Friday, 09 June 2000
.....
Line breaks in text are a very hard thing to get right. People are pretty good at detecting bad ones; so publishers must respect that ability and provide good breaks, but they can also exploit the ability and simply use people to get them right...up to a point.
The problem is that bad line breaks pop up quite easily, and it's very, very tedious work to find and fix them all by hand over and over. So people are out, the lazy slobs. There just aren't enough calm, mindless, ravingly anal perfectionist types to go around.
So, as with most things we know are boring but would like to be cool, we turn to technology---in particular, to computers.
Alas, as Donald E. Knuth has written time and again, finding a good line breaking algorithm---a good algorithm for good line breaks---is very hard. He's spent a good chunk of his career working on it, and it's still hard.
It's hard (avoiding the technical language for a moment) because fixing one bad break can produce a slew of new bad breaks. One's small effort to do local good can, and typically does, cause bad to ripple out. Typical word processors---and, to their great shame, desktop publishing programs---tend to examine one line at a time when trying to fix a bad break. This just doesn't even begin to work well. Don't even think about hyphenation
(Knuth's TeX does a more or less good job, but---aside from mangling the kind of text I typically write, it still hits a lot of problems---especially with page breaks and floating figures. The advantage of TeX, of course, is that you can tune the line break algorithm. For the average designer, this is also the disadvantage of TeX.)
Conrad Taylor pretty much gets all this (and several other things) right on the money (boy, I've been wanting to use that phrase). Of course, things have improved! Adobe's newish desktop publishing program, InDesign delivers the following "extraordinary innovation":
[Y]our choice of text composition engines to help you achieve elegant text flow. The single-line composer considers one line at a time, while the powerful new multiline composer can compare and adjust multiple lines at once. Both composers follow built-in rules that specify how they should apply your settings to achieve consistent visual color with optimal line breaks. Unlike traditional typesetters - often considered superior to desktop typography - the multiline composer can look both forward and backward in a paragraph to produce the best results possible. With the composition engines in Adobe InDesign 1.5, you gain the most flexible high-quality typographical tools available today.
Of course, ad copy doesn't mention that the other contender for the "most flexible high-quality" typesetting program "available today" (to wit, TeX) has been around (and been Free) since 1982. Oops! What an innovation!
The power of today's computers dwarfs those of 1977, when Knuth started TeX. With the disk space, memory capacity, and raw speed of current computers, it should be possible to vastly improve on his work...if simply by brute force.
Ah! the computer industry...that golden land of innovation. (Not just innovation, of course, but extraordinary innovation.) Of course, Adobe is to be commended for finally bringing in advanced, automated line breaking. Who knows, it could very well be an improvement over Knuth's work (and that of others). It is a little disheartening, however, to see the massive effort and investment put into things of little or no substance, importance, or merit, while things of substance, significance, and true value languish.
Productivity-enhancing technology is a failure if it merely enhances profit, for, in the end, production must ground out in the world of our eyes, stomachs, thoughts, and lives. The "output" of a standardized worker alone should not be the output of our entire existence.
A book's beauty, as an object, is a wondrous thing which gives genuine pleasure on its own, as well as making more readily available the worth of the writing. In Knuth's work (see Digital Typography for a good sample, with much that is accessible to the laity), he notes that it is important that when we computerize something like typography, that we do not forever accept radically inferior results. I fear that too often, with computers and with the whole of our lives, we're becoming conditioned to accept radically inferior results, and taught to think them good.
This is Done badly, it will be hell <http://monkeyfist.com/articles/563>