An Efficiency Comparison of Document Preparation Systems in Academic Research
journals.plos.org
journals.plos.org
I have created academic work both in LaTex and in Word and while I have not measured it in an objective way I also had the feeling that LaTeX could create better looking results once you've got over the first hump.
On the other hand, the integrated nature of a system like Word were spellchecking, annotations etc. are all part of the main tool you work with makes it very effective. It takes pretty long however to develop a workflow that avoids all the pitfalls.
In a sense, LaTex makes you believe that once you got over the initial hump, you have typesetting super powers but you will miss out on some of the integrated UI stuff that may boost productivity. Word on the other hand makes you believe that you already know everything while you still need to keep fine-tuning your workflow until you don't mess up constantly.
They didn't measure the quality of the typesetting in any fashion, though, which is what the GP's claim was about; they measured the rate of errors typists made under time pressure.
It's also unclear what LaTeX software they used, since the one I use has integrated spellcheck, etc.
That's a hugely important point and in a way it shows that they are comparing apples and oranges. Word is a complete document editing system (well, more or less) while LaTeX is a type setting systems and you need additional tools to use it. Those tools can range from a simple plain-text editor (say, nano) to totally integrated environments that are similar to word in their feature set.
As it turns out since the authors' fee was per-copy, while the typesetters' fee was a one shot deal, they made much more as typesetters (though this situation might get reversed if the book keeps selling for decades…).
My experience with LaTeX is that it enables and encourages my OCD behavior and it is very easy to be sucked into an enormous game of package dependency hell and whack-a-mole to achieve some desired aesthetic result. Whereas with Word I already expect an inferior result so I'm somehow more easily happy getting 95% of the way there and calling it a day. It's not too hard to learn "what not to do" in Word and how to use counters and references. Certainly if you've got the capacity to learn LaTeX, you can learn how to markup Word properly and use styles etc.
It is in a paragraph saying "different tools for different jobs"
https://news.ycombinator.com/item?id=8807497
https://news.ycombinator.com/item?id=8797002
Most of the study is, to be honest, bullshit.
IMHO, those are honest goals for a test. Your target may be your university's (publisher's, peer's) expected format which will cause a drop in your grade (pay, funding) if you miss. I also feel that we are frequently time constrained when trying to get work done, so I don't feel that the criticism against that is legitimate.
Now then, this is certainly not the definitive study to decide which is better, Word or LaTEx, but it's certainly an important study for those who want to be productive and have a particular format for a document that they need to match.
But when the document gets huge: ToC, key phrase index, quotes half-way in the margin, output to several paper sizes, multi-page tables, both footnotes AND endnotes, source code listings, a large number of references to journals... In that case I'm glad I used Latex for the smaller documents, because now I'm acquainted to it in time for the big thing.
> multi-page tables
You won't be able to persuade me that's easier in LaTeX than Word! I say that as an otherwise very proficient LaTeX user who finds himself much more productive with LaTeX than Word.
As for this study, I imagine it's a question of proficiency–good LaTeX users will be better with LaTeX, and good Word users will be better with Word. I have colleagues who can only barely use LaTeX, so it takes them forever, and they'd be much better off with Word. We do all our paper collaborations using Word, for example (which drives me nuts but there we go).
1) LaTeX + revision control is better for multi-user concurrent collaboration. Word + track changes can get the job done for one-person-at-a-time as might be seen more often when making revisions. (We haven't compared word+revision control). But the typical Word workflow involves emailing documents to collaborators - barf.
2) LaTeX+bibtex is better for references. As far as I've been able to tell, EndNote is a steaming pile of pain that crashes frequently and takes Word for the Mac out with it. And glitches in horrible ways.
3) Word users are more likely to get their table formatting the way they want it in less time. I hate playing that game in LaTeX. Particularly if you want fancy nested stuff that looks right, is space-efficient, etc.
The study linked in this article notably omitted two of those that are quite important at least in engineering disciplines -- multi-author collaborations are the norm.
But the recommendations are asinine. In the engineering and sciences, at least, the amount of time that goes into the physical preparation of the paper, seated at your computer, writing de novo text, is nothing. It's trivial. You could crank out a full-length CS paper (12 pages, 2 column, 10 point text) in a few hours, were you just limited by your typing speed and the time taken to format it. If that were the case, I'd have more papers than Erdos!
This is another important point.There are incompatibilities between the Windows and Mac versions of Office. I've never seen any such cross-platform issues with LaTeX.
1. They compare only "errors", but don't have any measure of overall document quality. It's certainly the case that it's easier to make mistakes in LaTeX, but I find that people who use LaTeX tend to end up with more professional documents.
2. They compare only on very short documents using very little time. In much the same way anything I'm going to type in 20-30 seconds is better served by Notepad than Word (mostly because Word isn't always even usable in that time window), they're failing to compare LaTeX to Word in the situations where I'd argue that the distinction matters - namely, large, complex document writing. Things with >40 hours of work involved, >100 numbered items, >50 internal references, etc.
First, TeX is archival. I can still tex documents from my work in the 80s and get the same results. My MSWord files won't load. Thus, MSWord documents die.
Second, TeX allows Literate Programs. I can write a trivial program to extract my source code from TeX documents that explain the code. MSWord documents are opaque. Thus, MSWord limits what you can do with the sources.
Third, TeX is a language. I can (and do) write style files containing macros for my markup styles. In MSWord, what you see is all you are ever going to get. Thus, MSWord is fine if you can't program.
None of these features were mentioned. All of them are important, at least to me. I care that my work survives (archival), that my code can be explained (literate programming), and that I can optimize my workflow (style files).
If these additional constraints were added to the study then I think you'd find that some tasks are impossible in MSWord.
That said, this is a "bikeshed" paper. Emacs vs VI. Noise. Worse, it suffers from "hasty generalization". Sigh.
The biggest challenge is I work in a field where Word dominates and collaborators don't do LaTeX and journals won't accept it. So, there is very little payoff for slogging over the LaTeX hill, unfortunately. The LaTeX to Word translators all blow and Track Changes rules the world for collaborative writing.
There is no workflow that is not a pain in the ass to collaboratively write LaTeX with someone that can't use LaTeX or that sucks at LaTeX.
I encountered this when working with a group of both applied math people and clinicians - the applied math people all wanted to use LaTeX, but the gains just weren't worth the learning curve for the clinical co-authors.
If you're writing a paper with other software developers who already use git, this is very true. If you are collaborating with people who have never used a version control system, it is very much the opposite.
For the study they used a predefined text, so there's no real way to measure if there's an inherent quality on documents created using LaTeX, or if LaTeX does something to your brain to conceptualize your ideas while you are punching them on the keyboard. Or not.
> Given these numbers it remains an open question to determine the amount of
> taxpayer money that is spent worldwide for researchers to use LaTeX over a
> more efficient document preparation system, which would free up their time to
> advance their respective field. [...] 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 or PDF format.
LaTeX is a good tool, but people who use it tend to abuse it. If you need LaTeX for your solution, there's really nothing better for the job, but you rarely need LaTeX for your solution.
I wonder if they extended the study to include Markdown would that tilt the balance somewhat. The overhead associated with LaTeX might come from the fact that there's a compile cycle to get the formatting right, although they mention Kile as an editor so that's not clear.
Markdown also has fewer features---therefore a smaller learning curve than LaTeX)---and as a frequent user of Word (and lowriter), I know I interact directly with fewer features of the wysiwyg editors, given that many features are implicit or default, whereas I must manually invoke all features that I use in a LaTeX document (those I haven't placed in a .sty file anyway).
I can write LaTeX in Vim much faster than I can edit text anywhere else. Plus I have automatic refresh on saving and rudimentary spell check in Vim. Why would I use Word?
EDIT: Plus I have autocomplete set up that makes life even easier.
https://pay.reddit.com/r/LaTeX/comments/2qz6ju/postpublicati...
One thing I never understood is why it has its own file format and doesn't operate on plain latex files. That makes interoperability with latex and in particular colaboration with non-latex users harder.
In other words, it can't operate on plain LaTeX files because someone not using LyX might edit the files in such a way that LyX can no longer handle them. By using their own format, while they don't prevent someone from editing the file without LyX, they can at least guarantee that a valid LyX file will always be loadable by LyX.
I am not even sure time writing is worth optimizing in the first place.
BUT the evidence in increased errors and time lost might mean that the cognitive load of writing latex is higher, and that might be really relevant
> The study was conducted when most of the participants in the Word groups used Word Version 12. Only 3 participants used older versions of Microsoft Word. Newer Versions of Word include a new Equation editor that makes the software even more powerful and easier to use.
The authors of the article are, obviously, completely unbiased.
I have, for example, never really felt the need, and learning LaTeX has always been binned for "It'll be slightly more of a hassle in Word, but it'll still be done today."
But the main problem of the study is not its reproducibility. The main problem is that it is designed in a completely wrong way and it cannot seriously claim to answer the questions it pretends that it answers.
guess what though
I prefer the sports car