TITLE: Learning From the Wheel of Reincarnation
AUTHOR: Eugene Wallingford
DATE: April 15, 2012  8:10 PM
DESC: 
-----
BODY:
Last week I was pointed in the direction of a forty-five
old <EM>CACM</EM> paper.  It had an innocuous title,
<A HREF="http://cva.stanford.edu/classes/cs99s/papers/myer-sutherland-design-of-display-processors.pdf">
   On the Design of Display Processors</A>,
and was outside my areas of primary interest, so I might
well have passed it by.  But it was co-written by Ivan
Sutherland, whose work
<A HREF="http://www.cs.uni.edu/~wallingf/blog/archives/monthly/2004-11.html#e2004-11-06T21_03_42.htm">
   Alan Kay has praised often</A>,
and it was recommended by someone on Kay's
<A HREF="http://vpri.org/mailman/listinfo/fonc">
   the Fundamentals of New Computing mailing list</A>,
where a lot of neat ideas are discussed.  So I printed
it out for my daily exercise bike ride.  I'm glad I did.

Myer and Sutherland tell the story of needing a display
system for a research computer.  That was a bigger deal
in 1967 than it is today, so they did some research of
their own....

<BLOCKQUOTE><EM>
 Finally we decided to design the processor ourselves,
 because only in this way, we thought, could we obtain
 a  truly complete display processor.  We approached
 the task by starting with a simple scheme and adding
 commands and features that we felt would enhance the
 power of the machine.  Gradually the processor became
 more complex.  We were not disturbed by this because
 computer graphics, after all, are complex.  Finally
 the display processor came to resemble a full-fledged
 computer with some special graphics features.  And
 then a strange thing happened.  We felt compelled to
 add to the processor a second, subsidiary processor,
 which, itself, began to grow in complexity.

 It was then that we discovered a disturbing truth.
 Designing a display processor can become a
 never-ending cyclical process.  In fact, we found the
 process so frustrating that we have come to call it
 the "wheel of reincarnation."  We spent a long time
 trapped on that wheel before we finally broke free.
 In the remainder of this paper we describe our
 experiences.  We have written it in the hope that it
 may speed others on toward "Nirvana."
</EM></BLOCKQUOTE>

A mantra from the paper characterizes the authors'
time on the wheel: <EM>"For just a little more
money..."</EM>.  I'll bet that sounds familiar to a
lot of researchers, not to mention all of us who buy
computing equipment for labs and end users.

I was really happy to read this paper.  It's an
experience report in which the authors share honestly
the mistakes they made.  But they paid attention,
recognized a pattern, and learned from it.  Even better,
they wrote what they learned, in hopes of teaching the
rest of us.

The wheel of reincarnation is not limited to display
design or hardware design.  It occurs any place where
we encounter complexity.  We try to tame it, first by
specializing and then by generalizing.  The design of
programming languages is just as prone to dizzying cycle.

(In software design, we have a related phenomenon,
captured in
<A HREF="http://en.wikipedia.org/wiki/Greenspun's_tenth_rule">
   Greenspun's Tenth Rule</A>.)

In language design, we almost have to look for a fixed
point at which we stabilize the pendulum between general
and specialized.  What we most often need as users is
the ability to grow systems gracefully over time.  This
speaks to the value of a good macro system and good
compiler design.

Reading this paper reminded me of a couple of lessons
I've learned over the years:
<OL>
<LI> I should read and watch everything I can get my
     hands on from Sutherland,
     <A HREF="http://www.cs.uni.edu/~wallingf/blog/archives/monthly/2004-07.html#e2004-07-19T10_52_37.htm">
        Robert Floyd</A>,
     and other computer scientists of their generation.
     They were solving hard problems long ago, in the
     face of resource limitations few of us can imagine
     today.  </LI>

<LI> As someone tweeted recently,
     <A HREF="http://twitter.com/#!/fogus">
        @fogus</A>,
     I think: reading these old papers makes me think
     that I will never have an original idea in your life.
     But then they also teach me a lot and prepare me to
     have more and sometimes better ideas.  </LI>

<LI> I need to do my best to hang around with smart,
     curious people.  It's old advice, I know, but it
     requires action and, in some environments, constant
     vigilance.  Simply eavesdropping on the FONC mailing
     list raises the level of my brain's activity by a
     few degrees.  </LI>
</OL>

These papers also remind us of a valuable role that
academics can play in the software and computing worlds,
which are also heavily influenced by industry
practitioners.  We need to keep papers like these alive,
so that the smart and curious people in our classes and
in industry will read them.  We never know when two ideas
will crash into each other and lead to something new and
better.
-----