About

This is the web version of the PhD thesis of Pierre Depaz, intending to facilitate access and readability of long-form written works beyond PDFs.

As such, it is also an inquiry into publishing research work online. It integrates modular systems of epistemic software (Zotero, Hypothes.is), hyperlinking of labelled components (headings, figures, listings, citations), document layout (user-accessible line height, line length, font family, font size, as well as responsive design and dark mode), asset delivery (optimizing requests for size and speed, manifesting as a progressive web app) and publishing pipeline (parsing, processing and templating).

It is also a practical attempt at translating the ideas of this thesis in a different computational medium. Some aesthetic properties can support a positive cognitive engagement. As a different material, emphasis is being put on the referencing graph of different components (attempting to build up a habitable semantic space by connecting things meaningfully), on inteface design (enabling epistemic and navigational actions, such as highlighting and toggling content visibility), but also on more traditional document layout principles (typography, spacing).

Others have also thought about and done things along the lines of literate programming, publication numérique, computational publishing, experimental publishing or even book crafting. A lot of good ideas in a lot of different directions, getting deep into the malleability of digital media.

If we were to take the conclusions of this PhD as a starting point, it turns out that some findings apply more, and some less, when translated to another environment. Some aesthetic properties of source code, such as metaphors and idiomaticity, apply more due to their evocative nature, and the high error-redundancy of the document. Metaphors happen mostly through the use of icons in order communicate the hybrid status of the document (partly text, partly software). Idiomaticity is related to the best practices of digital document design—in a way, speaking the language of the born-digital (still somewhat in becoming).

On the other side, precise token naming and structuring of document are less fertile grounds for stylistic innovations. These differences are, I think, also due to a more evasive assessment of the executed text and its function. Because the text is also mostly traditionally read-only at the moment of encounter syntactic highlighting doesn't make as much sense in natural languages than in programming languages; it might even detract from the linear flow of natural language reading. In terms of linguistic register and prose style, it might not be the most accessible, due to the context in which it has been written, and a need to adapt to such context through idiomatic terminology; but it would be fun to re-write in more vernacular terms! Ultimately, though, this system tries do the most with the least—that is, it tries to be elegant—and tries to conciliate optimization for a specific instance and accomodation with novel data structures.

One of the requirement to fulfill this degree is to hand in a specifically formatted PDF document, a format which, I assumed, people tend not to read (or not to have access to, for paywall reasons). I also assumed that, as information keeps growing in importance in our societies, the quality of the information delivered becomes more clearly related to the means of diffusion. Media saturation could also imply media differentiation, even at the slightest level. Such quality is, in its turns, relatively more dependent on its formal presentation.

What would be an appropriate medium for the presentation of this research? What does it look like to make a document more accessible as the result of media-translation? How much does the networked computer depart from print? And how can a static document be turned into a dynamic document? What are all implications of such a media translation?

Here, I take a rather practical and processual approach to those questions, which means that there isn't much documentation and that it's an on-going process with no rigid sequence. On the other side, there's something to look at, comment on and play with.

One of the conclusions of this thesis is that a form of beauty of source code is the extent to which it is shaped adequately to its material. This raises the question of media-specificity. In this case, it is the relation between the content of the document, its medium of representation, and its representation through this medium. What are best (non-invasive, fast, connected) ways in which a digital page is different than a printed page?

The original document is written in LaTeX. The parsing is done in Rust with the Pest parser, then a series of bash commands in a Makefile copy the JSON output and required assets to a Svelte application; the app is then built by SvelteKit and deployed to a virtual machine rented from Digital Ocean.

The serif font is Bespoke, the sans serif font is Inter the mono font is IBM Plex and the icons are from RemixIcon.

You can find the source code for this website (both LaTeX parser and HTML renderer) on this repository and you can find the source files of the thesis itself on this repository.