Why marginalia (repost)
Without the experience of [historical multiplicity] the sense of it will be lost. Whatever we are, we became in history, and only in history can we remain the way we are and develop therefrom: it is the task of the philologists, whose province is the world of human history, to demonstrate this so that it penetrates our lives unforgettably.—Auerbach, E. (1969). Philology and Weltliteratur. The Centennial Review, 13 (1), 1-17.
There are many phenomena now occurring that are all too familiar to both philology and anthropology, proving that a knowledge of behaviors does not necessarily translate into more advanced behavior. This post worries about the claim on knowledge experience of collectivized metadata where this effaces the voice of the individual gloss. It therefore has two parts: the problem of multiplicity in postdigital experience and the problem of marginalizing multiplicity.
- Experience, multiplicity, and mathematicized digitization
- Resurfacing marginalia
Experience, multiplicity, and mathematicized digitization topLanguage adoption and mimetic behavior are two examples of current behavioral phenomena that stand out to me. They can be conflated in unidirectional predictionism achieved through complicating signal to noise. Unfortunately, only the more simplistic models of trained classification are being mirrored, but these have already been rewarded over the past half-century as they are the easiest to optimize and reward.
In the mean time, the knowledge that we have externalized over the centuries has been digitalized, with bots granted more access to than humans. This knowledge is now partly enclosed in real-time learning models that cannot be audited because they cannot be fully evaluated. They cannot be evaluated because, for example, models that use abstract layers of statistical weights on data sets can simulate phenomena or situations that were already difficult to objectivize. These are compared to real-life situations and optimized or retropropagated for improvement—again, they cannot be audited.
“Creative discourse”—of problem solving, exploration, decision-making—is now essentially the promptism of an already programmed interface within a closed system hopefully at least spiced up by knowledge soup. What new data are we feeding the self-supervised model when our two decisions are to swipe left or swipe right? I wonder what this means for the future of innovation.
Yet here, language is as dynamic as language freed from legislation and auditing can be. It is more dynamic than pre-(digitally-)processed human communication that is categorical, reductionist, destructive. It creates actionable realtime decision trees. French philosopher Bernard Stiegler thinks this can short-circuit our own decision-making. This is to say nothing of the time needed to do the slow work of developing individual heuristics, considered in the light of the full range of possibility, sifting through good and bad.
This dynamic electronic language is also being used to furnish decisions when yesterday we spoke of the furniture of the mind. “A made-for-television script that makes no sense if there is nothing in our hands”, as Toni Morisson said in her Nobel Prize acceptance speech.
It could be that mathematizing knowledge into data is a form of language death. I don’t know. But I do note that how we are using language today and how we speak to each other is more calculating and categorical, just like the models that group behaviour. Is our language, to paraphrase Morrison, evacuated, leaving no access to what is left of human instincts for it speaks “only to those who obey, or in order to force obedience”?
The gift of this incredible time during which we live, with the potential to have more knowledge at our fingertips than ever before, depends on whether we agree that even though we may be holding nothing, every day we are deciding what we think about the future of knowledge through how or even whether we listen to the experience of multiplicity. Taking a moment to observe different local patterns, collections of meta patterns, meso patterns, and so on, after Christopher Alexander, might reveal a difference among patterns and recognition of those which permit “new knowledge” and the “mutual exchange of ideas”.
Perhaps the achievement of Paradise was premature, a little hasty if no one could take the time to understand other languages, other views, other narratives period.
In the scholarship of the pre-digital new era, studied marginalia was a dynamic process that took place on top of a static medium. Where this grew in to philological apparatus, it included a full range of historical and contextual references, filling in the picture of a knowledge artefact that a reader could see was distant from themselves, yet through effort, somewhat intelligible.
Resurfacing marginalia topHow can we safeguard difference in an information flow of dynamic data that is already just one type of knowledge? Within the digital realm, how do we protect discovery—an idea on my mind because of an exchange with rpg? Or, in the words of the Marginalia search engine founder:
Small websites just can’t compete against the SEO-industry on Google. So they’re almost impossible to find, and struggle to find visitors. It’s a real tragedy how many brilliant and fascinating websites are languishing in obscurity. Instead we get listicles, content farms, click funnels.
The Marginalia search engine, at the margins of mainstream data distribution, tells a different type of story of what can be said digitally. It is one of the ways to discover an idea that can surprise us out of an old way of thinking and see the world anew.
But the very idea of marginalia—here I speak of the elaborative practice and not the engine—also raises a deeper question about what indexing means in an age of dynamic text loading. Marginalia are comments or glosses, critiques, doodles, and such written, drawn, or scribbled in the margins of books. Especially where these were made in the once-prohibitively expensive manuscripts, the marginalia would be included to annotate the text, to make it easier for later readers to understand the text. In this respect, it can be compared with the critical apparatus of yesteryear. Collections of marginalia make it possible to retrace the complex history of a text. Scribes would compare manuscripts in an attempt to determine authenticity. It can therefore be asked: how will this be done in the postdigital world in which sideloaded text is that much more ephemeral and industry best-practices are not necessarily indexing or showing marginal or interlinear voices?
whether it laughs out loud or is a cry without an alphabet, the choice word, the chosen silence, unmolested language surges toward knowledge, not its destruction. But who does not know of literature banned because it is interrogative; discredited because it is critical; erased because alternate? And how many are outraged by the thought of a self-ravaged tongue?
Word-work is sublime, she thinks, because it is generative; it makes meaning that secures our difference, our human difference – the way in which we are like no other life.
We die. That may be the meaning of life. But we do language. That may be the measure of our lives.—Toni Morrison
Aurélie Jean (2021). Les algorithmes font-ils la loi? Paris: Éd. de l’Observatoire.
Sowa, J. (2000). Knowledge Representation: Logical, philosophical, and computational foundations. Pacific Grove: Brooks Cole Publishing Co.