Hi everyone, this is a room set up to discuss the following recent PLOS ONE paper that attempted to assess the productivity of LaTeX vs. Word.

It's clear from the paper's rhetoric that it had an agenda to show that LaTeX was an inefficient way of producing documents, but at the same time, it did provide a reasonably designed experiment which produced interesting results that are worth discussing.

@KennyPeanuts, @KeksDose I'd welcome your input here

By "reasonably designed" I mean the tasks themselves were clearly defined, there were three different tasks, and the dependent measures were also to a large extent not terrible approximations of the usability of the two programs.

@AlanMunn I'm not sure about 'approximations of the usability': most of writing a paper in whatever is just typing

@JosephWright Sure, but one can argue (and in fact it's what annoys me about Beamer, for example) that typing markup slows you down a lot (even if it has other benefits).

@AlanMunn What they didn't include, of course, is either a typewriter, hand-written notes to be typed up or something like Notepad/Markdown, all of which I think would beat LaTeX and Word for raw speed

@AlanMunn There's certainly a balance

@AlanMunn I'd be very wary of any XML-like solution to typing stuff (for example, in a LaTeX3 'native' context)

@AlanMunn beamer wasn't designed by the core LaTeX team, and one can argue that Till made some errors in the design

@JosephWright I agree, but (i) that seems out of the realm of current practice for anyone and (ii) the task was to reproduce the source document, which neither of those methods would be capable of.

@AlanMunn My point is that I don't think for an article to be published by professional publishers researchers should use any method to reproduce the layout

@AlanMunn Something like a thesis or other report is different

@JosephWright Yes, on this we agree (and remember you and I are I think the most pragmatic of the whole bunch of us when it comes to these matters.) :)

@AlanMunn You've obviously not worked for anyone more senior, where you (as the junior person) do the typing: that's very much like the old 'write by hand then hand to the secretary' approach!

@AlanMunn I maintain a class for a publisher: I get a reasonable number of e-mails of the 'How do I match X?' form, to which I say 'Don't unless the editors require it'!

@JosephWright Of course, given that, one might argue that LaTeX makes conversion to XML harder than Word, which means that for the purposes of producing publishable documents in today's multi-output world.

@AlanMunn I wouldn't like to say without seeing the back ends in use

@AlanMunn I guess that if one sticks to the instructions from the publishers, both paths are sorted at the 'back end'

@AlanMunn Of course, people don't stick to the instructions :-)

@AlanMunn I think it helps I use Word very regularly: it's more-or-less required in the chemistry field

@JosephWright Is word usable if you know what you are doing? Like, intermediate to expert level?

@FaheemMitha For a given value of 'usable'

@FaheemMitha Almost every paper I've worked on was done in Word, none in LaTeX, and they all got published :-)

@JosephWright sure. I was talking about the pain level.

@FaheemMitha My thesis is in Word (predated my involvement in the TeX world), and if you work at it you can get on OK

@JosephWright Mine too. The downside is that I can no longer open the file...

@FaheemMitha One issue with long documents is most users of Word don't ever pick up the things Word does offer, while for LaTeX it's hard to miss \section, etc.

@AlanMunn You probably don't have lots of graphics making life tricky

@JosephWright Does Word have a a macro equivalent?

@FaheemMitha Long docs get tricky, although that has improved, references using EndNote have got less risky than they once were

@FaheemMitha Two-column floats are a pain!

@JosephWright No. But I did have fonts.

@FaheemMitha VBA?

@AlanMunn Ah, yes: not an issue for me

@FaheemMitha AutoCorrect can be employed quite cleverly

@AlanMunn I made a PDF on submission (required nowadays, of course)

@JosephWright Yes, I still have a PDF of mine too.

@FaheemMitha I had one very tricky float that depending on the text would have to be split over two pages to avoid total disaster. That one was hard work.

@JosephWright I see. Doesn't sound too terrible. Of course, one has to pay for it. And use Windows.

I think we need to dissect the paper a bit... It seems that the authors suffered a severe disagreement that is pretty visible at this point.

Any researcher who has ever collaborated on such large
interdisciplinary projects has experienced the difficulty with reaching a consensus
about which document preparation system to use. Discussions about document
preparation systems are often unproductive and driven by preconceived opinions,
individual biases, and disciplinary traditions. A fair comparison of the efficiency
and usability of the different document preparation systems based on empirical
evidence rather than individual habits and biases may facilitate such discussions.

@FaheemMitha We had a student recently writing up in Word who did much more than I could to convince his girlfriend to write up using LaTeX. Her thesis was the best produced I've seen in chemistry, and she avoided almost all of the issues our 'typical' students have. But I can't say she was more productive that she might have been otherwise.

@FaheemMitha Every university has a license for all of this stuff, and we'd need to pay for it for admin, etc. quite apart from writing articles, so ...

@JosephWright Sure, I was speaking generally.

@percusse No: no discussion for me ;-)

@FaheemMitha So was I

Participants
were classified as ‘‘novices’’ if they had less than 500 hours of experience with the
respective program and ‘‘experts’’ if they had more than 1000 hours of experience
with the respective program. In the resulting groups, participants who were
classified as ‘‘novices’’ had on average 234 hours (SD5153) experience with the
respective program, whereas ‘‘experts’’ had on average 1909 hours experience
with the respective program (SD5211).

@FaheemMitha Article is about academic work => universities

This part is also problematic as workhours is correlated with expertness

@JosephWright Well, not everyone has MS licenses. Even universities. And there might be people working indepedently.

@percusse That's a very poor defn. I was meaning to look for that.

@FaheemMitha Not easy to work independently in chemistry :-)

@FaheemMitha I guess you can use OpenOffice or whatever and come to much the same conclusions

@JosephWright Also the task is to immitate what you see, not what you would like to have. So the task is biased towards visual approximation

I've used LaTeX since the 1990s. I'd still count myself a s beginner. That might be in the range of 10000+ hours.

Hard to estimate, obviously. I use LaTeX for everything, not just academic stuff.

So the hypothesis seems as follows;

@percusse Yes, and they didn't compare e.g. Quark

How convenient is it to replicate these aforementioned items via Word and LaTeX?

@percusse Yes, seems reasonable

So the input is a German text, table and some equations.

@FaheemMitha I've used Word (occassionally) from late 90'. I am still a beginner, because every new version is completely different. :-(

@PrzemysławScherwentke :-)

@PrzemysławScherwentke No, much the same if you know what to look for

So rigourwise, they don't reveal the approximation metric, namely; when is one particular table is better approximated the table than the other, given that both of them are not perfect?

Actually, I've learnt more about TeX/LaTeX in the last few years than the previous 15, larges because of tex.sx.

@percusse Yes, I think most of their error metrics are quite hard to figure out.

The denizens of tex.sx should get together and publish a rebuttal paper.

@percusse This is certainly one of the ways in which the task is biased towards a more visually oriented program like Word.

@AlanMunn Plus they are dictating a particular design, as their Fig. 1 reveals; columns, headers, spacing(!)

@FaheemMitha I've suggested that, and would be willing to run the experiment(s) if we could come up with reasonable ones.

These are all considered as bad practice in TeX part so it cannot be a neutral test.

Plus it is generally discouraged via package design.

I wonder how much of their time was taken up with actually writing the text as opposed to designing and running experiment, thinking about results, talking about them, doing teaching, committee work, reviews, ...

@JosephWright Judging by their Excel sheet, not much

@JosephWright Of course that goes to Clea's point in the comments to the question.

@percusse People still do that, though, so it's not entirely invalid

@JosephWright True but then expertness argument goes down the drain

it's like typing equations in Word, you can still do that if you have to but you won't

you can't excel at it.

excuse the pun

I think one of the crucial problems of the study design is exhibited in the wonderful xkcd comic.

For example, you can simply divide your document into pieces. (I'm deliberately not going into pgfplots, pstricks, tikz stuff as they might be considered as techy )

In Word, this simple task is a pain in the neck. As researchers they simply ignore this. Well you can simply copy/paste? No good sir you can't

@percusse Er, yes you can

@JosephWright Shall I send you some templates for try out? :)

It copies the template text too :)

@percusse I'm used to sorting out that sort of thing

Get in the Word Expert line then :P

@percusse I probably would fit into that category (other than VBA, which is hard work)

@percusse I do of course know what you mean

And on top of everything, how can a continuous text can make a difference on which editor you are using?

That's simply a reason good enough to reject this article.

LaTeX uses a different keyboarD?

Is it the screen color?

There is no correlation study.

Did they make Word users type on the other editor?

@AlanMunn It's hard to see how one could come up with a truly fair test here: to what extend writing up results limits research rate, etc. I can easily get held up for half a day by a small technical issue in the lab, for example: that's probably a much bigger hit than anything else.

The more you read the more it doesn't make sense

One third of this article can be summarized as MS Word has a spell-checker and autocorrects

@percusse Buy Mavis Beacon?

A proper test should have tested with auto correct on and off. To verify it is MS Word's contribution.

@JosephWright Right, I do agree with that assessment, but that's kind of an "even if it's true, it's irrelevant" argument, unfortunately. So if we (one?) were serious about responding to the article, we would either design an alternative test that would go the other way (showing that both side can be biased) or design a neutral(ish) test and see what happens...

Well, after reading it a bit better, I think I can safely discard this article. OpenAccess keeps worrying me sad to see.

@percusse I have autocorrect built into my Mac, so I don't see the difference.

@AlanMunn You have to turn the machine on and off see it matters then :P

Maybe the productivity goes up? Who can tell ?

If you think about it, it's kind of hard to justify that LaTeX is efficient. People use it because they like it better than the other alternatives. For whatever reason.

@percusse Yes, perhaps it can be discounted scientifically, but sociologically it might warrant a response.

@AlanMunn Yes, true

@percusse Certainly if my machine were off I wouldn't be doing this productive work... :)

People don't necessarily use software that they think will make them the most efficient. Except, really practical business types, that count the value of each of their minutes in money.

@AlanMunn 'Start with a new file, type up a (supplied) "hand-written" draft using whatever format you want'?

@AlanMunn @JosephWright Did you see the comments to the article on the article page?

@percusse there were comments?

which is suddenly a big plus for OpenAccess :)

check them on the right side bar

@FaheemMitha Quite true: Word for example is just the standard in a lot of areas

@JosephWright An unfortunate standard.

@FaheemMitha Not necessarily though. I think it's a good piece of software. It is its grandiose claims that makes it loathsome (I just learned this word from @Brent.L)

@percusse Yes, this one is especially helpful.

For example people like using free systems. But they spend lots of time messing with it. Maybe proprietary systems would be more efficient. For one thing, you can really mess with them much, so you might spend more time doing actual work. And if you want to do something custom, buy a program to do it.

@FaheemMitha I'm not sure: if you started from scratch writing a word processor, what would you end up with

@JosephWright I dislike proprietary software, and don't like the concept of a binary, very complicated, only partly documented standard.

@FaheemMitha Yes, one of the reasons I use a Mac and not Linux.

And I have used Word in the past, but not recently.

@JosephWright minesweeper

@FaheemMitha I can see that argument, but you could write a completely open word processor and I think you'd still have something looking a bit like Word (we used to have several competing word processors, and the ideas were picked up by all of them)

@AlanMunn Right. Having said that, free systems can become part os ones identity. You could equally reasonably say, don't have hobbies, you'll get more time to work.

I guess my overall point is that it isn't really about efficiency.

@JosephWright That would be an interesting way to measure raw typing skills. It occurred to me that perhaps LaTeX users are just more likely to be crappy typists...

@JosephWright Yes, possibly. I guess openoffice/libreoffice is somewhat similar. All word processors have some generic similarities.

@AlanMunn Back with the typing pool business. I should be OK: I did do Mavis Beacon while a student!

@JosephWright :) And I never have. I'm still a four finger + thumbs typist.

@FaheemMitha I was thinking back as far as WordStar via Word for DOS, AmiPro, etc.

@JosephWright I don't know any of those.

@FaheemMitha Really?

@JosephWright This is actually a big confound in their subject pool that as far as I can tell they didn't control for.

@FaheemMitha WordPerfect?

@JosephWright Nope. Can't say I regret it. I should also have added to the above that I don't (and never did) like the idea of a word processor.

Though TeX can be a pain some of the time, I'm still 100% on board with its basic concepts.

@FaheemMitha WordStar was the CP/M word processor, WordPerfect was best I'd say on DOS, AmiPro was in many ways superior to Word On Windows

Also, I really, really, like things to be entirely in plain text.

@JosephWright Ok

I mean, sometimes I wish TeX was more straightforward. For example TeX programming is a species of magic. It would be nice if this wasn't the case. It's particularly unwelcome when one is in a hurry.

@FaheemMitha Yes, but ideally end users shouldn't have to worry about that, I feel: that's why I'm on the LaTeX3 team

When I used word processors, I totally hated that all that information was hidden away inside the format.

@JosephWright Sure, but as an expert, you must agree there is a long way to go.

@FaheemMitha Yes, regrettably

@FaheemMitha For the topic in hand, of course, I'd argue that the user should leave well alone: for articles to be published, fiddling with formatting is a bad idea in Word, TeX or anything else

I didn't know LaTeX 3 has a focus on programming usability. I thought it was trying to make things more consistent. General cleanups, that sort of thing.'

@FaheemMitha Programming layer, design layer, ... Clear separation of different ideas

@JosephWright Would be good to get some funding. Have you thought of a kickstarter thing? Seems to be all the rage. Worked quite nicely for Joey Hess. Has worked for some other free software projects.

@FaheemMitha latex-project.org/papers/LaTeX3-architecture-2011-slides.pdf

@FaheemMitha Money not really the issue (at least, not on any realistic scale)

@AlanMunn Discovered the room owner tools?

@JosephWright :) Yes, I was just about to ask you and then I thought, hmm maybe I have the power....

@AlanMunn Indeed (I of course do as well)

@JosephWright Oh, really? I find that surprising.

Meaning it would take huge amounts of money to make a difference?

@FaheemMitha None of the team are employed in the typesetting business, so you'd have to find enough to pay at least one of us outright

@FaheemMitha And realistically you'd need to be able to pay FMi

@JosephWright FMi?

Oh, Frank?

@FaheemMitha Yes

@JosephWright And getting people from outside with the necessary expertise would presumably be next to impossible.

Assuming such people exist. How big is the LaTeX 3 team right now?

@FaheemMitha latex-project.org/latex3.html

@FaheemMitha Probably off-topic for the current room: one for the main chat

@JosephWright Ok

No sign of Keks or the OP yet

@JosephWright Nope.

@AlanMunn Shame

I think the PLOS paper is written in Microsoft word. I can see that by the bad text alignment of the paragraphs

I can spot that something is written in Microsoft word from a mile away.

@Nasser Typesetting of the article is down to the journal, of course

@Nasser PDF info says Arbortext

@JosephWright well, it was not written in Latex, that is for sure :)

@Nasser actually that's part of the PLOS ONE style

@Nasser Arbortext is quite capable of justification: this is probably a style decision

what's for sure is that the pdf looks quite ugly - I am currently formatting another PLOS ONE paper, and the latex output looks much prettier

Justified text always looks better, at least I think so.

(and I am doing it in Latex, of course)

@Nasser Tufte disagrees on the matter. I think each have their place. :)