🚀 go-pugleaf

In article <1919@glacier.ARPA> reid@glacier.UUCP (Brian Reid) writes:
>
>If you look at the interaction of technology and industry for the past few
>hundred years, you will see a recurring theme. A new technology gets
>invented. The in-place industry applies that technology to automate what
>they are currently doing. This is often inefficient, as the new technology
>is often better applied by changing the fundamental premises of the
>industry. Gradually new companies grow up, which use the new technology in a
>different way, and if it is more cost-effective, then the new industry
>drives the old one out of business.
>
   ...
>
>At the moment we are in a transition phase. The graphic arts industry is
>discovering computers, and they are molding them in their own image, taking
>the things that they have done by hand since the invention of cold type and
>putting them isomorphically onto the computer. Simultaneously, however,
>thousands of businesses are discovering that they don't NEED the graphic
>arts industry. With simple computer tools they can achieve their end
>results--the publication of books or newsletters or catalogs--without
>graphic artists. If history serves as any guide, then in half a generation
>the traditionalist approach will no longer be competitive and will have to
>pull out of those markets completely.

Brian is absolutely right that new technology tends to replace older
industry, and does this by delivering cheaper products.  The
operative word, however, is "cheaper", not "better".  Sad to say,
cheaper usually also means worse.  The new technology survives
because (1) it's usually only slightly worse, and so most people
don't care, (2) capitalism works well with mass markets, since by
definition the average customer is not as demanding as the more
discerning customer, and (3) eventually people get acclimated and no
one remembers the advantages offered by the older (higher priced)
product.  [Incidental note:  the new technology may also offer other
advantages, such as smaller size or weight.  Again, these are
engineering improvements and do not directly relate to the quality
of the final product (it being understood that the product is
*produced* by the technology, the technology is not the product
itself).]

This is exactly what we see with document processing systems and
technical typesetters.  Pick up a copy of Ullman's book on
Databases, typeset with TeX.  The typesetting is inexcusably bad!
(The standard TeX fonts are also terrible, but that doesn't affect the
typesetting.)  Why then was TeX used, rather than a conventional
typesetting?  Almost certainly it was to lower the cost of producing
the book, with the attitude that TeX output was "good enough".

(I have heard that Don Knuth developed TeX in response to his
publisher's statement that re-typesetting second editions of Knuth's
books would be too expensive.  I believe this to be true, but I
cannot remember the source.)  

By the way, don't take my word for it;  get Ullman's book and try
reading two consecutive chapters.  Don't just skim them (I suspect
you will find yourself wanting to do this, because of the
subconcious resistance to the bad typesetting), but make yourself
read and try to digest the book as a text.  Measure the results by
how you respond to the material in the book, and how little you were
bothered by the typesetting.  (In a perfectly typeset book the
typesetting completely disappears, so that it is never noticed by
itself.)

Of course it is not fair to judge a typesetter by only one of its
uses.  Look around.  Almost all of the documents (books, papers) I
have read that were typeset with TeX are awful.  If this is the
future of technical typesetting, I don't want it -- and it doesn't
matter whether it was done with TeX, Scribe, WYSIWYG, or chiseling
stone tablets.

Rather than continue in the style of a debate, let's look at the
good points of each of the two approaches.

WYSIWYG is good at:
	local things (i.e., a screenfuls worth)
	appearance
	user feedback

Text processors are good at:
	document structure
	textual computation (referencing, indexing, etc.)
	preserving intention

I see no contradiction in integrating both sets of good points into
one system.  What we lack is a good language to express both sets of
things conveniently.  What the WYSIWYG people (I confess I am in this
camp) ought to be doing is trying to find out how to incorporate the
good features of text processors into interactive systems.  That way
we wouldn't need those document processors (to be fair, we wouldn't
need any of the existing WYSIWYG systems either), and we could all go
on to more entertaining and more productive discussions.