C and the future of computing

E

Eric Sosman

A new C Standard is on the horizon, and since the gestation period of
a Standard is roughly that of six point three elephants it is perhaps
not too soon to start thinking about the next one. What will computing
be like in the late 2020's, and what role will C play? Prognostication
is an uncertain business for those who lack groundhog genes, and it is
virtually certain that my predictions will be incorrect in detail, but
I hope and believe that many of them will be true in broad outline.

First, I believe that the future of computing is rooted in its past.
Athene sprang full-armed from the brow of Zeus, but capitogenesis is
rare in computing: Somebody has an idea, various people kick it around
for years or even decades, and finally somebody else makes money off
it. Computing in the 2020's will be an outgrowth of something that's
already shivering on the chilly fringes today. Will it be Gallic
Asinine? Probably not: I find myself agreeing with those who say "It's
the future of computing: Always has been, always will be." Super-
parallelism? I doubt it, since Lobachevsky and Riemann have long
since shown the parallel postulate to be an arbitrary construct.
Artificial Intelligence will never match Natural Stupidity, RAD is
just a FAD, and The Singularity is only a trap representation.

What nascent development will nurture our future, or nuture our
furture? What mighty hint will we all ignore for the next several
years, until we suddenly all start claiming "We were There?" It's
impossible to be certain, of course, but I think the most likely
candidate is Quondam Computing.

Quondam Computing is thought of as a very modern development, but it
is in fact quite old. Some of the fundamental concepts can be found
in the works of Zweistein, although he himself professed disbelief
in the ideas and dismissed them as "Stupid action at a distance." A
few historians trace the origins of QC all the way back to Heisenbug
himself. But even they are too timid: The earliest references to QC
are in Genesis 6:15, where the Lord commands Noah to build an Ark
"three hundred qubits long, fifty qubits wide and thirty qubits high."
That's 450,000 qubic qubits (times a scale factor for the shape, but
programmers traditionally ignore O(1) multipliers), a quantity of
qubits that would be the envy of modern quomputer scientists.

I believe, though, that the day is not far off when today's physicists
will deploy exotic materials like selenium trixenide, gadolinoleum
emancipate, and perhaps thiotimoline to duplicate what Noah did with
cypress. (Space does not permit a recitation of the properties of
these compounds; for details consult Wonkypedia or Giggle.) When, not
if, this happens, we will surely see the rise of a new Zack Mickerburg
or Allen Learison to claim ownership of the idea and lock it up with
patents and the like. Even so, I believe QC will be too important to
lie dormant: The future Farceback or Borgacle will stand to make so
much money from QC that they'll be forced to market it, and then it
will be only a matter of time until someone open-sores it.

What, then, are the salient features of Quondam Computing, and how
well do they mesh with C? I've already mentioned "stupid action at
a distance," the phenomenon by which entangled quondam particles (i.e.,
those whose divorces aren't yet final) can affect each other without
communicating (e.g., by refusing to answer the phone). It turns out
that C has supported stupid action at a distance since its earliest
versions, with a construct called the "wild*" or "wild pointer!" Up
to now, programmers have tried to avoid SAAAD, and to stamp it out with
tools like Eclectic Farce and Poorify; in the new world of QC they
should learn instead to welcome it -- since it's pretty much inevitable,
they might as well.

Another feature of Quondam Computing is what's called the superstition
of states, where you believe your program is behaving as intended even
though there is no rational reason to think so. Again, C is admirably
suited to this attitudinal shift: Since a C program almost never does
what it's supposed to, superstition of states is practically a given.
Indeed, some C programmers can attain superstition not only of states,
but of nations or even of continents!

The most important QC characteristic of all, though, is indeterminacy.
A quondam system is useful because its state is undefined; the moment
you observe the state its waif function prolapses to a single outcome
and all the other possible outcomes are forever lost. Just as a cat
has maximum entropy when you can't see it (for all you know, it might
be chasing mice in the cellar or chasing squirrels outside or barfing
on your bed), so the quondam computer has maximum information content
when you haven't yet looked at the output. The moment you see the
output is the moment when the QC must commit to giving you just one
answer and losing the other billion. How can any programming language
avoid dissipating QC's indeterminacy and thus losing its value?

Here, it seems, is where C really shines. Languages like Ada have
rules like "Do X, and Y will happen." Java makes almost a fetish out
of such strictures: "Do X on any system anywhere, and Y will happen
on every system everywhere, maybe." Unlike these rigidly certain
languages, though, C is the language that deliberately erects shrines
to Undefined Behavior, Unspecified Behavior, Implementation-Defined
Behavior, and Impolite Behavior. C is thus perfectly suited for the
implementation of quondam computing algorithms: to get maximum benefit
from QC you must not ask what will happen, and C is the language that
wouldn't tell you even if you did.

America's best-known computer scientist, Dilbert, once explained
quondam computing as being like a rotating doughnut shot from a
cannon at the speed of light, but without the doughnut. That, it
seems to me, is the Spirit of C that will carry the language forward.

(Permission is granted for unlimited electronic transmission and
storage of this Intellectual Property. Permission is also granted
to any single person to print exactly one hard copy, no more. All
photocopies require individual licenses, from originator's own lawyers.)
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Members online

No members online now.

Forum statistics

Threads
473,744
Messages
2,569,484
Members
44,903
Latest member
orderPeak8CBDGummies

Latest Threads

Top