[Edu-sig] More on Dead Languages

kirby urner kirby.urner at gmail.com
Sat Sep 2 01:34:39 CEST 2006


So I was going over the M situation with Bernita this morning (a
health care data tech I know), talking up the predicament at the VA
(Veterans Administration), made known to me recently by UC Davis (an
active MUMPS/M center).

A short memorable form of my currently favorite strategy is "suMerian"
i.e. intersect the computer field's challange to recruit dead language
programmers with a problem already solved in Antiquities (one of the
humanities): make "knowing a dead language" (one or more) a really
cool attribute, a source of perks and kudos within the department.
That *doesn't* mean you can't also speak the current vogue lingo,
Ocaml or whatever.  Specialize in two why not:  one that's fresh on
the scene and one that's relatively ancient, and yet still running
some important machines.

This, I think, is a far better solution that rewriting and rewriting
working code, simply to "keep it in a language we understand or we'll
never be able to debug it or enhance it."  That *sounds* like a
reasonable strategy until you look at the cost in human resources.

"Hey, let's take a perfectly good operating system with this tiny
glitch, and junk it for this other one, because we don't know any C
anymore" -- that'd be the end of Linux right there, and just because
it didn't have some new features we wanted.  Nobody cared much for C
anymore, now that GUIs were so easy, so now it all comes crashing down
around us.  What a stupid development that'd be.  Like leaving the
Planet to Eloi, with not a Morlock in sight.

So here's my proposal:

First, the one I've already made many times (and not just on this
list):  that CS0 intro/survey courses be whiz bang and more fun, more
of a three ring circus, with lots of concepts coming and going, giving
candidate majors some dizzying experience of the wide-open field in
front of them, if they only apply themselves within our discipline.

Second, steal unashamedly from the humanities if there's already an
easy fix, in our case a failure to produce enough staff with the right
skills to satisfy academia, industry, government and the media (who
did I leave out?  just consider healthcare and entertainment as
industries, religion as all or none of the above).

What that looks like in practice might be frontloading with an "easy"
language such as Python (you've gotta admit, it's really clean OO),
then putting some forks in the road, one going to Java, but another
going directly to C, into the guts of our favorite snake.  Then bridge
to C# by dissecting the same snake yet again.  In Java, maybe go with
Jython and look at some of that source code.  What all these roads
have in common:  we're getting that an agile language on the surface,
is likely a system language under the hood.  Something faster to run,
but slower to write, now supports something faster to write, but
slower to run.  Since runtime is plenty fast, we go for the fast
writing.  But slow writing is still a very much needed skill, IF
you're going to go higher on the computer science ladder.

I was explaining to Barry and pbd on Barry's boat yesterday about
Knuth and his invented software/hardware for encoding the algorithms.
A passage of taut and sometimes funny prose, telegraphs to our minds,
but then the stupid machine has to be taught, and that takes a funny
language.  Knuth supplies one that's appropriately low level to keep
us living in hardware, just above the metal.  That's not a level we're
supposed to let go of.  We've got to keep a lot of focus there.

And so computer science needs that "nose dive" possibility, that after
CS0/CS1 a complement of courageous code jockeys will "head for the
basement" to study the inner gearworks.  Mostly others will do the
puppets and animations I mostly write about, given that's what I know
best.  But that doesn't mean I'm unaware of my peers, and their
admirable skills.

Assembler language is not really my bag unlike with Barry, a banker,
who makes money off it (on code he wrote himself mind you, inside of
some kind of power boat dimmer switch). I've never had to, even though
I studied it in college, and later in those Peter Norton books.

I do make money with Python though, though not yet as much as with
Visual FoxPro.

Kirby
Portland Knowledge Lab
Portland "open source capital" Oregon


More information about the Edu-sig mailing list