Thursday, October 23, 2014

Philosophy of Word

Escape from Microsoft Word by Edward Mendelson in NYRB:

The original design of Microsoft Word, in the early 1980s, was a work of clarifying genius, but it had nothing to do with the way writing gets done. The programmers did not think about writing as a sequence of words set down on a page, but instead dreamed up a new idea about what they called a “document.” This was effectively a Platonic idea: the “form” of a document existed as an intangible ideal, and each tangible book, essay, love letter, or laundry list was a partial, imperfect representation of that intangible idea.

A document, as Word’s creators imagined it, is a container for other ideal forms. Each document contains one or more “sections,” what everyone else calls chapters or other subdivisions. Each section contains one or more paragraphs. Each paragraph contains one or more characters. Documents, sections, paragraphs, and characters all have sets of attributes, most of which Word calls “styles.” [...]


  1. Anonymous said...


    The paragraph you selected doesn't speak of something that is unique to Word. The issues would apply equally well to any other word processor, and then, also to a typesetting system like Tex.

    Instead, the following passage from the same article perhaps better captures the actual problem that is distinctive to (a certain version of) Word, as in contrast to, say, LaTeX:

    Word, it seems, obeys the following rule: when a “style” is applied to text that is more than 50 percent “direct-formatted” (like the italics I applied to the magazine titles), then the “style” removes the direct formatting.

    The root of such problems lies in the fact that a word processing/typesetting/etc. software---let's call it the software for writing (SfW)---must carry two classes of mark-ups: (i) those for a logical division of a document (into parts, chapters, sections, etc.), and (ii) those for controlling the graphical look and feel of the output, or, the "style" of formatting.

    The real trouble is: both these elements can simultaneously operate on the same textual matter.

    In the event that they do, then, the problem is: how does the software resolve the conflict?

    The TeX and its derivatives (e.g. LaTeX) take one extreme end of the possible positions: in their case, the logical division reigns supreme. It completely controls the graphical (or stylistic) attributes.

    Certain other packages took the other extreme position: there was almost no provision for any mapping from the logical mark-up to the graphical mark-up at all. For instance, the early WordStar had only stylistic mark-ups (Ctrl K B for bold, and Ctrl KS for underlining, etc., for none for section vs. sub-section.)

    HTML vs XML provide a later analogy. HTML 1.0 had only stylistic markups, and XML has always had only logical division markups.

    The commercially most successful packages (e.g. Adobe and Mac products, and MS Word) have chosen to explore the entirety of the space in between these two extremes, but sometimes with strategies that are not at all obvious or intuitive to any one. The author points out one particularly bad example with certain version(s) of Word. No other package would do something like what he points out. (Not even Microsoft Bob!)

    But, frankly, even though the author displays a seemingly good command of the philosophy of Platonism, I don't quite see how it applies as an explanatory basis for this aspect of the software packages.

    No, on the other hand, both the above-mentioned logical extremes, and every choice in between (e.g. Style Sheets + XML) is really driven by the simple need to have some machine-interpretable means, simply because the task is being automated.

    You can't attribute Platonism simply because you have to have some abstract scheme or design to be agreed upon, if any automation is at all to be achieved.

    It would be like saying that the Royal Enfield Bullet motorbike has been designed in reference to Platonism because its gears are on the right hand-side, or the fact that one bike imposes first gear down and the other gears up (the old Escorts Rajdoot), or some other bike imposes all gears down (the old Ind Suzuki), etc., is an evidence of a Platonist engineer at work. No, it isn't. These are just illustrative of some choices that a designer of any machine must make, whether the machine is "hard" (e.g. bike) or "soft" (e.g. computer program). Simply because automation is always dumb, an abstract logical scheme concerning its processing or functionality or usage-style must first be agreed on.

    ... Anyway, Happy Diwali to both of you and your readers.

    [I will seriously think of posting this comment at my blog, as a separate blog entry, too.]

  2. Ungrateful Alive said...

    I am so thankful I am in the hard sciences.