[Edu-sig] Fwd: Do we "teach computers" when we write code?

[kirby urner]

Yes sir, exactly.  I believe that in calling computers "stupid", other
epithets, we're asserting our aggressive primate dominance over these
gray boxy things that in some science fiction have beamed aboard
spaceship earth in order to steal our jobs or something, because
they're smarter than we are.  This was the "HAL story" of 2001, which
you'll notice was way off target because actually we have no clue how
to code a thinking human-like intelligence that passes the Turing Test
(which some humans fail also).  Lots of doe-eyed venture capitalists
got led down that "AI lane" only to watch their money get slaughtered
out behind the barn, by wild-eyed computer scientists with BS for
brains (yeah, pretty crazy).

Sir Roger Penrose likes to show slides of really simple chess puzzles
that'd stump Big Blue.  I think chess playing is a good place to get
into the lore, as one of the earliest ideas of "the AI takeover"
(displacement of human intelligence) came in with The Turk, a
chess-playing gear-works, a box, like a computer, with a puppet player
over the board that would move the pieces.  This thing beat Napoleon,
likely provoking a crisis in consciousness in the latter tyrant slash
dictator, whereas that'd have to be one of the bravest dwarfs on
record, to risk humiliating a DFL in that way (he was stuffed in
amidst the gears, a kind of magician's trick).

Back when I was an ardent disciple of Ted Nelson, reading Computer Lib
/ Dream Machines and believing in hypertext, I would yak about how key
words might make a few links, but really only human intelligence is
able to perceive what we call connections, which might be based on far
fetched analogies, buried allusions, the kind of stuff we use in the
humanities for glue language (psychoanalysis... greek mythology).  You
get these kinds of networking effects in Google though, because that's
not just a keyword index.  It has measures of "where the humans find
relevance" i.e. what searches mapped to what pages -- becomes
self-fulfilling in awhile, as we tend to get lazy and pick one of the
"top 10" (what others have picked) but in any case, it's not like "AI"
is in control of the hyperlinks.  Real thinking humans have to make
those URLs across disciplines.  The Web is not the product of AI, but
of humans collaborating (dream weaving).

Probably CS types who love computers, consider this whole area
fascinating and wonderful, have a hard time getting into a mindset
where computers are looked upon with fear, even dread, because of all
those science fiction projections.  Remember 'The Matrix':  the
terrible future is one "run by machines" as these misanthropic men in
black guys who torment our heroes are only represented as such in the
simulation, have no actual form outside the bleak landscape of AI-dom,
whereas the humans are encapsulated within pods, their energies
harvested, their lives a mere dream.  Likewise 'Terminator' is about
machines taking over, starting huge wars without our permission.

So calling computers 'stupid' and thinking of programming them as a
rather condescending activity, where we can't use our pedagogical bag
of tricks, have to be formal and direct, use a "stupid computer
language" is in some sense an intellectual counter.  Gauss is the
right hero to bring in, as he makes those "noncomputable leaps" Sir
Roger likes to talk about, whips Big Blue's ass.

This whole picture gets flipped on its head in Kubrick's 'Artificial
Intelligence' though, as here the predatory humans start to seem evil
and fascist, given how they exult over their own supposed superiority
-- a projection from many a human past (and present), wherein one
caste (class) of human is made to play "subhuman" to prove the other's
sorry tale of inherent superiority.  Our hearts go out to the
oppressed computers in this case (like I said, a reversal).

Once past this stage of asserting king-of-the-hill dominance (i.e.
we're done feeling threatened), I think it's OK to go back to the
music idea (e.g. laptop as guitar, Portland as FOSS Nashville).
What's so liberating about today's music software, a lot of it on the
Mac, is even if you don't have the reflexes and training to play "in
real time", you can drag and drop notes, based on whatever intuitions
and theories, and run the thing back at speeds even a professional
couldn't match (should you wish to do so).

In that sense, we're right to see computers as "better", in the sense
of "faster".  We take all our file cabinet skills, what it'd take in
terms of paperwork to look up a balance, deduct an amount, write out a
plane ticket, mail it, and code those into our little "musical
numbers", then set those running, making computers our "energy
slaves".  We each have the equivalent of many secretaries working for
us, researchers in libraries, mail delivery people, thanks to the
Internet and intranets.  With a few clicks of the mouse, we mobilize
virtual armies, enjoy a sense of power thereby.

Our powers have become amplified.  We are better served, by clerical
and information resources.  This "energy slave" concept didn't start
with computers, arose in the 1940s w/r to earlier "ephemeralizings"
(using tech to do more with less, improving living standards by
optimizing, applying comprehension of principles), but it's still as
relevant as ever, why Dr. Bob Fuller grabbed on to it for his 'First
Person Physics' initiative.

I bring this up because of the "slave" meme, which engineers keep in
play within a benign context/namespace e.g. this hard drive is slaved
to that master one etc. (people worry when they hear talk like this
though -- we need to remember what's scary about engineers, as we move
forward with STEM recruiting).

Speaking of energy slaves doing clerical work, I have some business to
attend to...

Kirby

Related:
http://www.4dsolutions.net/ocn/numeracy0.html  (includes the Gauss story)
http://mathforum.org/kb/thread.jspa?threadID=1917995&tstart=0 (lesson
planning @ Math Forum)
http://mybizmo.blogspot.com/2008/04/philosophy-posting.html  (about the AI bust)
http://physics.unl.edu/~rpeg/First_Person_Physics/urneressay.html
(First Person Physics)


On Tue, Apr 7, 2009 at 6:32 AM, Lloyd Hugh Allen <chandrakirti at gmail.com> wrote:
> I haven't posted in a while -- forgot to reply-to-edu-sig :)
>
>
> ---------- Forwarded message ----------
> From: Lloyd Hugh Allen <chandrakirti at gmail.com>
> Date: Tue, Apr 7, 2009 at 09:26
> Subject: Re: [Edu-sig] Do we "teach computers" when we write code?
> To: kirby urner <kirby.urner at gmail.com>
>
>
> As a math teacher, using the particular example of summing a finite
> set of consecutive integers:
>
> To give students a formula, in particular n(1+n)/2, and then have them
> do a set of practice problems where they apply that formula, is not
> teaching. It might be training.
>
> Instead, consider the case of telling students that: when Gauss was in
> elementary school, his teacher needed time to work on some other
> matter and so told the students to add all of the numbers from 1 to
> 100; and that Gauss instantly looked up and said 5050; and the teacher
> hadn't actually yet done the problem himself and so denied Gauss'
> answer. Gauss, as an ~8 year old, said, no, look, and wrote
>
> 1 + 2 + 3 + ... + 100
>
> and then below that wrote
>
> 100 + 99 + 98 + ... + 1
>
> and showed that there were 100 columns, and that each column summed to
> 101. However, he then noted that he had written the series out twice,
> and so had to divide that product by two. The 100 columns is the n;
> the sum of the first and last number is 1+n; and then divide by two.
>
> And then to have the students try to represent a similar problem, and
> to check their answer against the formula, and THEN to have them do a
> set of practice problems, that might be teaching.
>
> If the computer were able to understand the story about young Gauss,
> then we could teach it. Instead, we can use it to confirm that the
> formula seems to work (because computers can add numbers in the
> fashion that Gauss' elementary school teacher expected just as fast as
> we can apply the formula), and we can show that using the formula is
> still faster for the computer than actually summing the list, but no,
> we are not teaching the computer.
>
> Perhaps if the computer were then able to, of its own volition, wonder
> what we would get if we were to sum consecutive squares, then we could
> teach it. As hard as it is to get students to wonder about things,
> it's even harder to create that state in computers.
>
> -Lloyd
>
> On Mon, Apr 6, 2009 at 18:05, kirby urner <kirby.urner at gmail.com> wrote:
>> I'm wondering what others on this list think of this non-standard use
>> of "teaching" when talking about programming a computer.
>>
>> The authors say we're "teaching" the computer....
> _______________________________________________
> Edu-sig mailing list
> Edu-sig at python.org
> http://mail.python.org/mailman/listinfo/edu-sig