Home

Daniel I. Scully

pixels

The Pixel Perfect Fantasy

When writing my PhD thesis recently, I departed from the standard practice in science of writing documents in LaTeX. The resulting printed version does not look as perfect as it might have done had I used LaTeX, leading many of my colleagues to conclude that I had used an inferior method. But less perfect is not necessarily bad.

LaTeX came from the old world of science documents where everything was printed. When things are printed there’s a definitive version, and it’s probably A4 in size. Yes the document was authored on a computer, but that was only to more effectively produce a printed document. That document would then be printed as a thesis or included in a journal (which was printed and posted), and people would access it by visiting a library and picking it up. Everyone, all over the world, got an identical copy of the definitive, printed version.

But that was yesterday. Today everything is distributed electronically, because it’s faster and cheaper. Today, we aspire to a paperless office, where we make, store and read everything on a screen. Today, we want to read articles on our desktop, laptop, tablet or phone. In this world, an A4 paged document makes very little sense (even less so when there are columns on that page!).

LaTeX simultaneously describes both the content and layout of the document. So it’s very good at creating well crafted and pixel perfect layouts on a set page size. But it leaves the reader constrained to that page size. The user can zoom the page but nothing else. This can lead to a particularly terrible user experience, for example for those with poor eye sight or small screens, to name just two.

My solution was essentially to use the language of websites (HTML/CSS). The modern website approach of “responsive design” allows such a document to change and adapt itself to the screen it’s being viewed upon – so that it always makes the most of what’s available to it. It will also respond to changes by the user who could, for example, increase the font size and the document would react so that it’s still readable.

The result is sometimes less than perfect. In LaTeX you could introduce a line-break to tidy up the layout of some text, push an image around to prevent a paragraph breaking over a page, or tweak it in any number of ways to get the page just right. In a responsive design approach, these tweaks make no sense because you can’t know where those line breaks occur, and there are now pages.

This is where my PhD thesis came in for criticism, because some of the layout on some of the printed pages wasn’t quite ideal, when LaTeX would have done better. But the mistake is in thinking the printed A4 version is the document. In fact it’s just one view of the document. My thesis looks pretty good on a phone, looks pretty good on a desktop computer monitor, and looks pretty good printed on A4. In contrast a LaTeX thesis would look perfect on printed A4, and absolutely awful everywhere else. The price of my thesis looking good everywhere, is that it doesn’t look perfect anywhere.

Since however you can no longer assume your user is reading from a piece of A4 paper, it would seem to me that the universal approach is the only way to go. A LaTeX A4 document may seem to offer you the ability to craft a pixel perfect representation, but since you can’t control when or where that document will be read, or how, or by whom, that offer is an illusion. While the pixels may be perfect, you’re reader’s experience won’t be.

This idea isn’t new. Web designers have been talking about “responsive design” for years now, and it’s long been understood that the designer cannot have complete control over how something appears to the reader. The question is, how long until this understanding filters through to the reluctant scientific world. While an alternative which meets all the requirements for scientific documents is not yet in place, that’s partly due to a lack of acceptance that the problem exists. Only when we recognise that our readers are suffering will the necessary momentum build to alleviate it.

Tags