Hacker News Reader

Comment by CJefferson | Hacker News Reader

While LaTeX makes a good language for writing maths, I always feel I want to add it is an awful system for accessibility. The PDFs / outputs created by other systems like Microsoft Word are much more accessible.

For blind users, LaTeX Tables come out as an unreadable mess. The large-scale use of Latex to generate PDFs makes academia extremely unfriendly to anyone who needs accessible documents.

View on HN

Editing LaTeX documents is actually far more accessible, since they're all plain text. Every formatting instruction in a LaTex document is just text, while Word and PDF documents have non-text markup like bold, underline, italic, etc, which is WYSIWYG (with the emphasis being on "see" -- you must be able to see to be aware of the markup).

There are aspects of it which are extremely annoying to use, however - for example including and resizing an image could be a 10-20 minute googling session. Setting margins, headers, and footers not only requires including and invoking some package in exactly the right way, but actually choosing which package to use for the job as well. These are just a few examples, the list goes on and on.

LaTeX produces great looking documents, but I find it a giant pain to use.

When formatting a large document, those costs are amortized. In addition, many changes are set-it-and-forget-it. Once you set the right dimensions for an image, table, header text size, or margins, table of contents layout, bibliography style, etc., you wont have to mess with them again.

In a word processor, an image size can be changed by inadvertently clicking on something, the style of a text selection can be changed by moving the mouse scroll wheel on the font drop-down menu, page numbering may not refresh when working on a large document, inserting a figure may screw up table or image numbering, etc. You have to be constantly checking your multi page document to make sure everything is still ok.

As a teacher, my heuristic is simple: if I'm authoring a document that is less than 3 pages, I use Word. Otherwise, I use LaTeX.

For example, I write exams and syllabi in LaTeX. Quizzes, assignments and handouts in Word. It has worked well for me for a decade.

In MS Word, an inadvertent click won't change an image size. You have to click and drag. And inserting a figure certainly doesn't screw up table numbering.

> You have to click and drag.

Or type a letter, or open the document on a different computer, or set margins, paragraph size, printer, whatever.

But yes, just clicking an image won't change it.

Just because this is the way that most word processors currently work, does not mean it needs to be this way. You could imagine a word processor where the default mode was to edit content only, and editing structure/layout was hidden behind a second button or something.

LyX is, for me, the perfect word processor. It doesn't have those flaws. The only reason I use Word instead is because it makes collaboration and sharing with colleagues extremely simple.

I'm actually curious what you're doing that causes this. Basically all my figures are \begin{figure}[ht] and everything works out fine. Yeah sizing can sometimes be odd but I save it for the end (I use \includegraphics[width=0.x\textwidth] and adjust x. I get it in 2 or 3 attempts). I only invoke graphicx. I use \FloatBarrier frequently but that's trivial. I use jpg, jpeg, png, and pdf images almost exclusively (wish gif worked and wished mp4 would save into the pdf). I do also invoke tikz and pgfplots when I'm making vectorized plots and graphs (I know this is "advanced"), but I'm not writing tikz pictures in most documents. I do all my homework, papers, presentations, etc (PhD student) in LaTex. Been using it for almost a decade now and I've only faced placement problems when I was just starting and didn't know about [ht] (or [h!]).

I'm not sure what you find difficult about margins. You define them in document class. It is literally top, bottom, left, right, or height, width. Or you use the template that a publication is giving you. Are you doing something more complicated? If so, I can't imagine this is common.

Headers and footers? I never use headers. But I use footnotes (\footnote{this is all you need}).

Honestly I'm curious what issues you face. I'd gladly lend a hand (LaTeX is far from trivial) and I'd like to learn from others. But I'm honestly confused at the complaints. There's plenty of things to complain about LaTeX, but these aren't the challenges I face.

Yes, this is the same way that I use LaTeX, and have been using it for almost as long as you. However, when you first started out, how long did it take you to figure out this stuff? My original comment was in response to the “accessibility” of LaTeX, and honestly I don’t think it’s that accessible.

As for margins, I’ve had issues where changing the margins makes the page numbering “fall off” the page, and as for headers and footers, asserting that you “don’t need” them is not a valid option for many people producing documents.

Also to be clear, I’m not saying LaTeX is impossible or even hard to use, I’m complaining that certain tasks in it should be so much easier than they are.

And also tables are a major pain in Latex. Also that it does not separate physical and logical formatting of text. Asciidoc is what I prefer now, as it is semantically equivalent to XML DocBook, which I find is a better designed system than Latex.

DocBook is not a typesetting system, but I agree it's almost ideal to compose a highly structured document.

Word and PDF also support semantics readable by assistive technologies. With a LaTeX file, the user has to go through the raw code, often character by character.

Don't most people consume the document by reading say the pdf that is ultimately output at the end?

Yes, but I was responding to a post that wasn't about "most people." For people who use a screen reader because they have a visual impairment, a PDF is better than a raw LaTeX file if the PDF has been produced in a way that includes document semantics. My understanding is PDFs produced by LaTeX converters do not include such semantics, the code is only meant for printers, so also having the raw LaTeX file available is better than not having it.

>Editing LaTeX documents is actually far more accessible, since they're all plain text.

The conclusion doesn't follow from the premise.

Who said "editing plain text" is more accessible (in the accessibility for people with impairments sense) than a dedicated editor that can help them with assistive technologies?

>with the emphasis being on "see" -- you must be able to see to be aware of the markup

Not anymore than you must be able to see to edit a text document. The markup can be read back to the user (and is , e.g. "bold emphasized words end bold".

One trick I found that makes things WAY easier is by breaking sections of a paper into different tex files. You can share them on github and each collaborator works on their part. You don't end up with collisions and it works beautifully. I honestly believe LaTeX is easier to collaborate with than Word or even Google Docs.

Disclaimer: I'm mostly a nix user and so doing anything Word generally ends up as a pain for me, but I understand it isn't for most others.

Agreed. Having all the usual versioning tools at your fingertips handily beats Google Docs and Office 365 for collaborative paper writing.

One thing that I find useful is to have semantic line breaks in the source. That is, instead of using automatic or static word wrap, introduce line breaks at the end of sentences or clauses. This makes it somewhat more convenient to look at diffs.

I break on sentence end in HTML as well

Last time i edited some latex document, i ended up in dependancy hell with diffrent packages excluding one another or distorting the results of elements used by other packages. Has that become better?

For example if you use

  \usepackage[spanish]{babel}

to change the hyphenation tables, it also change a lot of things. For example it makes < and > active characters, so they can be used as a special kind of quotes. This breaks other packages like xypic that expect that < and > are normal characters to recognize the arrows.

Yes, now most people use online services that run on a standard platform with the dependencies already sorted out.

I've been using LaTeX for almost a decade and never faced this problem. Always done ```sudo <insert package manager and options> texlive-full``` and been good. Only dependency issues I've faced are forgetting to copy a cls file (which are downloaded).

I have run into that problem, but only when merging Latex documents with different packages. If you find something that's really difficult and requires a lot of packages, ask why other users haven't made it easy already - maybe it's not a good design.

Here are some thoughts on this topic, together with software designed to help make LaTeX more accessible, by a blind mathematician.

http://latex-access.sourceforge.net/

The software is designed to work with the LaTeX source, rather than with the ensuing PDFs. From what I can tell, it's not very complete -- but it certainly suggests to me that LaTeX provides a platform which can be built upon.

Given the academic world we are in, there are people making heroic efforts to make latex more accessible, and I don't want to discount that work.

However, none of these systems work well, and I think it'd an unsolvable problem due to the fundamental foundation of TeX, which is actually a text rewriting system, so when using packages there never is a nice semantic description of the page. The only thing to do is to rewrite every package in another language, which describes semantically how to latex should be interpreted.

Since you seem knowledgeable about the topic, could I ask --

What would an ideal solution look like? Say I have a complicated equation. How could I present it (in any format) so that a blind person could read it, and understand it as easily as possible?

Currently, the best practical option is do it in latex, but ensure the raw latex file is available as well as a pdf.

My question is more hypothetical than practical: what would an ideal solution look like? I literally have no idea.

As a developer I know this is bad, but I’m not familiar with accessibility tools. Surely this is a result of the compilation of the TeX source, right? In theory it seems like a text-based format would be more amenable to processing by screen readers and the like.

What can authors do to make their documents more accessible? Distribute the source? Stick to some formatting guidelines?

The best thing authors can do is distribute their latex.

In fact, concretely, upload your paper to arxiv which will demand the latex source of your paper.

In my academic mathematician life the path was clear: write a paper, place on the arXiv then submit. However I have sets of notes I'd like to put out there (some day), but are in much too fluid of a condition to commit to the arXiv.

I haven’t seen it lately, but there was a time where it was common for latex systems to not use space characters and instead add extra spacing after the last character in every word, because apparently that looked better ...

I hate the banal lack of imagination that goes into producing documents that look the same as dead tree documents have looked for the last 100 years, even though few people actually print them out any more.

This isn’t the 1980’s. We have better technology.

Given source code there’s no reason the document couldn’t be made available in a variety of formats, including more accessible formats than PDF.

But we stick to avatistic document representations because that’s what everyone has always done.

> documents that look the same as dead tree documents have looked for the last 100 years

In many cases we are talking about documents written using brand-new notation never used in print, but only on chalkboards or scribbled on scratch paper (by the author).

It just happens that LaTeX is a more expressive and precise tool for expressing arbitrary semi-structured 2-dimmensional symbolic formulas than available alternatives.

What better technology do you have in mind? HTML? Because as much as I like it, I find myself converting HTML to PDF every now and then so I can annotate, paginate, and yes, print it. PDF isn't perfect, but the art of typesetting is informed by much more than the physical properties of paper. Reducing communication to a static artifact divided into page-sized chunks is a useful feature for scholarly writing. Being able to produce physical copies of a document that you can give away for free, even if most people will read the document on an electronic device requiring a power source, is a feature. Writing static content that really takes advantage of the web's unique features [1] is hard!

[1] https://meltingasphalt.com/interactive/going-critical/

I’d prefer a format that automatically adjusted things like page and font size to whatever I’m reading on, whether that’s reactive HTML or an ebook format.

And some e-readers do support two column layouts, although it only makes sense if you have a wide screen.

I hate having to pan and scan a PDF file on my phone, and the fixed pagination only makes sense when printed. Even on most desktop screens a letter/a4 page is too long.

I'd like those features too, but unfortunately for reference material I find annotation and pagination more important. Annotation is broken in HTML/EPUB because there's no way to make my annotations portable across devices without vendor lock-in. Pagination is broken because HTML documents rarely contain paragraph numbers or other useful pinpoints, and browsers aren't as good as PDF readers at keeping track of your position in a long document. These are solvable problems, but until they are solved I find HTML unusable for anything that takes more than 15 minutes to read. I suspect that most people who consider PDF obsolete rarely spend that much time on a single document.

Agreed. For anything complicated, I’ll always download the pdf. I don’t really find it burdensome pan and zoom a pdf, even on mobile. It seems to work a lot better than publishers attempts at producing web versions of papers, which I find generally useless.

> reactive HTML or an ebook format

Genuine question but can you think of a website that uses reactive HTML properly and mind sharing it? I find that most websites seem to just put a maximum width on a page and call it a day.

Everybody should be writing and sharing org-mode files.

Generally I find myself wishing more webpages acted like a static document than the other way around. It's always guaranteed I can scroll and zoom in a PDF, which guarantees it's readable even if they layout is not the best.

I utterly despise the layout of academic papers. They’re garbage. Two column layouts have no place in the digital age.

Academic papers should have rich media. High resolution images and video.

Bah hum bug!

Strange, I find the compressed width of two columns perfect for mobile phones.

Really?

I find myself scrolling down then up, down then up. I also have to manually control the zoom so text fills my screen width. Then I had to scroll left and right. And when I want to scroll vertically I have to be careful to not screw up my horizontal position. It’s super frustrating.

I don’t think a single blog in the world uses a two column view. If it were an actually useful format for sharing information digitally then it would be more prevalent.

> I don't think a single blog in the world…

Well, yeah, you don't typically print blogs. Remember that academic publications are still printed, and columnar presentation can be very useful for annotation.

Double tap the text column and it auto-fits. When you reach the end give it a big ol’ swipe from the opposite corner and it will pull the other column into view.

At least in iOS...

I also find it easier to read on physical paper. Honestly I read faster when things are two column (speed readers will tell you why)

> because that's what everyone has always done

And we still print them out. People still prefer to read on paper. This doesn't change no matter how much the digital revolution wages on.

Couldn't that be fixed though? Would be a shame to have to give up on the Tex language because pdftex is out of date.

I suppose new development would happen as part of luatex. It would be really nice if latex could compile to more accessible documents!

Yet another reason to upload your (La)TeX source as well rather than only the pdf output.

We need a program to extract an accessible representation from PDFs produced by LaTeX.

Check out Mathpix snipping tool https://mathpix.com (disclaimer: I’m a founder)

I work in disability services and I can add a (perhaps slightly less biased) take that MathPix really is the real deal!

How does MathPix compare to EquatIO?

EquatIO is a MathpixOCR customer.

Cool! But how do you make money off of it? I understand MathPix is a company.

Snip is currently free. We charge companies for MathpixOCR API. We will probably charge heavy users of Snip in the future, so we can keep Snip free for light and medium usage.

On the Mac, the Equation Service or LaTeXit (?) included the source in the PDF so that it remained editable. Maybe the source would be easier to parse, especially if no macros were used.

How hard would it be to do some kind of OCR (maybe combined with machine learning) to “scan” anything that looks like a table into a more accessible format? It seems like there’s a lot of room for ML to improve various aspects of screen readers and other assistive technology.

I claim such a solution would be morally wrong (I am not going to comment on whether such a solution would be good in practice or not) because latex largely throws away the data it has on the table structure and this solution tries to piece it together again. I claim a morally better solution is to have latex not throw away the information it had, but I concede it’s often harder to modify something big and old and complex than to produce something new which is supposed to always work.

Reconstructing content and formatting from a PDF through OCR, machine learning, etc. may not be the best possible solution. But it would still make more information accessible to more people. So I must vehemently disagree that it would be morally wrong.

Can you elaborate on the moral principles you're employing here?

I assumed your parent was using 'morally' as it is sometimes used in mathematics: https://math.stackexchange.com/questions/1434043/sources-of-...

Morally wrong? What the actual f. The proposed is a perfectly reasonable and decoupled solution that would solve the problem for troves of existing, unstructured paper documents, too. In fact, it likely already exists.

Tabula does this. There are Python and R wrappers.

-Java: https://tabula.technology

-Python: https://github.com/chezou/tabula-py

-R: https://github.com/ropensci/tabulizer

Abbyy Finereader does an excellent job

Quite the contrary: A blind friend of mine studied math. He got scripts and exercises as latex source code from his professors and read them on his braille display.

Yes, getting the latex source is fine. However your friend would have been in serious trouble if they then went on to do a PhD, and then couldn't get the latex source to every paper they wanted to read.

Reply on news.ycombinator.com