extremely intelligent and that I simply didn't have the
> >> ) intelligence to understand your post... but then I just looked closely and
> >> ) decided to try understand what it was that I didn't understand, at which
> >> ) point I realized that I was simply unaccustomed to your language, and your
> >> ) unspecific references. Both of which make you come across, at first, as
> >> ) extremely intelligent.
> >>
> >> I should point out that, in the past, a lot of his pompous language and
> >> usage of unfamiliar verbiage has turned out to be plain wrong. He has
> >> never admitted this, but claims we are unable to grasp his unique usage.
> >> Not so recent examples include NP-completeness and fascism.
> >
> >You failed to thoroughly understand NP completeness in that
> >discussion, as far as I can remember. You stuck to what the text book
> >said.
>
> Whereas as far as I can tell from a look through Google's archives,
> you don't remember what the "text book said".

I have replied separately to your request that I describe my
understanding of the applied theory of NP-completeness. It was
probably a mistake to do so, since anything other than what on usenet
corresponds to moronized rote memorization (cutting and pasting a
Wikipedia) will be "deconstructed" in a Sophistical, but moronic,
fashion, to "prove" that "he doesn't know what he's talking about",
when I do.

The "text book" in 1971, when I started reading comp sci and took my
university's first class in comp sci, almost did not exist. Our "text
book", Sherman's Programming and Coding for Digital Computers, was
useless because it assumed that the university teaching the class had
an IBM 7094 running IBSYS as an OS and with a Fortran compiler. My
university had an 8K 1401 running the OS diddly squat; each program
carried its OS on its back.

Another "text book" was Donald Knuth's first editions which contained
material on what became the theory at a later date, and, as I
mentioned, I recently reviewed the mat'l in Stephen Skiena's book The
Algorithm Design Manual.

However, one attractive feature of computer "science" in 1971 was the
fact that it hadn't "matured" to the point wherein knowledge is rote
memorization and cutting and pasting from Wikipedia.

It was probably a long-term mistake for me to try to escape
authoritarian structures by in 1971 entering an "immature" field in
which "mere" programmers (forgotten men, for example Bob Bemer of IBM)
were advancing human knowledge in a collegial fashion. I'd seen the
psychic price paid by my neurosurgeon father, who succeeded at Harvard
and med school only by 24/7 hard work, and neglecting his family.

But as "high" tech became oh, so very important in the Reagan era to
America's declining sense of self-worth it too became divided into
theory and praxis.

The turning point, for me, was when I noticed the difference between
the code examples in Knuth's 1968 Art of Computer Programming and
Numerical Recipes in 1978, for in the latter there were unacknowledged
errors and stylistic details that made it clear that the "mere coding"
had been farmed out to graduate students who were so alienated...that
(for example) they used a font which confused I and L in code that
used both variable names frequently.

My own father, in fact, helped me to understand what was happening.
When he started brain surgery, the field was still to some extent a
matter of combined theory and praxis and of individual endeavors such
as those of Dr. Cushing in Boston. These men wrote books in which
their humanity was simultaneously a scientific limitation and an
organizing principle that made the material understandable.

Later on, however, my father increasingly encountered what Gary
Cornell of Apress calls "Frankenbooks". These are mulitauthor texts in
the medical and programming fields alike that are collections of
disparate views which manage not to present a comprehensible
framework.

Thus today I am confident that computer science students learn "all
about" NP-completeness and are able to regurgitate the material on
their examinations. I am equally confident that they then, on their
first job, develop programs that behave in the NP complete fashion.

I am equally certain that comp sci students learn "all about"
compilers in Compilers 301. But I also just know that the A students
in classes of this nature take away from class one lesson, and one
lesson alone.

This is that they will never write a compiler if they can help it, for
the class was so boring, and the professor so sadistic, that for them
compiler design is a metaphor for pain. In particular they will never
consider solving a problem with a "little" compiler for a "little"
language.

>
> Edward, maybe I didn't read the right parts of the right discussion
> (I searched for postings in this group written by you and containing
> "NP complete" or "NP completeness", and the threads that came up were --
> long, very long, and not exactly focused on NP completeness). But in

Right, because moron math is less important than other issues.

> the posts I read, Willem seemed to be using "NP complete" in the way
> it is used in its field of origin and in one of his posts provided
> a link to a definition/discussion in the comp.theory FAQ. I don't
> get the sense from the way you use the term that you've read this
> formal definition, and you seem not to have explained anywhere what
> you think it means. Why don't you do so? You seem to be claiming

See separate post

> to have a practical understanding of the underlying ideas that's
> more valuable to a working programmer than any formal definition.
> Could be. What is it?

When I encountered the theory in Knuth (who is to me of far greater
stature than most younger computer scientists, in part because he does
his own goddamn programming and is man enough to admit error) it was
in the context of "aha" rather than the context of "ho hum".

The context of "aha" is where a "theory" makes sense because you see
its instantiation in the real world.

O(n*n) is where the program starts slowing down for larger inputs and
in the real world it means "it works in test, why is the user hopping
mad?"

O(n) is where the program runs at constant speed. For example, the
simple "recursive descent" algorithm in my book "Build Your Own .net
Language and Compiler" (Apress) empirically confirms that it is O(n)
by providing an indented compiler progress report.

Almost never provided by "real" compilers, this progress report is
emitted for each successful parse of a grammar symbol and has the form
of an indented outline. It is provided at constant speed which is an
empirical confirmation that (1) the recursive descent algorithm is
order(parse depth) and (2) there are no "hidden" higher order detours
in the code.

You "waded" through usenet postings and found nothing to confirm or
deny a pro or anti view. This is because most of the sludge was
created, I am now convinced, by people who use this ng not for the
purpose it was intended but to shore up shaky senses of self-worth by
abusing newbies and outliers.

Part of the "maturation" process of any field is in fact the
elimination of charlatans, snake oil salesmen and quacks. When I
started out in compsci, one of the superficially attractive features
of the field was in fact its domination by amiable rogues, but today
one of its depressing features is the absence of same. The problem is
that today, knowledge production in the field has been formalized to
the extent that knowledge becomes like wisdom in the Bible as mediated
by Prince Hal in Henry IV: "thou sayest well, for wisdom crieth out in
the streets, and no man pays it any mind".

Prince Hal is commenting on Shakespeare's Falstaff, who has told the
Prince that the Lord Chief Justice, Falstaff's nemesis, had berated
him in the streets, and the Henry trilogy in Shakespeare is a
meditation on the costs and benefits of political "maturity".

Falstaff, like an old time computer consultant, is one of those
bullshitters who occasionally comes up with a useful insight. But the
problem for students is that their sources of knowledge have to be
accepted by society as a whole.

Thus any recount "in my own words" of NP completeness is suspect
without editing and review.

However, I do know that this newsgroup is meant for the discussion of
programming and that unless discussion of a technical field by its
actual participants is now "wrecking" in Stalin's sense, it is
misusing this newsgroup to use it instead to destroy reputations.

In a court of law, it would be found that while Willem is a good
student, he's unqualified to pass on my capabilities, and, far more
seriously, he unwittingly at first participated in a conspiracy
initiate by Richard Heathfield who has since absented himself from
these discussions, to focus this ng on low-level C programming issues.

>
> [ snip ]