Typesetting and Document Formatting

I originally posted this on tumblr, but decided to copy it here as well.

Have you ever had to make a 'document'? Usually it boils down to something that gets turned into a PDF, something printable at least. Generally they're not made to be adaptable to screen sizes, but rather to some standard of aspect ratio, margins, letter types etc.

If you're not using a printing press for this (kudos if you are, please send pictures), you probably use a computer program for it. For most people, this will be Word from Microsoft's Office suite (do they still market it like that?). Some people might use Google Docs, and Apple probably has a comparable program. I personally call these programs 'typesetters'.

They suck to use.

They're extremely opaque, only presenting some basic elements to the user (headers, lists, tables, etc), with no upfront indication of where those elements should go, and what their sizes and margins are exactly. The rules for when something wraps, or moves down a line, or gets moved aside, etc, are also very unclear, and because it all seems to be calculated from a single origin, one small font size change can entirely mess up your project three pages down. This all is not even mentioning images (although tables got a brief mention), or any formatting beyond 'wall of text with particular alignment'. When such 'advanced' formatting is possible at all, it is three times more finicky and opaque.

People might say "Well you have to really learn a tool to get good at it!", but the typesetters in question don't present themselves as one tool with a particular paradigm. They present a couple features that the typesetter's internals can do something funny with when they're used (and it usually does not share your sense of humour). Learning a tool like that is extremely unnatural for people like me, because there is no solid ground to stand on in terms of operational paradigm.

There is an existing alternative to this, and that is the markup language. Popular examples are HTML and Markdown, but particularly relevant here are the ones popular for use in academia, like LaTeX/TeX (mostly only LaTeX, but Typst deserves an honourable mention). These are great, as they give very explicit and direct control to the user, and almost every aspect of the formatting and layouting. On top of that, the languages have a (usually small) learnable core syntax. In my opinion, such markup languages are superior to the common 'What You See Is What You Get' (WYSIWYG) typesetters.

However, they're still not really great to use for layouting. They become esoteric quickly when you start looking into how things are managed, with legacies going frighteningly far back. Images still move your text, or the other way around, or both.

Markup Languages are not the ultimate answer.

While the Typesetters are built on shaky ground, the Markup Languages live too much in the dirt. Raw text is largely one-dimensional, and allows for a limited second dimension for slight nuance. A markup language has to contain:

Most of these are not in their 'native' language, and thus a markup language will always have to sacrifice either the clarity and ease of the content text, or that of the metadata text.

(As an interesting aside, raw text languages make more sense for interactive UI, because as far as we know, ordered instructions are clearest in a linear textual format.)

So what now?

I've thought about it a lot, and it has puzzled me for a while that doing such layouting has become so frustrating and convoluted. I think the main problem (as given earlier) is how opaque it has become, and not so much that it's an inherently difficult problem. It's very straightforward to do on a typewriter, for example.

I think these typesetters should look more like either image editors, or likely better, parametric CAD design tools (at least optionally).

Another fun idea I had is a 'rectcut' typesetter.

In it, the limits and positions of different types of content is entirely given by the arbitrary subdivision of rectangles, with layering like an image editor. So, you start with your paper size, and then you (for example) give margins by subdividing in three horizontally, then vertically, and setting those boxes to required size. This can be done fractionally or in absolute units (both should be possible). You could then further subdivide to make a little box just to the side, which you can select as 'image content', and a box below with some info. Then the other space you can select for 'text content' or maybe you want a title, so you make a box, and put in your title, etc.

Line or page numbering, tables, headers, indentations, footers, graphics, it all becomes something that slots in way more straightforwardly.

In my mind, this is rectilinear/axis-aligned subdivision, but one could argue there is value in expanding beyond that (I have certainly seen posters and books that could use it)

A small disclaimer:

I haven't looked at the history of typesetters or layouting/graphic design tools yet, and I don't know about all the options that are out there, so do let me know if you know of something similar as some of the things I described (I believe Adobe (I cry) might have some, but I wouldn't want to use those regardless).

All of my problems outlined are also at least a little bit a 'me-problem'. In general, it is difficult for me to adhere to someone else's workflow, and such programs definitely require a certain workflow that is not my own.

Thanks for reading!