from Hacker News

An Efficiency Comparison of Latex vs. Word (2014)

by uger on 12/26/21, 1:45 PM with 38 comments

  • by cs702 on 12/26/21, 5:23 PM

    The authors tested it on researchers working in Physics, Psychology, Computer Science, Mathematics, Electrical engineering, Business (MBAs), Sport Science, and others.

    But... they did not break down results by field (!), and also did not ask the researchers to compose a complex, highly structured document akin to something that they would actually publish (!), and also did not ask the researchers to reorganize a structured document repeatedly, as is typical in research (!), and also did not ask the researchers to recreate tables multiple times with updated data (!).

    Not a very useful study.

  • by fsh on 12/26/21, 5:35 PM

    I think the methodology of this paper is quite weak and the results do not support the strongly opinionated conclusion the authors draw from them.

    The first example is a plain text with a few footnotes in a two-column scientific journal layout. The participants were asked to "reproduce" this text, but the article does not explain what is meant by that. Is it a mistake if the font sizes and spacings don't exactly match? This would certainly explain why LaTeX users had a lot more "formatting errors and typos" and wrote less text (presumably they spent a lot of time trying to google a suitable template). This is not representative of typical paper writing, where you either have a template provided by the journal or you just pick something you like.

    The second text are some tables and they get the unsurprising result that making tables in LaTeX really sucks.

    The third text is heavy on equations and they get the unsurprising result that the Word equation editor really sucks.

    From this they somehow conclude that using LaTeX is a waste of taxpayer money.

  • by beloch on 12/26/21, 5:45 PM

    Some concerns with this study:

    1. 40 is not an adequate sample size. Not even close.

    2. Continuous text entry in LaTeX and Word is very similar, so why are LaTeX users far behind Word users here? Did the researchers measure typing speed and control for that, so that the Word group wasn't given a set of typists who were, on average, faster?

    3. For the equation entry portion, did they control for user expertise when assigning participants to groups? e.g. Asking a psychologist to typeset a physics equation is going to result in slower work and more mistakes. With a sample size of 40, an imbalance of just a person or two could easily throw the whole study off. (Note: They did find that LaTeX was better for equations, which is not surprising. Their methods are still suspect though.)

    4. This study does not address the specific tasks most people use LaTeX for: papers and books. Both are longer than a quick and dirty copy job, and such jobs benefit greatly from LaTeX's ability to produce consistent output when edited. Word documents break in all sorts of strange ways once they become long and complex enough.

  • by JCWasmx86 on 12/26/21, 4:38 PM

    LaTeX by itself may be less productive than Word, but it has (at least for me) several advantages that make it for me more productive:

    - The standard unix tools (E.g. grep) do work really well with LaTeX, if I have to find something. I would now know how to find a string in like 30-40 .doc(x) files

    - Works with git

    - Programmatically generating LaTeX is really simple as it is just text (I use this e.g. in my Vala Project. Sure for other language there are libraries to generate .docx files, but LaTeX works for me here better, as I generate dozens of diagrams and so on)

    - FLOSS and fully documented

    - Documentation is far better (E.g. How do I do X? The results for Word are often spam-like)

  • by mcswell on 12/27/21, 2:45 AM

    I'd like to see the test people try to do automatic numbering in Word. I'm a long time Word user (like 30 years), and numbering in Word has NEVER worked right.

    A few weeks ago, I was tearing my hair out trying to get simple list numbering (1, 2, 3. Some non-list text. 1, 2, 3, 4...). Word seems to think that if you start numbering a list at the beginning of a doc, and then you go back to Body Text (or the horribly named Normal text), and then twenty pages later you start another list, this next list must be a continuation of the first list. What idiot decided that? You have to right-click on the first list item of the second list and do "Restart numbering at 1". And even then I was getting weird numbering of list items, like 1, 2, 5. No visible reason, and very hard to fix.

    Legal section numbering in Word? Like 1, 1.1., 1.2, 2, 2.1,... Don't even try. It used to be a built-in, but it was so buggy they removed the built-in. There are instructions on the web for how to create such section numbering in Word ex nihilo, but the instructions are so complex that I haven't even tried. (And probably just as buggy as the list numbering.)

    Both list and legal section numbering work out of the box in LaTeX, of course.

  • by zdw on 12/26/21, 5:14 PM

    A good analysis of why this article is BS: https://lemire.me/blog/2015/01/14/knauff-and-nejasmic-recomm...
  • by marcodiego on 12/26/21, 5:49 PM

    This is a pro-word article. The irony: it needed corrections for size and placement of the figures: https://journals.plos.org/plosone/article/file?id=10.1371/jo...
  • by mturmon on 12/26/21, 10:22 PM

    This article is trash. They can learn nothing by the methods used.

    There were 40 subjects, not randomly selected, and they had no way to normalize for skill or motivation.

    The tasks are mimicry, not scholarly document preparation.

    The evaluation metric is not really even given.

    And the biases of the researchers are evident — last line of the paper:

    “preventing researchers from producing documents in LaTeX would save time and money to maximize the benefit of research and development for both the research team and the public.”

    There is no way the data they have is sufficient to reach this conclusion. What an embarrassment.

  • by infogulch on 12/26/21, 5:42 PM

    Does the Word comparison include spending hours fiddling around fixing indentation, formatting, and headers in 10+ page documents? When the whole document becomes a patchwork of conflicting formatting so any change you or any collaborator adds has some incomprehensible default that you thought you purged from the document weeks ago? The time spent merging changes made by 3 people at once and emailed around with document names that have no coherent pattern, where half of the time you can't even tell what changed?
  • by dynamic_sausage on 12/26/21, 5:17 PM

    It seems this article actually measured how well people could use their editor of choice vs Word, which has nothing to do with LaTeX.

    I especially liked the claim that "expert LaTeX users performed even worse than novice Word users", which classifies a user as expert if they used either typesetting system long enough. It is pretty clear though that most LaTeX users have exactly the knowledge required to typeset the exact kinds of documents they normally author with maximal efficiency, rather than to typeset all kinds of documents. For an alternative measure of expertise, how many of the "expert" users actually read, say, Knuth's TeXbook? (not that they need to, of course)

  • by kevinventullo on 12/26/21, 5:29 PM

    I personally use the two respectively as the authors suggest: LaTex if there’s a lot of math, and Word/Google Docs otherwise (I will note that GD actually supports a surprisingly large amount of LaTex syntax in their equation editor; not sure about Word).

    However, their suggestion that journals should ban submissions in LaTex seems rather bizarre. It is not the journal’s job to ensure research is done as efficiently as possible.

  • by analog31 on 12/26/21, 6:50 PM

    Granted it was 30 years ago but when I was a grad student in physics, there were a variety of options for how to write your dissertation. From what I observed, my guess would have been that Latex takes about twice as long as Word.

    But it wasn't a fair comparison, because the students using Latex were also using Unix computers, and were genuinely interested in the nuts and bolts of the process. They were also settled on being able to push a button and have a beautiful book slide out of the laser printer.

    The students who used more primitive tools simply accepted that their formatting would be more primitive. I used an antiquated scientific word processing program called T3 on a MS-DOS machine, and covered for its deficiencies by fixing my manuscript with pen and ink before taking it to be photocopied.

    In my dad's dissertation from the 1950s, the equations, chemical diagrams, and graphs, are all drawn by hand. You wrote your manuscript by hand, and paid a typist to make the final copies after your committee had approved it.

    Today, I think formal typesetting is disappearing. They should compare Latex, Word, Jupyter, and the e-mail editor.

  • by chromatin on 12/26/21, 4:14 PM

    > On most measures, expert LaTeX users performed even worse than novice Word users.

    Big Oof

    > LaTeX users, however, more often report enjoying using their respective software.

    Ok so there's that as a highly redeeming feature =)

  • by i_am_proteus on 12/26/21, 5:16 PM

    >The texts were selected based on a pilot study so that an expert could reproduce around 90% of the text in thirty minutes.

    One big advantage of LaTeX over Word, for this user, is consistency in formatting for larger documents, which gets completely lost for a one- or two-page piece that takes just a half-hour to type.

  • by Finnucane on 12/26/21, 5:02 PM

    To match output quality, you would need to run your word doc through a separate typesetting step, losing your time advantage.
  • by ianbooker on 12/26/21, 5:00 PM

    "Efficiency" in this study is measured by counting errors in reproduced texts and that is a fair operationalization for "typesetting". Yet, it seems that the design omits a central variable as there is no information on spellcheckers. While its a easy default in Word, you can only speculate about the usage in thr Latex groups..
  • by zoomablemind on 12/26/21, 7:31 PM

    I thought in science fields most publications are based on styled templates (for Word or LaTeX). So there should not be any need to 'recreate' a styled layout from scratch.

    Figure and table placement is still manual per pub requirements, but page/section/font/size styling should be already defined.

  • by scrubs on 12/26/21, 5:17 PM

    Latex+pdf ... Ultimate combo. Case closed.

    Now ... imagine Bill Murray with a tie and glasses who then gives you the follow pep talk:

    For the slow pokes on latex we recommend coffee, 80s R&B to help you get in there and get done.

    Have a hot date setup ahead of time ... This helps you to not play with every configuration and possible import option. Yes latex is awesome but that not that awesome. Keep it simple. Don't let latex weakness build into relationship problems by running late. That takes time away from the keyboard later.

    Also remember the words you write are almost as important as how latex displays them so keep some of your focus in reserve for clarity, thesis, stuff like that. Latex is awesome but alas clarity counts too (unless you heard something different? ... Please IM me asap. Clarity isn't so simple it turns out).

    Well that's a wrap. Another problem solved!

  • by mmcgaha on 12/26/21, 4:45 PM

    So you have 1000 man hours of research and you worry about how long it takes to write the report? As far as grammar goes, yeah I guess the grammar checking of Word helps. The number of words produced is a bit of a strange metric unless we measure the importance/quality of research by the pound.
  • by bodhiandpysics1 on 12/26/21, 4:45 PM

    I'm not sure it was a good test... Latex users do appear to be significantly faster at mathematical typing... which is obviously Latex's strong suit. What Latex is in general weaker than word at arbitrary formatting... its goal is to provide semantic markeup for large scale academic writing. Also, apparently its difficult to use Latex's table system compared to word (though tables in word are particularly easy!)
  • by cozzyd on 12/26/21, 4:50 PM

    Yeah until word supports vim bindings it's not going to be very efficient.
  • by noisem4ker on 12/26/21, 4:45 PM

    (2014)