Re: [xml-dev] seduced by markup

[Steve Newcomb <srn@coolheads.com>]

On 11/15/2013 10:10 PM, John Cowan wrote:
> 
> 
> 
> On Fri, Nov 15, 2013 at 4:32 PM, Steve Newcomb <srn@coolheads.com
> <mailto:srn@coolheads.com>> wrote:
> 
>     The DTD syntax was never about machines.  It was about human beings, and
>     it is still, even today, and as crummy as it is, the most humane way
>     available for human beings to communicate about data design in a diverse
>     collaborative environment that inevitably must include non-programmers.
> 
> 
> In my opinion, RELAX NG compact syntax is far more humane than DTDs.  It
> is as readable, if not more so; it has greater power; it has far fewer
> arbitrary restrictions.

It is *more* readable for you and for me.  It is *less* readable for
those who haven't yet fully internalized that XML schemas are all about
element types.  For us,

"<!ELEMENT..."

is absurdly large and redundant, and

"<!ATTLIST..."

is even worse.  Why on earth is it separate from its corresponding
element type declaration, with its own enormous, 50% redundant billboard?

The answer, in both cases, is that it's more humane.  Let's first
remember that the design was done not by a computer scientist or other
academic who was trying to demonstrate his awesome command of his field
of expertise and/or his artistic/poetic gifts, but instead by a lawyer
who was trying to improve the ease with which ordinary people could make
useful and enforceable agreements among themselves about their
communications.

Now, here's a rule of thumb that works for spoken, written, and musical
communications:

The most effective communication streams have about 15% information and
85% redundancy.

More redundancy is bad because people get bored and lose interest.  Less
redundancy is bad because people lose track, and loss of attention
follows very soon thereafter.

I studied music for a long time, and many of my views are informed by
that study.  Please let me explain about this in terms that are familiar
and compelling for me.

In their heyday, 12-tone rows made perfect sense as the Next Big Thing
That Would Influence Everything That Will Ever Happen From Now On.  So
why did tonal music survive, while the 12-tone row is, in retrospect, a
bizarre blip in the history of music?  I think it's because tonal music
tends to be around 85% redundant, while the 12-tone row technique packs
more information into the stream.  The 12-tone row music sails too close
to the chasm of entropy.  For most people, the music is simply
inaccessible.

If you want a demonstration of what I'm talking about, let me recommend
some of the most potent music I know, the expressionist opera *Wozzeck*
by Alban Berg.  In it, the level of redundancy is significantly lower
than the norm for operas.  You'll start to love it only after you've
heard it a several times.  Ten times, maybe.  Prior to that, you'll have
trouble resisting the temptation to turn off those repellently
incomprehensible orchestral squonks, biggles, and portentous blaring
crashes, not to mention the off-putting, semi-tuneless sprechstimme of
the singers.  Another ten times and you'll be whistling pieces of it for
your own amusement, but at that point I'd advise you not to whistle too
loudly around other people.  Nobody else will enjoy the tunes, and you
may see some eye-rolling in your vicinity.

Most of us cannot afford to listen to an opera 20 times before we can
appreciate it.  We find it easy and enjoyable to absorb songs with
beginnings, middles, and endings, larded up with familiar tonal
formulas, and with all manner of redundant signposts everywhere.  Those
features help us to "get" it, and most importantly to be *confident*
that we're getting it.  We humans need that stuff.  The redundant
features facilitate the drama's spirit-purifying power, which of course
is the whole point of the artform.

DTD syntax is like that.  It's as boring as Bellini's *Norma* for the
pit musicians and the singers, but the audience goes home whistling the
tunes, having heard every one of them several times, and each time with
plenty of what might, in another field, be called "Cyclic Redundancy
Checks".

I claim that the best music is the most-whistled music, and I think XML
isn't being whistled nearly enough.  It should be facilitating the
purification of our public and private institutions, as those
institutions move through their dramas of revelation and adaptation.
Instead, our efforts have been far too focused on machine-to-machine,
system-to-system communications.  Machines do not whistle.

>  
> 
>     If the purpose of a programmer's task is to make machines facilitate
>     human communication, nobody should care how hard that task is, least of
>     all the programmers
> 
> 
> Perhaps not the programmers, but definitely the people who pay the
> bills, because
> hard = expensive.
> 
> -- 
> GMail doesn't have rotating .sigs, but you can see mine at
> http://www.ccil.org/~cowan/signatures