The trouble with TeX

I am involved in a Twitter discussion about this classic article about why one should eschew Word processors and instead choose to use a markup language like TeX. The basic argument is that creating a text is and should be a separate task from typesetting it, but programs like Microsoft Word spuriously combine the two. The result is crappy typesetting and a constant distraction from the document’s logical structure to how it looks.

Fair enough! To all those who use TeX, I wish you nothing but the best. But I agree with Voyou that TeX does not actually solve the stated problem — instead, it adds a whole additional layer of making you learn a clunky mark-up language. I tried to use TeX for a while and found it to be incredibly frustrating, not to mention the fact that it’s a pain in the ass to set up on a Windows machine (since you have to set up a Postscript translator to actually print your document). And I’m kind of a best-case test here, since I have always been fascinated by the more hands-on aspects of computers and in fact used Linux as my sole operating system for a long time in college.

Word processors might not force you to think about logical structure, but I assume that the TeX mark-up language does not actively disallow “manually” formatting headings in detail rather than using a heading style. The solution to not thinking about logical structure is just to think about logical structure, and all standard word processors provide tools adequate to the task: namely, styles. If average users don’t use them, the solution can’t be to make them learn how to use this much more complicated system, but rather to convince them to use them. I use them consistently, and the results are of course absolutely amazing.

Similarly, the solution to worrying too much about formatting during composition is not to worry about formatting during composition. I leave all styles in their default state, then I can easily manipulate them for printing.

Finally, for the vast majority of documents like class papers, etc., there’s no need for good typesetting. The fact that double-spacing is the standard should be a hint that camera-ready “beauty” (and don’t even get me started about how obnoxious it is that the party line on TeX is that it produces “beautiful” results) is not desired.

For those documents that do need typesetting, i.e., published texts, most often your publisher is going to take care of that. My experience with books is that the copyeditor will put in the markup relevant to the publisher’s particular system. They could give it to me in advance, but then they’d still have to check it, right? Nothing would really be gained. (Things may be different in math or natural sciences, but it’s notable that none of the people I’m debating with work in either.) Why not just use the good old “Heading 1” and “Heading 2” styles in Word and be done with it?

People might lament the fact that publishers often require submissions to be in Word format when they could have the beauty and elegance of TeX, but maybe that’s their way of saying that they prefer to take care of the typesetting element on their own — which they should, since they’re the experts.

So other than people who are self-publishing or those who are using it for something that it’s particularly great for (such as math), I don’t see any advantage to using TeX. I mean, use it if you want! For the stated reasons, however, it seems to me that it’s no better or worse than Microsoft Word — and using Word requires less intellectual overhead than TeX, which in turn seems to provide no genuine value-add for the majority of uses to which humanities scholars would put it. (There is the matter of citation management, but that’s not inherent to TeX itself — Word has add-ons that can do the same thing.) That is to say, it doesn’t seem to me that the time spent learning the TeX mark-up language would ever “pay for itself.” If you tell me that I can use an automated system, then it seems like you’re just telling me to use a word processor again, and thus there’s nothing to keep me from following the path of least resistence and manually formatting headings.

There is, in short, no necessary connection between using TeX and focusing on the logical structure of the text. Indeed, for most users it’s initially going to add a significant additional amount of distraction. The payoff, on the other hand, is unnecessary (at least for scholars in the humanities) — if you have something that needs professional-grade typesetting, 99% of the time you’re going to have a professional typesetter doing it for you.

For my own personal purposes — and as readers know, I’ve worked on a lot of fairly large-scale writing projects — Microsoft Word 2003 is more than adequate. It’s customizable and it provides all the organizational tools I need. (I refuse to upgrade to later versions because they’ve removed the macro features.) Turn off all the auto-formatting options (which honestly should be off by default), use the pre-defined “Heading X” styles consistently, and you’re golden. Or if you don’t like using the tacky corporate software and really like the hands-on aspects of computers and programming, use the mark-up language. Either tool is more than adequate to the task, however, if properly used. Pitting TeX properly used against Word sloppily used is stacking the deck.

46 thoughts on “The trouble with TeX


    But, for the moment:

    (since you have to set up a Postscript translator to actually print your document)

    TeX and LaTeX and friends can produce pdf output directly.

  2. That is a better system than was available (or at least than I had found) nearly ten years ago when I tried to set it up. It still preserves the necessity of a separate program to print.

  3. I’m glad to have left behind a long-running argument between my previous employer and myself and a web-designer colleague over whether we would be more productive using Dreamweaver than hand-coding HTML. Thank goodness he at least had the sense not to try to force Dreamweaver on us.

    At least Dreamweaver will let you drop out to the markup if you want to, though. IIRC, WordPerfect also used to let you see the underlying formatting codes, and correct them manually if the need arose. I dislike very much the fact that Word doesn’t have a similar feature.

    I’ve used Word’s styles, outlining etc. for years, and found them usable but beset by small frustrations. Everything has small frustrations, of course, but what amplifies the small frustrations of using Word is that the GUI won’t get out of the way and let you fix the things it breaks on. Word has various crap ways of getting its knickers in a twist, and provides no clean and obvious way for you to untwist them. Users who expect no better will perhaps not find this so infuriating, but it annoys me intensely.

    I would also like it, a lot, if Word’s underlying markup was one that could be processed easily by standard text-processing tools – if Word documents were good citizens of the command-line universe, where you pipe text out of one program and into another. The ability to write macros in VBA is rather poor compensation for not being able to do this (and I’ve written a *lot* of macros in VBA…)

  4. I definitely prefer coding HTML by hand rather than using any kind of WYSIWYG tool (particularly in-browser tools), precisely because they fuck things up and produce ugly, inefficient code. Microsoft Word is obviously one of the worst imaginable offenders.

    I know the little frustrations of which you speak, though they basically no longer arise for me (and I’ve become adept at fixing them when they arise in others’ documents I need to edit). The option to drop down into a plain mark-up language and straighten things out would be a beneficial feature, and the fact that Word has a binary document format is a pain in the ass for the reasons you mention. I may not be a good sample, though, because for the most part I rarely have more than one level of headings within any given chapter or article. Most of the small frustrations, however, seem to me to result from doing what the presenting article says one should not do initially — namely, fiddling with formatting in too much detail.

  5. It seems to me that the problem TeX treats is obsessive-compulsive disorder with respect to formatting. I have this problem, but I’m also too lazy to learn new things.

  6. I use LaTeX for lecture notes, presentation notes, letters, course outlines, and my dissertation. I use TextEdit in plain text for things (e.g., publishable) that others will read. Having said that, the integration between LaTeX and BibTeX is far superior to anything I’ve ever seen for word processors: I wish journals, for instance, would accept LaTeX documents.

  7. The more or less irenic tone of most of the post makes it hard to get too up in arms about (which is no doubt too bad since Adam probably wanted this to increase his traffic), but it seems to me to overstate the difficulties of using TeX and its relations (and concomitantly never really to tell us what the trouble with TeX actually is).

    It is true that switching to TeX from Word involves learning something new, but that seems not to be of much moment as a complaint; switching to anything from Word would, even things like Markdown or reST, which are much lighter-weight than TeX, involve learning some new things. And for me, learning to use “the good old “Heading 1″ and “Heading 2″ styles in Word and be done with it?” would also be learning some extra stuff on top of “just typing”, especially adding in whatever is involved in keeping track of what bits of text are styled how (presumably bits of text with different logical styles could come out on occasion being visually styled the same way), keeping styles consistent, etc. (I have no idea how easy or difficult it is to do this in Word or other WYSIWYG editors, since I mostly don’t use them, but however one does these things, it’s still “a whole additional layer”, just one which isn’t represented by things one types—I’m also not sure what you mean by a whole additional layer. TeX isn’t on top of a Word-style WYSIWYG system.)

    It is also pretty easy to get started using TeX using a micron of the resources technically at one’s disposal. I am still hardly a LaTeX guru, but when I started using it I basically knew, and needed to know, only the following:

    – an invocation to begin and end a document: “\documentclass[12pt]{article} \begin{document}” followed at the end by a bibliographic invocation if necessary (since forgotten) and “\end{document}”. I’m calling these “invocations” since you don’t need to, and I didn’t, understand at all what they’re for, you just put them there and Bob’s your uncle.
    – To get italicized text, put it in “\textit{}”.
    – To get a footnote, put it in “\footnote{}”; for an endnote, “\endnote{}”.
    – To get an ampersand, write “\&”, not “&”.
    – Paragraphs are done like this:
    “this is one paragraph.

    here is the next.”
    – And a passel of stuff related to the non-bibtex bibliographic facilities.

    My editor took care of correctly converting ” to either “ or ” depending on context.

    And that’s it! This is neither a particularly clunky markup nor particularly hard to learn, and it’s easy enough to add to it. (If I’d been using LyX, presumably the old keyboard shortcuts would have worked.)

    I have since become somewhat more sophisticated, but not much more, and the markup one actually has to use is often pretty high-level, and things are set up so that it’s much easier to work at that level. So, while it’s true that there isn’t a necessary connection between using something like LaTeX and a high-level approach to the structure of the document (after all, the high-level macros are also implemented in TeX, at least until one gets to whatever is simply primitive in the system), so that instead of using, say, \section{} to indicate a major section in what you’re writing, you could do everything the already-present \section{} command gives you by hand, that would be an absolutely insane way of proceeding. (If you actually wanted to change the way \section{} did something, you would either redefine it, or write a new command in terms of it, so that what you actually used in writing your document, call it “mysection{}”, would still be high-level; doing everything piecewise at each point at which you wanted a section-marker (or whatever) would also prevent you from taking advantage of other packages that hook into the already-present and canonical name.)

    My displeasure at having to use Word or Wordalikes in preparing documents for submission isn’t that I want to be able to control the typesetting. I don’t control the typesetting: I make practically no changes to the default LaTeX behavior in almost all of my documents.

    I do add some behavior to the default: a command \ibid{} that if invoked with no argument generates the string “(ibid.)”, or with one (e.g. \ibid{p 3}) the string “(ibid., [argument])”, and for my dissertation one to aid in cross-referencing footnotes that generates “footnote N” for same-chapter references and “footnote N, chapter M” for references to notes in a different chapter. (One does “\nref{descriptive-note-label}”.) (The built-in cross-referencing facilities suffice for referencing sections, etc.) The only marginally formatting-related thing I did was add a bunch of stuff to create a nice-looking analytic table of contents, where the precis for each section is defined immediately after the section itself begins, rather than where the TOC goes: so you have:

    \subsection{This is a subsection}
    \subsectionprecistoc{Several objections are considered and shown to be absolutely meritless.}

    And the precis shows up in the TOC with the right font size, alignment, etc.

    Instead, I prefer editing plain text: it’s easier, surprisingly, to put off worrying about formatting when the formatting just ain’t there as one types. (Not that there isn’t some information presented visually.) One can edit it with whatever editor one pleases. It’s easy to deploy unusual characters or subscripts or superscripts or whatever. “φ1” -> “$\phi_1”. It’s easy to use with revision control systems. There’s a built-in citation system. (Yes, there are citation systems that work with Word, but that would be another layer to learn, doncha know.) If there’s a Wordly way to refer to footnotes or sections independently of just writing in whatever number they happen to have at the time (so that if a new footnote/section is added in earlier, the reference becomes incorrect), I don’t know it.

    My preference mostly has to do with the plain-textual nature and the fact that I am using explicit markup—not with the appearance of the output. (Though, well, I do prefer it.) I want to be able, for instance, to write the text of the footnote or endnote in the same place as the text to which it’s attached, not at the bottom of the page or end of the document or wherever. I want to be able to get en and em dashes without the word processor getting in my face about it. I’d like to be able to make lists and sub-lists explicitly, so that I don’t have to try to outsmart my word processor to get the behavior I want. Etc. This has nothing to do with having control over the formatting in the real ultimate typesetting sense. (Markdown and reST don’t offer that, for instance, but they also allow (some of) what I’m after here.)

    So, uh, basically, I still don’t know what the problem with TeX is or why it provides no benefit over Word. The answer can’t be that you have to learn TeX but not Word, since I would have to put in a lot of effort to become as proficient with Word as you are (styles? what’re those?), and TeX is not hard to learn for day to day use; and it surely can’t be that TeX doesn’t make it impossible to fiddle around at the low level.

  8. I’m also a huge LaTeX fan. My reasons are basic formatting, as Word can definitely do those things without learning something new. However, I’ve found the ability to put in internal references (e.g. ‘see p. 32 above’), changing citation formats (and overall handling of citations), indexing, and the occasional advanced typesetting (e.g. while I’m a theologian, my thesis has a few mathematical equations on display).

    I’ve used TeX for all articles, lectures, notes, my CV, and my thesis. I have the advantage of my primary computer being Linux-based, but I generally use Kile in KDE. For Windows (as my laptop is a dual-boot and I get lazy), there are now some quick-and-easy installers (e.g. MikTeX) for the TeX base and some decent GUIs (e.g. LyX). I’ve found that there are plenty of packages around for TeX that most styling can be dumped into the preamble (or made a separate command), leaving the document itself mostly pure text. I do like, also, that TeX can handle multiple files, which makes for re-arranging chapter layouts a bit easier.

  9. Does anyone else use OpenOffice? I’ve used it since my freshman year of college and it works fine for me. Pretty basic, intuitive, and doesn’t come with all kinds of ridiculous pre-loaded general settings. I’m repulsed whenever I try to use campus computers with Word 2010 or whatever. Plus, OpenOffice is an open source program (I would guess that TeX is as well).

  10. With MS Word 2003 don’t you have huge problems with RTL text? And Word 2010 is not much better. Just different bugs. But I agree – you’ve gotta forget formatting sometimes. Presentation is a whole artistic skill of its own. Get it down – clean it up – then decide if it’s worth presenting or is just more delightful compost.

  11. I’ve never used RTL text.

    The cross-referencing features in TeX do seem better. I’ve gotten close to getting cross-references to work in Word, but then given up in frustration — thus undercutting my point above.

  12. A more important point: going to a lot of trouble to get the cross-referencing right seemed useless given that that would be lost in the transition to whatever system the publisher was using. With that in mind, my cross-references tend to be to chapters (which are unlikely to change places in my writing process) or sections (which I refer to by title rather than number) — and my cross-references are minimal in any case.

  13. Well, it isn’t a lot of trouble. I say \label{foo} in one place and then \ref{foo} in another. Various packages for emacs make it easy to quickly survey all the labels in context if I forget what one of them is. And sometimes I want to send stuff to entities other than publishers who’ll have an interest in checking to make sure the references make sense (which a copy editor might not be up to, at that—if I say “as we saw in section whatever”, is it the copy editor’s job to ensure that section whatever is indeed the section in which we saw that?)—for instance, to referees, or as a writing sample, or what have you.

  14. Well, that does seem simple enough — but odds are if you wind up publishing any of this, you’ll need to do the cross-reference by hand, likely on the proofs (or in copyeditor queries).

    For my uses, the primary documents I create are either documents to submit to publishers, in which case the added TeX structural information is going to be lost, or lecture notes or conference papers, in which case there’s really no need to worry about section headings and all that — even the article advocating TeX says that there are documents for which it’s overkill.

    The one big appeal is citation management, but I don’t find that to be a big problem — it would take a really long time for the time spent learning the system to “pay for itself” in saved time, especially given that my “default” citation style matches up with most publishers and journals in my field or else is close enough that copyeditors can take care of the changes.

  15. Speaking of citations (first sparked by ben’s comment), I find two packages in LaTeX indispensable: biblatex-chicago and csquotes. It means learning a few different commands, but I like the payoff. First, csquotes adds the \enquote{} and \textquote{} commands, each which put quotation marks around the text. The difference is that \textquote has an optional argument to put in a citation. The beauty of the two is that it switches between American and British punctuation (as well as any other language) differences, handles nested quotes, and moves ending punctuation where it belongs for the language. The other packages handles citations (generally through the \autocite{} command) as well as the ibid bit. Also (for ben), I’ve read that using \emph{} is better than \textit{} because not all outputs will be able to handle italics (granted, for our purposes, it will be almost always italics).

  16. MS Word nearly killed one of my postgraduate theses, and I wasn’t prepared to take the risk with the other one. Plain text files are portable and pretty much bombproof. The formatting (the hyphenation engine, in particular) is a bonus.

  17. Awkwardness looks like it was laid out using Word—is that right? Amongst other things, the text has no ligatures, and seeing the words “The Office” used over and over is enough to make type nerds claw at their eyes.

    My more or less considered view on the Word vs TeX thing is here and at greater length here. Or, more or less, what Ben said. The typesetting is a nice benefit, and certainly a potentially giant time-suck. But having everything rooted in plain text is at the core of it for me.

  18. XeTeX handles unicode splendidly (making languages other than English much easier to use), allows the use of most (if not all) the fonts installed on your system (not just the TeX-friendly fonts), and is included with most TeX distributions. Also, LaTeX documents only need a few heading changes (the rest of the document can be normal LaTeX, so no changes are necessary).

  19. Ultimately, whatever works for you :)

    TeX has two big advantages (for non-mathematics use)

    1) Lots of documents are written to be shared, but will not be typeset by professionals. When people with the visual sense of dead trout write using TeX, they produce something reasonably pleasing. When they use Word…
    Hence the self-interest behind the evangelism.

    2) TeX documents, being plain text, are much more readily and transparently converted, shared, put under the version management system of your choice, easily useable even after 20 years have passed, etc… Something like Markdown also does the trick, for simple stuff.

    I’d second Kieran’s comment, I also suspect a lot of “professional” documents are produced in Word these days to save costs.

  20. I’ve typeset camera-ready copy for a publisher before as a paid copy-editor/typesetter [using LaTeX], and also taught the Word for Thesis course at the institution where I used to study/teach. Given a choice, I’d choose LaTeX any day, although you can do more or less anything using either. The learning curve for LaTeX isn’t that steep, you can learn it in an hour or two. Maximum. And it has certainly saved me a lot of working hours since.

    I expect it comes down to what you prefer, and whether you are a text-editor or wysiwyg type of person. Word is/was a decent enough package before they broke it and would still be my first choice for correspondence or short things; anything over a couple of pages and I’d go to LaTex, without a second though.

  21. The post misses a major point of (La)TeX/LyX (or comparable systems like lout), which is that, used by author and publisher, they obviate a lot of work. Things like:

    “maybe [saying publishers should use TeX]’s their way of saying that they prefer to take care of the typesetting element on their own”


    “but odds are if you wind up publishing any of this, you’ll need to do the cross-reference by hand”

    The point of using LaTeX at both ends is that you don’t have to do either. You can write your LaTeX file, and the publisher can apply their stylesheet, and, without changing your file, with the probable exception of fixing a few page and line breaks, typeset the result. As for cross-references, I’m mystified: why on earth would you ever want to do them by hand? This is the 21st century, we have machines that do that…having said that, I have proof-read journal proofs which, the submission being in Word, have come back to me as PDF, which is basically like living in a souped-up version of the 20th century where paper has a search function. Cross-correlating a proof, a list of corrections and an original document is still necessary. What a waste of time! I presume the publisher didn’t typeset using Word because they found it unusable.

    When I translated a book written in LaTeX, the publisher sent back a list of corrections and I made them in collaboration with the author. Two documents to cross-correlate: better, but still less than ideal. (LyX allows Word-style inline corrections.) Note however that because author and publisher used the same system the division of labour was more flexible, which works well for academic books where the budget is smaller and corrections are less easy to make without adding errors, so they had better all go through the author in any case.

    But LaTeX is definitely not just for would-be typesetters: its major strength for the publishing process is precisely that it encourages you not to fiddle, and if you don’t touch the stylesheet (“class” in LaTeX-speak) you can ignore the details completely.

    In nearly ten years of encouraging non-computer-scientists to use LyX, I’ve had three major sticking points, usually fatal:

    1. You can’t import from Word. At least, not without being a techie (Word→RTF→LaTeX+a certain amount of tedious fiddling).

    2. You can’t export to Word (aka “the journal doesn’t accept LaTeX”). Again, there’s a techie route to this, but it’s not pleasant, and necessitates much work if repeated revisions are needed.

    3. BibTeX isn’t as good/easy-to-use as Endnote. I have not used Endnote, but from what I’ve seen this is true; I’ve tried various front-ends to BibTeX but they’ve all been rather Mickey-Mouse, which is a pity as BibTeX itself is plenty powerful; it just needs the LyX treatment.

    I might be a little out of date, since I stopped using LyX 10 years ago (though I have followed its development) and haven’t dealt with bibliographies in anger for the same length of time. Being happy, as a programmer, with LaTeX, and always seeming to require features that LyX doesn’t directly expose, and working almost exclusively with short documents (rarely longer than a few pages), I am happy just to use LaTeX in an editor that has good support for it. LyX’s major annoyance is that for LaTeX feature it doesn’t directly support it is harder to use than a good text editor (because it just leaves unknown LaTeX commands in plain red, and has none of the understanding of their syntax that good editors have). However, LyX’s grasp of LaTeX, even ten years ago, was plenty broad enough for anything not requiring specialist typography and had a lot of support for individual specialist packages; this support has broadened considerably since then.

  22. Ok, this is a thread I have to add my two cents to. I probably used LaTeX for as long as some of you have been around (since 1988), and I was quite adept at it. I still think it’s great, or all the reasons it has been defended here. If one is doing a standalone document like a dissertation that is not going to be sent to a publisher, it beats all other systems by far. However, the minute you want to get the document published, you are in the seventh circle of Hell. First, you have to destroy the beauty and simplicity of your document in order to get it into Word format. Second, you have to use Word. Adam is right that Word works mostly ok, but when it doesn’t, well, just abandon all hope and enter the dark woods.

    I still don’t use Word, though I have given up LaTeX. I use OpenOffice. I loathe and despise the Word layout. I admit OpenOffice has some drawbacks (you can’t change footnote to endnotes, so you need to Convert them in Word), but it is Purgatory compared the the Inferno of Word.

    And, finally, the hand of God has reached down into Hell to rescue us: RefWorks. It is better than BibTeX (where is the translator field in bibtex?).

    For historical purposes, here are the systems I have used in my grad student to now life: (1) 1978: IBM Selectric with a Greek font wheel (very expensive); 1981: IBM computer, two floppy disks, no harddrive, DOS, and WordPerfect; until (3) 1991 (same IBM), when I was dialing up to get access to my Unix shell account at the university, using LaTeX; then (4) 1991: a new PC, running Linux, with LaTeX; until 1998 (5) a new PC, running Linux, and OpenOffice; until (5) 2009, a new PC running Windows, and Puppy Linux on the CD. I simply could not do without Word any more on my own computer. I was never happier than when I dialed up to my Unix shell. It was cloud computing avant la lettre.

  23. (you can’t change footnote to endnotes, so you need to Convert them in Word)

    You can actually do this with the find/replace dialogue, IIRC.

  24. where is the translator field in bibtex?
    It’s used in biblatex-chicago (and I’d wager in BibLaTeX generally) as is the URL access date.

  25. I am a total geek and I have never, even for a moment considered learning TeX. I just don’t see the point. If I want to write a document straight out in markup I use Markdown which is so easy to use it is insane. To create short texts I use Textmate and then, when I am done, copy and paste into Word, format and then its call over. For my thesis, I use Scrivener – something I have been encouraging someone on this blog who uses it (eg myself or Anthony) to write an account of for a while – it is actually writing software, as opposed to a Word Processor and is very impressive. Once I am done here, I’ll slap it into Word, if I am feel fruity, In Design, to give it a final polish – I put temporary citations in using the Endnote format straight into Scrivener, then run the bibliography maker in Word to finalise this.

    What conceivable advantage I could get out of having to type out markup into my writing as well as writing I really don’t get. If I do use Word, and want to make an section of text italic I just Apple-I, type, Apple-I – how is this hard? There is a pure geekiness element to writing in TeX, I’ll admit, but outside of this I just plain don’t get it.

    For the record since Word 2007 Microsoft have used the Open Office XML format for documents. Drop a Word document into a zip file opener to see the content – it is a bloody mess though reading it raw.

  26. What conceivable advantage I could get out of having to type out markup into my writing as well as writing I really don’t get

    I thought you said you used Markdown?

  27. I don’t. I use Scrivener, and occasionally Markdown – but I have so say *syntax* wise Markdown is much easier than the designed in the 80s TeX language.

  28. Oh, I must have been confused by the part of your comment where you wrote “If I want to write a document straight out in markup I use Markdown”. I guess you never want to write a document straight out in markup.

    I know that “from the 80s” is kind of a general purpose slam, but I wouldn’t have thought that its applicability could possibly extend to the syntax of artificial languages.

  29. I am surprised no one has mentioned Apple’s Pages. I’ve tried using Word, LaTeX, and Scrivener, but find Pages to be the best word processor for a couple of reasons:

    1) It loads documents (including Word documents!!) in roughly a second on my laptop, much faster than either Word (horribly slow) or LaTeX (slower, but not bad for a non-native app). Moreover, its impact on the amount of RAM used while running is basically negligible.

    2) It lets you easily switch between different languages assigned to the document, and also in terms of specific paragraphs. Basically what this means is that I could be writing a paper in German and then set the document’s default language to German so that Pages knows to format quotation marks, dates, and SpellCheck/grammar differently (or, if I am writing in English but citing German, I can select specific German blockquotes and change those on a case by case basis). I have tried to do this in Word–before I angrily trashed the entire Office suite–and it simply doesn’t work.

    3. The UI is incredibly lightweight, straightforward, isn’t bloated at all with useless features, and more or less “gets out of the way” during the writing process. There’s basically one toolbar and that’s it (aside from the popup toolbox, which more or less remains closed during editing).

    4. It produces PDFs (and even Word documents!) elegantly and quickly. It also does ligatures (e.g., Shift+Alt+5, Shift+Alt+6), and you don’t even need to master an esoteric bullshit language that is unreadable, cluttered with code, and basically unusable in its original format for the vast majority of people.

  30. I’m like Hill but Ben got me interested. Seems to me that it’s an advantage that what you type does not correspond at all to what somebody would read. It breaks the promise of something like Word that the two are related and therefore the point of compulsively obsessing about e.g. filling lines up with characters.

    Is there a command that forces the lines to be as filled as possible with characters even if that requires proposing a synonym or hyphenating?

  31. I don’t write much directly in latex either, these days. I mostly use org-mode, which exports to tex or html very easily. Citation management is the achilles heel of lightweight markup formats, though. If org-mode could export biblatex-style cites to HTML it’d be perfect. But it’s still very good. Pandoc’s citation processing is also very promising.

    What conceivable advantage I could get out of having to type out markup into my writing as well as writing I really don’t get. If I do use Word, and want to make an section of text italic I just Apple-I, type, Apple-I – how is this hard?

    Depending on the editor, regular users of tex don’t type out their markup, either: they have keyboard shortcuts, just like you.

    If your writing incorporates results from any kind of quantitative data analysis, the case for using plain text + latex gets much stronger.

  32. ben

    Jeez! If I wanted to write straight out in markup I’d use something like Markdown. I wouldn’t use TeX because I don’t think it is very elegant. Of course it extends to all sorts of stuff! (and I use the command line for near everything!).


    If no one really types out the markup, then I understand even less why anyone would code in it! Incidentally your website is very very cool.

  33. PS For the record, if I had my time again, I’d do my thesis with version control it with git and host it online at github. Scrivener apparently saves thing as a Subversion repository, but Subversion is from the devil.

  34. I don’t actually write anything anyone reads or would publish, but when I was trying to, I liked using markdown/pandoc. For simple stuff–humanities papers, for example–it seems perfect. Decent citation management, seamless export through LaTeX into attractive PDFs, and it’s all in a small, human-readable text file with *intuitive* formatting.

    I’m a bit nonplussed by the idea that publishers want Word documents. They don’t use Word all the way through the process, do they? I know many math/science journals want prefer tex files; which sort of publishers insist on Word?

  35. I’m kind of confused by statements like this:

    you don’t even need to master an esoteric bullshit language that is unreadable, cluttered with code, and basically unusable in its original format for the vast majority of people.

    Here is a LaTeX file; the text is Robertson Davies’ “The Ghost Who Vanished By Degrees”; here is the PDF it produces (complete with prissy ligatures, because—aside from the fact that they’re kind of intrusive—why not?).

    Is it possible to maintain that the tex file is “unreadable” and “cluttered with code”? Surely it is not.

  36. “IIRC, WordPerfect also used to let you see the underlying formatting codes, and correct them manually if the need arose.”

    It still does. This is why I still use it. (I’m too lazy to learn anything new.)

  37. ben

    If everyone is inserting the tags using shortcut codes, which is exactly the same as Apple-I, why is it any different in terms of user experience from using a Word Processor proper?

Comments are closed.