Edward Tufte has a famous essay on the Cognitive Style of Powerpoint, which should be required reading for anyone involved in communicating basically anything. I think Rand has already linked to this essay elsewhere, but I'll link again just for emphasis. There's an excellent, if a little technical, essay here which covers some similar issues in word processing (hat tip to an anonymous commenter on this post).

The author makes three points about WYSIWYG word processing:

1. The author is distracted from the proper business of composing text,
   in favor of making typographical choices in relation to which she may have
   no expertise (``fiddling with fonts and margins'' when she should be
   concentrating on content).

2. The typesetting algorithm employed by WYSIWYG word processor
   sacrifices quality to the speed required for the setting and resetting of
   the user's input in real time. The final product is greatly inferior to
   that of a real typesetting program.

3. The user of a word processor is under a strong temptation to lose
   sight of the logical structure of the text and to conflate this with
   superficial typographical elements.
   

The technical communication tools we use direct our thinking about the problems we are working on into certain channels. Bad (or inappropriate) tools encourage bad thinking. Good tools make it easy to understand the semantic content of the communication. This is one of the reasons good tools are less popular than bad ones. Bad tools allow sloppy thinking to fly under the radar. Good ones make it harder to obfuscate, requiring a higher level of discipline and clarity of thought. It's possible to be clear in PowerPoint and Word, just as it's possible to say stupid things in LaTeX or ASCII. The thing that makes a tool good is what it makes easy (clarity) and what it makes hard (obfuscation).

The thing that is lost in many of these discussions of technical communication is that for the majority of users, ease of obfuscation is a feature, not a bug. Most people are average or below. They want to be able to pass off their work without subjecting it to excessive scrutiny. Tools which make this easy will always be more popular than tools which make it hard. The customer for the software is the person writing the BS, not the person reading it. The distractions of futzing with typesetting make it harder to focus on generating good content, but the flashiness of the presentation makes it easier to paper over the weakness of the content. It's telling that the second of these "features" is more important than the first.

Having bitched about the problem, I feel I should offer some attempt at a solution. In this case I think it comes down to little more than requiring technical memoranda be written in ASCII or LaTeX (or some similar method that separates content generation from presentation - even HTML might work). Of course, this implies that the boss wants technical memoranda, which is the root of the problem in the first place. The technical tools are just a low-order symptom. Another approach (of which this post is a sample) is to try to propagate the meme that excessive fondness for PowerPoint and Word is a warning sign for technical mediocrity. The CAIB report has certainly helped with this, as has the Edward Tufte essay.

The typesetting algorithm employed by WYSIWYG word processor sacrifices quality to the speed required for the setting and resetting of the user's input in real time. The final product is greatly inferior to that of a real typesetting program.

Huh? While I completely agree with the rest of the article and the essays, what is this doing in there? I was a typesetter for several years and watched that industry die because it thought it had some kind of artistic lock on what "good typography" meant. The only think typesetting really gets you is intelligent kerning and that has been built into word processing packages for a while.

But the point is that no one ever expected WYSIWYG to produce good typography since that's not what people use it for. It sounds like someone is lamenting the fact that no one is paying them to typeset newsletters anymore. Since when has anyone cared about the quality of the typesetting of a technical memorandum? Is he suggesting that a typewriter is a better instrument from a typographic standpoint?

That one entry cheapens the rest of the point by making it sound more like a pining for an old IBM Selectric.

Michael - IMO the output from earlier generations of WYSIWYG was noticeably uglier than custom typesetting programs like TeX, though it's gotten a lot better, but I think the point the author is trying to make has more to do with the fact that WYSIWYG compromises typesetting in part by getting the user involved in messing around with details while writing that are better left to the end of the process and then applied globally. I think the author is also conflating typesetting and layout, which confuses things a bit. The essay could certainly be tightened up a bit.

All this is no big deal if you're just pounding out business letters (which is one area where WYSIWYG is really the right way to go - lot of business boilerplate benefits from WYSIWYG, especially if it includes templates). The application to technical memoranda is my own spin on things - You really don't need to worry about typesetting for technical materials unless they contain a lot of math (where typesetting can make the difference between clear and impenetrable).

What??? The alternatives you present are all FAR more difficult to produce technical documents in than Word. The only good argument I've heard for the TeX variants is that, since you are PROGRAMMING your document and not writing it, you can get precisely the layout you desire. I, for one, have no desire to learn a programming language in order to simply write. (And if you make TeX or HTML easy to use, you've reduced it to a WYSIWYG word processor anyway.)

Besides, the days of people playing with bizarre font choices are largely 10-15 years in the past -- the novelty of WYSIWYG has long since worn off. That argument about people focusing on 'fonts & margins' is largely invalid. Using something like Word is a matter of 'fire it up and start typing'. Word and AppleWorks, etc. all pretty much are set up with reasonable defaults when it comes to formatting. And many organizations propogate their own 'Normal' template...

I *might* buy that allowing easy document production increases the amount of slop that's out there, but it also allows an increased amount of good content to be produced. I'm not exactly sure where the balance point ends up.

- Eric.

And another comment...

Long ago I did my PhD thesis in NROFF, which follows this absurd notion of 'separate the writing and formatting', because there was no alternative back then. And guess what -- it was nigh on impossible to discern the logical structure of the document without doing trial prints to read the thesis as it would look on output. You certainly can't examine logical structure when it's obfuscated with slash or dot codes. Written documents are a visual medium and their logical structure is primarily visual -- you can't really effectively separate the writing and the formatting, except perhaps for really simple things like short memos.

BTW, when one does the right thing and starts a Word document by using the outlining function, you ARE putting the logical structure foremost, but you can SEE the structure.

- Eric.

Eric - I'm beginning to think I should have spent more time on the post than I did - the result was a little sloppy. I blame the software :-)

Anyway, back to the point - for boilerplate, like the in house templates you mention, you are right - something like Word is really quite good, as long as you stay inside the box. For a heck of a lot of technical documentation plain ASCII is just fine, too (the majority of my technical memos are nothing more than long, detailed plaintext emails). The tipping point where LaTeX's steep learning curve becomes worthwhile comes only if you are doing either relatively complex math or doing something which the WYSIWYG programmer hadn't considered. I find that nearly every time I use Word for anything more complex than a form letter I end up fighting the program's assumptions about how things should be done. There's a learning curve for Word, too, if you step off the beaten track.

OTOH, simple TeX documents are very straightforward to write, you need to know about 10 fairly straightforward commands, though in that case, might as well use Word. It only gets really tricky when you start to typeset math, and all the math editors I've seen suck compared to TeX, with the exception of Scientific Word, which is just a glorified LaTeX front end. Scientific Word, incidentally, is semi-WYSYWIG, sort of What You See is Closely Related To What You Will Get When you Are Done (WYSICRTWYWGWYAD), while observing the separation of content generation and formatting. Unfortunately SW isn't available for Mac, so I'm installing LyX to see if that does the trick.

As for NROFF - ouch. I sympathize.

Case in point: The Internet Engineering Task Force has a long standing policy of all of its documents being in 7 bit ASCII, 72 columns x 58 lines. I've written 21 RFCs and it has made me a firm believer in making sure that you can express your thoughts in words long before you attempt to make a picture. A picture is worth a 1000 words, but you have to get those 1000 words right before you know what picture to use. Or, in the case of Powerpoint, what slide animation plus sound effect to use.

"the novelty of WYSIWYG has long since worn off. That argument about people focusing on 'fonts & margins' is largely invalid. "

BTW, go and listen to a beginners course on using Word. They start from stuff like how to bold your text and insert pictures. Use of styles and proper document structure is taught only on "advanced" courses.
And who wants to sit through a lengthy course, if all i wanted is to write a short technical paper, right ?
IMO, it would be a good policy in organization using Word or somesuch to _disable_ or _hide_ most of the features that do damage to proper document structure. Stuff like futzing with fonts ( i.e. not using a predefined style ) and page layout should be strictly forbidden.

I guess as a physicist, I have a much different feeling about what constitutes a technical document. As in, pretty much, if it's just plain text, it ain't technical...

I'd be interested to hear how LyX works, and I think I may look into it also.

- Eric.

i've been using latex for technical writing since 1987, and powerpoint for presentations since sometime in the mid-90s. one thing i've always hated is using the math editor in powerpoint (same as in word) when i've already typeset all the math in latex.

behold: tex4ppt is an add-in for powerpoint that allows the user to use latex snippets. it's a beautiful thing, and you can find it easily enough with google.

of course, i am speaking as someone who does not have any problem with powerpoint, and indeed views it as a useful tool for preparing presentations, something that most professionals, sooner or later, need to be able to do. i commented on this ppt issue at rand's earlier post, and have posted on the topic a couple of times myself.