Two German psychology professors, Knauff and Nejasmic, recently published a paper on the benefits of Microsoft Word over LaTeX. The paper was reported in Nature.
They show that you can copy a simple document (containing little mathematics) faster and with fewer errors when using Word. Of course, those of us using LaTeX are aware of these shortcomings. LaTeX is not suited for quick-and-dirty jobs.
I should stress that Knauff and Nejasmic did not have authors compose a document, let alone craft an actual research article, let alone a sophisticated scientific article. They do not even try to assess the tools in the scientific workflow (data generation, analysis, processing, figure generation, and so on). They compare Word and LaTeX on a data entry job akin to what you might ask from a secretary. They also make no attempt to measure how much of this type of purely secretarial work scientists do… or whether it is representative of what scientists do.
It is clear that Knauff and Nejasmic have been frustrated by their collaboration with computer scientists that expected them to use LaTeX. In this story, they are not objective observers. Their frustration runs deep as they urge publish to restrict or even ban the use of LaTeX. Their main message is in the conclusion:
We therefore suggest that leading scientific journals should consider accepting submissions in LaTeX only if this is justified by the level of mathematics presented in the paper. In all other cases, we think that scholarly journals should request authors to submit their documents in Word (…) 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.
According to Knauff and Nejasmic, scientists who use LaTeX suffer from cognitive dissonance. To help them and improve science, we should force them to use Microsoft Word.
Given the obvious methodological gaps in their manuscript… how do Knauff and Nejasmic answer? Basically by describing LaTeX users as an irrational sect:
It is astonishing how some commentators ignore the basic principles of scientific decision-making that is, collecting facts, control over variables, using systematic methods, careful measurement, connecting causes and effects, and making rational evidence-based decisions, instead of generalizing personal impressions or opinions. (…) Why do so many people disagree with our conclusions? (…) from the beginning on we were aware that the issue is a highly emotional issue for many LaTeX users. (…) we think that the passion is a special habitus of the LaTeX community.
In the answers to the comments, their main objection regarding LaTeX is collaboration. They write: Word offers the helpful track changes tool, which makes collaboration very easy and efficient. In comparison, LaTeX produces text files. Text files are silly things that do nothing on their own. Except that we know a lot about how to collaborate on text files. It is called version control. Version control allows many people to work a the same time on the same files. There are builtin conflict resolution mechanisms. Everyone has instant access to the latest version of the file and there can be no ambiguity about the revision history of the document. You should be using version control in any case, to store your data and your software. You are not relying on data that has been saved on one of your students’ hard drive, are you?
I applaud Knauff and Nejasmic for trying to improve the productivity of scientists. But I give them a falling grade because there is an unacceptable gap between their conclusion and their actual experiments. It might be that your productivity as a scientist depends critically on whether you use Word or LaTeX for producing research articles. But their experiments tell us nothing about this question. Their paper is an opinion piece, not science.
36 thoughts on “Knauff and Nejasmic recommend banning LaTeX”
You are considering Knauff and Nejasmic as scientists, expectations they are not up to live for. IMHO they are fine test subjects, to explore the cognitive limitations due to which some people are less efficient in LaTeX than in Word. Think about all the time that could be saved in bureaucracy (filling up fields in plain text!) and the money that would be saved (except for Microsoft, but they will be fine anyway). Such research would be crucial, if only it was possible to tell whether the fault is in the software usability or the user incompetence.
@Leonel do you organize your ideas based on how do they look like when written (in paper pages) or based on what they *mean*?
I divide my papers in sections and sections in files. I use macros and commands for everything, with representative names and avoiding boilerplate. Reading the source files is often easier than reading the paper (at least for me).
As usual, there is an “easy” way to find out: split the research in Word and LaTeX adherents and let’s see which group is more productive.
I wholeheartedly agree with you. I have never had a basic text editor suddenly reduce my 100+ page document to a single dense blob in the corner of the page, but word has done this to me – repeatedly and repeatably.
I have never successfully used version control on a word document. I have never had any success creating dynamic contents and indices in word, nor cross references.
Latex consistently produces clean formatting, without me having to remember fonts and sizes, without worrying about page breaks etc.
Perhaps I have been overly influenced by the emotions of losing my data in word, or spending hours checking my page numbers for creating tables of contents, or tearing my hair out with weird formatting errors, or having to split my documents up into chunks of no more than 30 pages in fear of corruption.
When tracking changes are enabled, earlier or later MS Word fails to save the file. It is that simple: you can’t save the file.
Furthermore, formulas turn into images. I had it in my practice. My friend’s PhD thesis was once completely destroyed this way (and they had to retype it in Latex). According to reports of some of the Word users, formulas still keep disappearing here and there.
Another problem is that links are not working properly. No matter how carefully your proceed, eventually some links get broken. People who use Word often employ a professional editor who does two things (1) fixes broken stuff in the final version (2) reformats documents if they need to be resubmitted to a different conference/journal.
I tried to use Word myself and, unfortunately, came to a conclusion that this is an editor that simply doesn’t work. It is that simple.
I’d challenge these authors to produce a report with inline citations, figures on a specific page, tables at a specific locations, and then reorganize the sections and citations. All in Word!! Without uttering a single curse!
If they can do it without resorting to uttering any 4 letter words, I’d hang up my LaTeX hat.
Until then, LaTeX it is!!
(I have nothing against word … it’s nice for writing say a letter. But for a report/scientific paper with citations, Word is a horrible and time consuming way of doing it. Takes up 10 times more hours to fix a word document, when this can easily be done in LaTeX in a jiffy)
In general, in my writing, actual typing and typesetting takes a tiny fraction of overall time. It wouldn’t have helped me, if this time were zero.
In general, in my writing, actual typing and typesetting takes a tiny fraction of overall time. It wouldn’t have helped me, if this time were zero.
This is the gist of the matter for me and it is my main objection.
We spend a lot of time deciding what to write and how to write it. Once you do have the text in mind, the actual writing is not hard.
It could be that this is just a false impression, but then their paper offers absolutely no evidence to the contrary. It does the opposite: it makes the implicit assumption that copying and formatting a text is representative of a significant part of what we do.
A thoughtful response to a flawed study.
I was brought up on document preparation via mark-up and consequently find Word and its equivalents terribly unintuitive and difficult for large or complex documents.
Having said that, I have no problems using Word or similar tools for simple documents.
A properly-configured editor, with autocomplete for LaTeX commands and handy snippets for frequently used structures such as figures, tables, etc, goes a long way toward addressing the issue of document preparation speed. Add version control in there and you have a much more powerful and sophisticated environment for writing documents collaboratively.
It’s a step in the right direction. Use of Latex ignores the fact that visual formatting is not just for pretty effects. It is part of the writer’s mental organization of ideas, and there is mutual interaction between writing text and formatting it. Using Latex, whose side-by-side alternatives are far inferior (at the moment) disregards this and assumes that nothing needs to change from the electric typewriter days.
Sure, we need better experiences. But I fear Latex aficionados won’t like the results.
Then use LyX – it does both the WYSIWYG of Word and the structure of LaTeX together.
Here is another review of the PLOS one paper. Seems like the author agrees with your points …
There are usability issues with LaTeX. I find some of them deeply annoying.
However, objectively, Word behaves much more like a typewriter than LaTeX. Word processors have evolved straight out of typewriters.
@Daniel Word processors evolved obviously from hell. Until they make them stable, any talking about usability of Word is void.
“It could be that this is just a false impression, but then their paper offers absolutely no evidence to the contrary. It does the opposite: it makes the implicit assumption that copying and formatting a text is representative of a significant part of what we do.”
Knauss and Nejasmic coded incomplete documents as errors; if you look at their data, some LaTeX submissions for the table exercise were empty and had hundreds of “errors”. As far as I can tell, they never acknowledge this choice in their paper, and it accounts for most of the variation in error rates between LaTeX and Word users (relating to their finding that LaTeX is not only slower but more error-prone).
That being said, are their findings that hard to believe? It is genuinely difficult to typeset tables in LaTeX. Obviously LaTeX wins when it comes to mathematics, which they verified. I’d rather typeset tables in LaTeX than equations in Word, though; since most of what I write has equations, Word is a non-starter.
The authors also claim at some point that although the output from LaTeX is more attractive than that of Word, the gap has narrowed considerably. It is actually pretty easy to produce ugly documents with LaTeX; I have no sense of what the most beautiful Word documents look like. In any case, most journals are not prepared in Word, even if they accept Word submissions. (I assume they use InDesign, Advanced Print Publisher or some similar commercial product.) It could be that those journals have to do more work typesetting Word articles and preparing them for publication than those that publish LaTeX articles.
There’s also latexdiff, which works brilliantly 🙂
I’ve worked on group-written grants with Word many times, and it is awful. People are forever changing fonts, spacing, and other formatting. They have many different ways of doing citations, some of which are not compatible. I have at times copied the whole paper into the clipboard, then pasted into a text-only editor to remove all the formatting, then put it all back into word and formated it consistently. Ugh.
Comparing Latex and Word is like comparing a movie and a movie script… Certainly editing Latex directly is not time-efficient. That’s why there’s interfaces on top of it, like Lyx. I recommend anyone trying it once, it’s extremely good in particular for making Latex snipplets like complicated equations or tables.
@trylks, I and you and the entire human species (apart from the possibility of existing a rare individual with a specific, distinct mental organization) react to our environment. Having larger or smaller text, having it indented or slanted, having more or less spacing or kerning, etc. – all of this REALLY impacts meaning. And the use of capital case in the word is a basic demonstration of the fact.
Alas, it is not easy in the least. As research in human-computer interaction has shown repeatedly, humans are more productive using the tool they are proficient with. And benefits from proficiency are not directly related with benefits for novices. So removing bias from such an effort is a challenge.
As for your proposed setup, again, it depends on how you measure “productivity”, first and foremost, not an easy task (is it number of pages? is it clarity of arguments? is it the persuasion effectiveness of arguments?) and definitely not one depending on only one variable. But more critically, that proposed setup would only yield correlation. Not causality. Regardless of the results.
If the authors of this paper spent as much time in learning basics of LaTeX as they spent in formatting this paper in MS Word, we would not have to read through this.
There is no question that LaTeX has usability issues.
Having larger or smaller text, having it indented or slanted, having more or less spacing or kerning, etc. â€“ all of this REALLY impacts meaning.
I agree that presentation and content are not so easily separated, but that is not entirely in favour of Knauff and Nejasmic’s argument. Experienced LaTeX users tend to produce documents that are more finely tuned. Most Word users barely know about unbreakable spaces, let alone thin spaces and so on. Many of us create LaTeX documents with internal hyperlinks so that if you click on a reference it brings you to it, and so on. LaTeX makes it relatively easy to fine tune your presentation. If you want to use Word to achieve this same level of fine tuning, you will have to work hard.
Accordingly, Knauff and Nejasmic state that professional typesetting is easier with LaTeX. But then they dismiss this with the following remark:
we think that the appearance of the text is secondary to the scientific merit of an article
So which way is it?
It could be that those journals have to do more work typesetting Word articles and preparing them for publication than those that publish LaTeX articles.
In many of the journals I work with, papers are expected to be publication-ready by the time they are accepted. So the LaTeX document provided by the author is actually what is published with sometimes a few minor changes.
@Daniel I agree that the shortcomings presented about the K&N paper make it almost irrelevant – but not entirely, of course. The concept itself of collecting empirical data is what I meant by stating “a step in the right direction”. It is about the need to seriously confront the almost cult-like status of LaTeX in some circles, which tend to dismiss non-LaTeXers with disdain, ignoring HCI issues as trylks as done above, pretending that only the content matters. And the way to do it is with empirical data. So I can only hope more rigorous inquiry is made subsequently.
I fear the problem is open-ended… Most Word users would have less of a problem if they cared about non-visible chars (paragraph marks, line breaks, etc.) and entities (styles, style hierarchies, etc.). Sadly, most don’t care the least about that, only about the looks… and I wonder if using LaTeX with a WYSIWYG editor wouldn’t originate a similar situation.
For what it’s worth, I create my equations using LaTeX notation, generate pictures from that, and then insert the pictures in Word.
*has done, not as done, sorry.
My main criticism is that the authors are not comparing like with like. LaTeX is a markup language, not a word processor, or editor. There are many tools that can be used in conjunction with LaTeX, including WYSIWYG editors that let you see the processed LaTeX document as you enter the text.
In their comments about cost they make no mention of the fact that Word is a commercial product that has to be purchased, whereas LaTeX and most of the associated tools are free.
@Sven, people have staff editors just to deal with Word quirks. I guess it greatly outweighs costs of buying the software itself. Why do they need them, you ask? For two reasons
1) Reformat papers for another journal/conference
2) Fix a lot of things that almost inadvertently occur when Word is used to create large and complex documents. In particular, the editor fixes broken links.
I agree with @Leonid. As much as I prefer free software, the Microsoft tax remains relatively small. Moreover, though I would not know how, I am quite sure that pirating Microsoft Word is not terribly difficult.
Hi Daniel â€“ I’m chiming in late here, I know, but I’m in the process of discovering and loving your blog, so all your posts are new to me. (Also, I had no idea WordPress could be used to such great, minimalistic effect.)
I agree with you about the ecological validity of the K&N study. However, I think LaTeX is much worse than they argued, and that more rigorous, ecologically-valid research would not help LaTeX’s cause.
To use *TeX, one must use *TeX software, and that software is just devastating. It’s as though all these *TeX distributions have asked our civilization for a special exemption from the customary standards of software usability, quality, and even size constraints â€“ and that a small corner of civilization has granted this exemption.
I’ve never been able to install a distribution south of 1 GB in size. I’ve never been able to have anything go smoothly with any of them. MikTeX simply wouldn’t work. I have TeX Live on a Windows 10 machine now. I’ve never been able to get it to successfully update packages. Nothing happens when I click on the button. This is typical of TeX distributions I’ve tried. The user interface looks like something from Windows 95 or an old XWindows metaphor.
Though I installed something analogous to a “minimal” bundle, scrolling through the packages I see things like “Fonts for making barcodes”, “Bibliography style for Chalmers Institute of Technology”, and “Bullshit bingo, calendar, and baseball-score cards”. I’ve never understood why TeX distributions do this to people, why they install this enormous bloat of styles and schemas for specific organizations and subfields. It’s such a terrible way to package and distribute software.
I’d love to see more data â€“ or any data â€“ comparing typesetting quality between Word and LaTeX. I don’t know of any systematic testing there. People have lots of beliefs, and K&N are surely right about the religious and evidence-free zeal of many LaTeX proponents. The current version of Word is 2016, and one problem I encounter is that LaTeX fans only ever speak of “Word” non-specifically, seem to have an automatic ten-year timelag when they refer to Word (or Windows), and often turn out to be referring to Word 98 or 2007 or some other very old version. I’d like to see blind tests of modern Word typesetting compared to modern TeX distributions, along with looking at usability other dimensions.
I’ve never been able to install a distribution south of 1 GB in size. (…) I’ve never been able to have anything go smoothly with any of them. (…) I’ve never been able to get it to successfully update packages.
My favorite editor is texpad and it is quite smooth (https://www.texpadapp.com/). It even works on my iPad.
I realize that you are probably a Windows user, so you can’t buy texpad.
This brings me to my next point.
I find the usability of Windows quite a bit worse than what is offered by Apple. My Macs “just work”.
It’s as though Windows has asked our civilization for a special exemption from the customary standards of software usability, quality…
We can’t deny the religious and evidence-free zeal of many Windows proponents.
So, really, for their own good, Windows users should be forbidden to use Windows and force to upgrade to Apple products. Think of all the productivity savings!
Daniel – Your criticism of the Knauff and Nejasmic paper is that the task does not fully depict the activities of composition, editing, and publishing. You also criticized its unwarranted recommendation, based on the results, that journals should only accept submissions in LaTex that are heavily mathematical. I agree with both points. Two problems you didn’t mention, which I think are the main limitations of the study, are the self-selection of participants and its measure of expertise.
As to the first, an intimate relationship develops between user and gear. That’s true whether the instrument is a carpenter’s tool, a chef’s knife, a musical instrument, or a piece of software. To the extent that people use what they are comfortable with, they put up with inherent limitations and even come to embrace them. A case in point is LaTex users, who are using a typesetting tool to edit text. They are not unlike spreadsheet users who compose text documents in Excel, or a wine aficionado I know who keeps his wine database in a flat file.
I suspect that some of the school spirit that LaTex users exude is owing their having been forced or expected to use a piece of gear that wasn’t right for them, then wanting to spread the word once they found LaTex. The result of the Knauff study that most interested me was that notwithstanding their comparatively poor performance on the basic typing task, LaTex users were more satisfied with their typesetting program than were the users of word processing software.
As to the second point, the authors could have developed a practical test of expertise. Instead, they assumed that persons qualify as an expert if they have used a piece of software for over 1000 hours. Many experienced MS Word users think of the program as you do — as an advanced typewriter. That’s a novice view. Another novice belief is that Word is only WYSIWYG. I wonder what the performance results would have been had the two groups been screened for expertise more adequately.
No study is perfect and I think that despite its limitations, Knauff and Nejasmic have done an interesting piece of work. I do wonder, as you do, about their policy recommendation. If a journal is willing to accept LaTex submissions that are heavily mathematical, then why draw a line around “heavily?”
You don’t fight science with a blog post; you fight science with more science. You could even add a little note at the end of your paper saying that it was written using LaTeX.
I presume you are assuming that science is whatever happens when people publish peer-reviewed articles in journals?
I agree with the assessment that it is an opinion piece, and not a scientific one, or a meaningful study.
So here’s my opinion: Word is crap. Even in the class of programs it belongs to (WYSIWYG word processors), it is, at best, 3rd rate. You cannot use Word as a tool for writing â€“ you operate Word to hopefully generate an output that resembles the manuscript you envision. You have to adhere to a strict discipline, akin to magical enchantments, to keep Word doing what you want it to do. It is incredibly buggy, and you need to know and learn the hard way where the bugs lurk, lest they jump out and bite you in the ass. For instance, Word does not have dedicated workflows for applying styles just at the cursor, or everywhere. Or editing a style. Or just editing the style for your selection, and nowhere else. If you don’t follow the procedures for each case exactly, anything else can, and does happen.
I have written a diploma and PhD thesis in LaTeX, and loved every minute of it. I could actually write, because I didn’t need to care if the numbering, formatting and look would come out fine â€“ I knew it would, and any error I encountered where my own typos, easily found.
That said, I don’t use LaTeX anymore, unless I have to do loads of math.
Neither do I use Word. There are better tools out there â€“ unless you stay on Windows, where Microsoft has raped the environs clean of anything but Word clones, which do behave the same shitty ways.
Nowadays, I use Scrivener (agreed, that’s on Win, too) to work on complex documents. For WYSIWYG, I use Nisus Writer, Mellel, or Pages (on my Mac), and can actually write using those tools, and do not need to summon demons to keep Word at bay. (Pages works surprisingly well, and can deal with LaTeX-math in beautiful way.)
Oh, and here’s the most surprising thing: I have written for publication in scientific journals. With, sigh, Word, since .doc is the only file format all journals accept (some hard core physics journals accept LaTeX, too).
Except that there’s nothing in Word you actually need. Most journals don’t accept embedded graphics â€“ the quality is, mostly, too bad. You’ll have to submit separate TIFFs. Nor do you need to use styles. You can, but most journals prefer you don’t. Because the .doc file is mostly just a raw substrate, and nothing of its functions is used in the actual publication process. From what I told, most journals hire a student intern who copy’n’paste from Word into Adobe InDesign (or, very very rarely, LaTeX). You do not need Word for all that. All you would need is a plain-text editor. Maybe with Markdown, but not more. You only needed Word for that pesky .doc file â€“ but all the 3rd party programs that are out there got so good at it recently that they can generate a .doc file that can be opened at the other end. In fact, they gotten better at it than Word itself, since they avoid the undocumented or badly documented parts that Word uses and that cause much of document crippling.
I have only recently found the Knauff and Nejasmic (2014) study. There is one glaring error in their statistical methodology that completely invalidates their methodology. In statistical terms this error is known as self selection bias. If one was to do a comparison of Word and LaTeX the participants should have been allocated to Word and LaTeX at random. Perhaps one would then need to provide some instruction and to control for previous experience. This would not be easy. My own thinking would be that users of MS Word are likely to do a lot more typing than users of LaTeX. I would also suspect that they type faster. The results of the study may be confirming this and not making a statement about the relative merits of Word and LaTeX.
LaTeX users are said to have made considerable formatting errors. One of the advantages of LaTeX is that one inputs the raw text to LaTeX and leaves the formatting of the output to LaTeX. It would be almost impossible to use LaTeX to reproduce the exact format produced by for example a Word Processor. Trying to reproduce such results in LaTeX would be very time consuming, reduce throughput and lead to other errors. This would contribute to the bad results by the LaTeX users.
Clearly this is an example of statistics done wrong from the point of study design and statistical analysis. The results are worthless.
An interesting approach to writing using a computer is the one taken by TeXmacs (http://www.texmacs.org/tmweb/home/welcome.en.html), which combines structured and WYSIWYG text composition. Here is a presentation of the systems written by its author, Joris van der Hoeven: http://www.texmacs.org/Data/TeXmacs.pdf, where he explains why combining the two “paradigms” is useful and possible.
I have seen other work in the same direction, as far as I know TeXmacs is the most complete system at the moment (extensible with macros written in Scheme).
You may subscribe to this blog by email.