Word Processors: Stupid and Inefficient
ricardo.ecn.wfu.edu
ricardo.ecn.wfu.edu
Word and OpenOffice can be used intelligently, if you avoid the design-as-you-write trap. I've never really seen a diffing tool (for prose, mind you) with the readability of a word processor's 'redline' or trackchanges feature. Anyone have recommendations on that front?
The post's tone is objectionable, I suppose. I'm sure LaTeX is not for everyone. The learning curve's too high for your standard high-school kid writing an English paper, and the payoff is minimal in such cases since word processors have many of the same features for managing bibliographies, header formats, etc. But compared to just typing it in plain text, using a word processor feels horrifically obtuse to me.
I just press the ä key.
If I were planning to first-draft a long document and didn't want to be interrupted with nonsense, I would write it in markdown and render it to whatever output format using pandoc.
Nonsense. When you program, does your code look like meaningless.jibber(jabber)? I certainly hope not, and your TeX shouldn't look meaningless, either. TeX is a programming language, and you need to learn to use it effectively. If you are doing non-trivial formatting, then abstract out your directives and macros. Give them meaningful names.
That said, I actually agree with your points about LaTeX, and I gave up using it a long time ago. Plain TeX works exceptionally well, though. I even wrote a small and straightforward macro package for typesetting screenplays. It has a \direction macro, for typesetting script directions; a \speaker macro, for typesetting text spoken by a speaker; and a \scene macro for switching scenes. Aside from these directives, the entire script source is readable plain text.
For most people, structure in writing is visual. It's a paragraph because it has an indent or a blank line before it; it may also represent one complete idea or all that good stuff they say in English class, but that isn't what makes it a paragraph. It was that way before they started using word processors, and it'll continue to be that way - any other means of describing structure is always going to feel like a distraction to most humans with a need to put one word after another.
That said, most people don't care about typographical quality. It satisfies their emotional needs for a memo or letter to look typeset at all.
For these people, telling them to take what's now a one-step process and turning it into two steps, to gain some supposed advantages that either don't demonstrably matter or that you won't ever persuade them to care about, is not gonna fly.
That said, for those writers who do care about structure and typography, Ulysses looks pretty cool.
For long documents, there is a big opening in the market.
An ideal word processor would have powerful features for organizational structure. This means drag and drop organization of chapters, subsections, and the ability to add meta-data to them—main characters, affected departments, whatever is appropriate. Color codes attached the the metadata would allow at-a-glance determinations of length, flow, etc. It would also be able to create visualizations of use of names and terms, and assess reading difficulty. The table of contents of your document should be a map that helps shape the work.
It would also need a context-sensitive thesaurus, powerful grammar tools, and revision control in the tradition of programming revision control. I should be able to cut a chapter and effortlessly bring it back in later in the manuscript two revisions later.
There are a few programs that attempt the basics of this for Mac, and nothing that I can find for Windows. Not including typesetting is just the beginning, and only necessary because it gets in the way of the rest.
That's where a tool like Scrivener or Ulysses comes in, one that gives you the ability to rearrange the constituent units of your piece (paragraphs, chapters, scenes, arguments, whatever) to shape and reshape the overarching structure and logical flow. (Besides, both Scrivener and Ulysses allow you to use markup and export to LaTeX, etc., so you're not necessarily losing out on the formatting benefits.)
I think that those ad-hoc, one-page documents represent a majority of documents created by users - even if you count the pages, not the complete documents.
<META NAME="GENERATOR" CONTENT="TtH 1.93">
Presentation is a whole other step that you get to when you've actually written something worth reading.
That said, I think having a catalogue of suitable elements for each context and a (modifiable) document structure tree is better than either of the above alternatives.
I'm more for than against the author's main point fo separation of concerns, even in the realm of text processing, and I think especially for corporate communications that have to be all branded and specified, a LaTeX style document is probably the way to go. However, the overhead of having to code your document, instead of write it, is a big interface barrier, both in terms of adoption, and in terms of productivity. Also, in the cases where you do care about presentation during the construction of your document (which do arise), having a sharp divide between the two processes makes for a lot of annoying context-switching. I'm sure I'm not the only one with a horror story of having to compile a TeX document dozens of times to make a damn picture align correctly. Word isn't perfect in this respect either, but my worst-case horror story is not nearly as bad (5 minutes of drag-and-dropping).
You can, if you are careful, achieve effects such as changing the appearance of all section headings with one command. But few users of Word exploit this consistently, and that is not surprising: the WYSIWYG approach does not encourage concern with structure.
Instead, I think if more document processors took the approach of the early DTP apps, users would have a much easier time. Quark XPress 3.3 is an excellent model for how to do it: boxes, that you put on a page and then fill with text or pictures.
If you're just bashing out text, it can works largely as a word processor does today, just with one text box filling the page. But if you want to position or format things with any accuracy, you can manipulate them directly into boxes without the heartaches of doing so in any contemporary word processor.
It's a very simple conceptual model, easy to grasp and easier to use. I skipped both of my parents straight past word processors and after the vaguest of introductions to XPress they're now both producing complicated documents that are effectively impossible to create in Word.
Of course, something like XPress takes processing power, and it wasn't available in the early GUI days -- typing a document straight into XPress or Pagemaker was an exercise in frustration. So developers and users went down the "enhanced typewriter" route of MacWrite.
If we could start again, however, Mid-90s-style DTP would be the way to do it -- you don't need the complexity of something like InDesign: just a few simple tools and a floating palette for fonts and styles. Most of the common frustrations people have creating documents today would just vanish.
Apple has taken steps towards this model with Pages, but they haven't had the gumption to go the whole hog just yet, and what Word Processor DNA remains hampers it a bit. But it's going in the right direction.
Awesome program. Cumbersome, but ungodly powerful in the right hands.
There is no good reason to compromise nowadays.
The only problem is that LaTeX is typically required for math/science publications.
At this point, it becomes clear that this makes no damn sense at all for most people. Who wants to add steps to their writing process, add a delay between when they make a change and when they see it, and learn all new software to do it?
And then there's the remark about conserving disk space, which might have made sense in 1999, but it's a worthless consideration now.
If your document can be ugly, you can just open word and get cracking. The occasional weird formatting problems that show up in word aren't a huge deal. But if having a proper document is important, you need to either actually know latex or actually know word. Knowing latex is way easier than knowing how word works. MS made word easy to use by hiding all the gears. This is a problem when the gears don't work how you need them to (which sadly still happens too often).
The number of people for whom formatting matters this much is probably limited to publishers and academics, so I guess it probably doesn't matter much in general.
1. No other program really, truly understands it, although large steps have been made in other "word processors".
2. Can't do a "diff" between versions.
These two factors mean that "Word" format is absolutely and truly useless for any kind of versioned text. Whether "gears" are visible or not doesn't come into it.
How come MS Word is not able to detect that there were no changes?
It's more aggravating if we take into account how popular, old and big the application is.
By ignoring the rich rendering capabilities of WYSIWYG word processors in favor of simple monospaced text, a potentially valuable dimension is lost.