Chance, Desire, and Pierre Menard

“From now on, concepts such as “true” and “false” refer only to unattainable horizons”

In May 1939, Jorge Luis Borges published a short story called Pierre Menard, Author of The Quixote. Menard, we are told, was a novelist who, at one point, embarked on a mission to recreate Don Quixote word-for-word. “Initially,” Borges writes, “Menard’s method was to be relatively simple: Learn Spanish, return to Catholicism, fight against the Moor or Turk, forget the history of Europe from 1602-1918 – be Miguel de Cervantes.” But for Pierre Menard, this was too simple.

Becoming Cervantes seemed as if it would lead too naturally, too predictably to the production of the Quixote. Instead, Menard resolved to arrive at Quixote through his own experiences. At first glance, this seems impossible. How could someone sit down and, in a creative act, reproduce a 1000 page novel? In fact, it turns out that Menard was only successful in recreating a few scattered chapters. But Menard recognized the improbability, rather than impossibility, of this endeavor: “The task I have taken is not in essence difficult. . . if I could just be immortal, I could do it.”

This is a variation of the classic thought experiment known as the Infinite Monkey Theorem. Given enough time and enough attempts, random chance (in the form of monkeys and typewriters) is capable of producing anything that is not outright impossible. Naturally, this thought experiment leads us to a question about the nature of human creation. Fundamentally, what is the difference between the storytelling of a single Cervantes, as opposed to the probability-engine of the infinite Monkeys (or an infinite Menard)?

In Into the Universe of Technical Images, Vilém Flusser provides an answer: “I distinguish myself from chimpanzees and other ignorant beings in that I produce the same things they do but in a much shorter time.”

Of course, he continues: “But the matter can be presented differently. Whereas the typing chimpanzee is immersed in a blind play of chance and necessity, I transcend this play. As I type, I see past the game. . . to the text to be written.”

While this may seem obvious, it’s a key point to make. Intelligent creation requires us to look into the stream of possibilities and choose a favored outcome. We “see past” the surfaces of texts, images, and other objects in order to interpret them. We develop (and receive) a deep understanding of the relationships between the object at hand and the wider cultural and information environment in which it exists. Through this lens it is possible to understand what the object implies. With a purely random approach to creation, your output is a predictable function of your throughput, with identical chances of producing any given text, image, or object. With an understanding-based approach, you can tilt the scales towards extremely unlikely outputs. As production of information becomes fully automatic, the extant information inexorably begins to saturate the probability distribution of the possible.

Flusser deals with the notion of seeking out the desirable within the possible and automatic. Given access to an infinite stream of automated, probable information, how do we curate it, capture it, and tame it? Where do we direct and redirect it? What influence will these automatic processes have on the structure of society? Flusser’s approach to these questions centers on a still-emerging-but-rapidly-maturing form of media he dubs Technical Images.

Particles, Apparatuses, and Envisioners

“The rules that once sorted the universe into processes, concepts into judgments, are dissolving”

Flusser portrays the history of media, in very broad strokes, as a ladder with 5 rungs:

1. The world of unmediated and concrete experience,
2. The invention of tools which hold the world at bay, mediating the world through hands,
3. The invention of images which depict the relationships between objects, mediating the world through the eyes,
4. The invention of text which can order the world according concepts, mediating the world through rules,
5. The collapse of rules into “particles that must be gathered up”. These particles are the material from which technical images are produced.

This idea of the “collapse” of rules strangely foreshadows the recent success of implicit and learned computer systems in the fields of, to name a few, coding, translation, conversation, and of course, image creation. Modern neural networks have achieved remarkable success by shedding the explicit rules which people previously used to understand the world, and replacing them with implicit rules, learned rules, rules which are defined only by the flow of information through the network as it learns optimal behavior for a given task. While traditional images “arise through depiction,” technical images, Flusser writes, are the result of “a peculiar hallucinatory power that has lost its faith in rules.” His reference to hallucination feels especially prescient in the context of neural methods.

The differences between technical and traditional images are subtle but important. Technical images are “mosaics assembled from particles. They are therefore not prehistoric, two-dimensional structures but rather posthistorical, without dimension. We are not turning back to a two-dimensional prehistory but rather emerging into a posthistorical, dimensionless state.” Technical images, once solidified, cannot be separated from the amorphous, dimensionless flows of information from which they were snatched. Theyare that flow of information, seen at a particular moment from a particular angle.

From a human standpoint, a lack of dimensionality and the arbitrarily-high-dimensionality of neural networks might look pretty similar. Images assembled as a mosaic of high dimensional tokens are pretty familiar to users of the internet in 2026. Additionally, in modern image models, the internal hidden states, the layers which contain the model’s implicit understanding, contain thousands or tens of thousands of dimensions. The particles, or curves, or data points, which a modern image model uses as a raw material undoubtedly exist in a space that is beyond the human capacity to visualize, dimensionally speaking.

Writing in 1985, how was Flusser’s analysis so prescient? He seems to have “seen past” the results of photography, film, and computer imagery, and derived the existence of a category of raw conceptual material which all technical images have in common. Flusser seems to have been informed by the uncanny ease with which cameras transform a twitch of a finger into a fully-formed image. In fact, the ability of people to control events, even large-scale events, via “fingertips” and “keys” is a central point in his theory of the emerging “telematic society”.

Just as AI image models haphazardly transform text prompts and directions into likely images, cameras transform “the effects of photons on molecules of silver nitrate into photographs in just the same way: blindly. And that is what a technical image is: a blindly realized possibility, something invisible that has blindly become visible.” Even though a camera only outputs an image when it is directed to, whether by a person or by a machine, its aperture takes in light constantly, in different angles and directions. From this viewpoint, the process of taking a photo is just the selection of a certain slice of all possible images that the camera could be exposed to, just as the output of an image generator corresponds to a certain slice of its data distribution.

In Flusser’s words:

“For example, if we look at the gesture of a photographer with his camera and compare it with the movements of a fully automatic camera (as in a satellite), we are tempted to underestimate the task. For it looks as though the fully automatic camera is always tripped by chance, whereas the photographer only presses the release when he approaches a situation that cor- responds to his intentions. But if we look more closely, we can confirm that the photographic gesture, in fact, does somehow carry out the apparatus’s inner instructions. The apparatus does as the photographer desires, but the photographer can only desire what the apparatus can do”

Flusser proposes the existence of “envisioners,” a new class of creators who use all forms of apparatuses to create technical images:

“These are people who press the keys of an apparatus to make it stop at an intentionally informative situation, people determined to control the apparatus in spite of its tendency to become more and more automated and so to preserve human judgment over the machine. Envisioners are people who try to turn an automatic apparatus against its own condition of being automatic. They cannot create illusions without the automatic apparatus, for the stuff to be envisioned, the particles, are neither visible nor graspable nor comprehensible”

In other words, Envisioners are the people who “see past” the apparatuses, and the objects produced by apparatuses, and who use the apparatus as a tool to secure desirable outcomes from the raw material of the possible.

Meanings, Redundancy, and Pierre Menard

“Forgetting must achieve equal status with learning and be recognized as equally critical to information strategy. Third, it will become possible to delete redundant information (that which is already stored elsewhere) from specific memories. Redundant and informative situations will

have to be systematically distinguished.”

What makes an outcome desirable? Flusser proposes that the primary criterion should be an axis between the redundant and the meaningful. An object is completely redundant if no new information can be gleaned from it.

“One can perhaps classify these images according to what level of information they contain: whether they are more or less informative, surprising, predictable images. I could say of a photograph of the cathedral in Florence, for example, that I had seen similar things many times before and that the image means almost nothing to me, and I may be able to say of a computerized image of a four-dimensional cube that I had seen nothing of the kind before and that the image was therefore meaningful; that is, I cannot distinguish between depictions and models, but I can distinguish redundant from informative images. Of course, I have not said what but only how the images mean—and that is the ap- propriate way to look at technical images”

Because of the impossibility of determining the means by which an image was produced, the only criteria we can evaluate an image by are the criteria pertaining to our experience of the image. Even Flusser’s example here feels like an incomplete exploration of this idea. It’s not sufficient to simply compare the contents of a technical image, you have to compare its effect on its receiver. Did this particular image of a cathedral, seen at this particular moment by this particular person trigger a cascade of meaning, a reframing of implicit rules? Or did it slide right over their awareness, triggering nothing? Importantly, the informational and cultural context frames the object. This means that the same object, seen again in a new light, can take on new meaning.

When Pierre Menard produced his Quixote, he had to adopt an entirely different mindset from Cervantes to produce the same words. Readers of Menard’s Quixote, armed with the knowledge of its origin, see it in a totally new light. The narrator of Pierre Menard informs us that “the Cervantes text and the Menard text are verbally identical, but the second is almost infinitely richer.”

Heat Death

“It is a mistake to close one’s eyes to the telematic revolution. . . it contains possibilities for real dialogue of unprecedented richness”

One of Flusser’s main concerns is that these new technical methods of visualizing substances which were previously “neither visible nor graspable nor comprehensible” would necessarily reorganize society, the same way that text reorganized society around rules, codes and laws. Flusser foresees a society in which raw conceptual material is easily, quickly, and automatically visualized, disseminated, and responded to. The result is a society in which information traverses cleanly through all previously understood boundaries. Class, nation, location, and language will provide no obstacle to the reorganization of society around imagery. A society governed by technical images trends naturally towards a structure of senders and receivers, in which the senders provide the receivers with a constant stream of increasingly stale and meaningless information. This will result in ubiquitous social isolation:

“A technical image radiates, and at the tip of each ray sits a receiver, on his own. In this way, technical images disperse society into corners.”

As well as a pervasive sense of emptiness:

“Everyone all over the world will shortly be accessible to us; we’ll be playing chess with someone in the Antipodes and spending an amusing evening with geographically scattered friends around an electronic round table. Only, what will we talk about with these people, when we all have the same, centrally programmed informa- tion? When we are served by the same central memory? And when we are so neutralized that even as our interests appear to conflict, the conflict has been fed into us from the central memory? Even our arguments are empty chatter”

This is what Flusser describes as a “discursive” as opposed to a “dialogic” society. In the discursive mode, information primarily radiates from a few central locations, “the sources of information threaten to dry up from an absence of dialogue, and the society is threatened with entropy.”

The antidote to the “discursive” society is the opening of “dialogic threads” which run directly from person to person. Flusser likens the power of dialogue to a lively chess match, in which a zero-sum game results in spontaneous and novel situations, where the participants derive meaning from “making the most” of the situation. In a chess problem, “both opponents ally themselves against the problem: polemic becomes dialogue.” In a dialogic, or “telematic”, society, information propagates between all points in the “global brain.” The result is the constant dissemination, uptake, synthesis, and redistribution of information in a form of cooperative play.

Entropy and heat death are Flusser’s primary tools for understanding the failure modes of technical images and automated communication. Automated image systems, if left unhindered by any means of forgetting and discarding, threaten to completely saturate the information environment with meaningless information. Since dialogue is so essential to the synthesis of new and useful information, the discursive society is particularly prone to informational heat death.

A fully automated society is also likely to succumb to entropy. The role of the “envisioner” proposed beforehand is to sort the improbable, informative, and desirable outputs of systems from the probable, uninformative, and undesirable outcomes. Without a good system of discernment, the probable and uninformative are destined to dominate. Whether any given person or system of persons will actually be able to withstand, let alone sort, the upcoming torrent of information remains to be seen.

Utopia

“Utopia means groundlessness, the absence of a point of reference”

Etymologically speaking, utopia means “no place.” This is the definition most relevant to Flusser’s predictions for the future. In no uncertain terms, the “telematic society” he predicts is one completely alien to all former conceptions of human life, a “cerebral” society in which all available time will be spent dialogically making and receiving images in a “feverishly involved and passionate state of mind” as the body is medically and technologically abstracted away to minimize distractions from image-making. It goes without saying that automation and robotics are the productive basis for the utopian societies of Flusser’s imagination.

The experience of reading this section was eerily similar to the experience of reading Dougal Dixon’s Man After Man, a “speculative anthropology” book which imagines the genetically-engineered descendants of humanity far into the future. The future humans in Man After Man seem somehow degraded and diminished, despite their improved adaptation to their environment, and I can’t help but feel the same way about passages such as:

“Telematic society is a unique sort of ant colony: an ant colony be- cause it has a mosaic-like structure in which all functions interact cybernetically, and unique because rather than working, a telematic ant will sit in its cell and spin apparitions, technical images, pure art. There will be brains that are linked through a dream-secreting superbrain to each other and to artificial brains. And yet there will be bodies attached, like anachronisms, to these brains, bodies that demand to be nourished, to reproduce, and to die: spoilsports. These bodies, these spoilsports, these pretelematic participants in the telematic game must be pushed to the margins of view, behind the back of the player staring at the screen, because they cannot be completely eliminated. And this consideration for bodies, this regard for them, this looking back to pretelematic conditions will make them appear continually smaller, less interesting. They will shrink. Everything physical, everything voluminous is already beginning to atrophy”

I feel pretty strongly that the human experience of well-being, the experience of embodied personhood, is linked not only to the brain and central nervous system, but to the health and integrity of the entire body. Linking myself as a “brain” into a “dream-secreting superbrain” feels unnatural and perverse in a way that’s hard to describe. The last third of this book is a pretty wild ride with a lot of interesting science-fiction speculation on the reorganization of society.

Chamber Music

“But only synthesized images are really conceived musically and made musical with visualizing power. It will become pointless to try to distinguish between music and so-called visual arts because everyone will be a composer, will make images. The universe of technical images can be seen as a universe of musical vision.“

After the “utopian” visions which dominate the last third of the book, Flusser focuses on a dialogic point of reference which is much more human in scale and implication. He likens the dialogic society to a society of musicians without a leader, an improvisational group such as a chamber music group or a jazz ensemble. In improvised musical environments, every action is based on, and interpreted through, a shared and evolving context of theory, rhythm, and dynamics.

This vision of the dialogic society is a lot more palatable, as a world in which people create and share statements, melodies, challenges, and responses at the edge of their abilities as their primary activity. Presumably, these musical dialogues will take place at various levels of abstraction which can be communicated to the machinery of the automated society.

For right now, this is a pretty far-flung prediction. Flusser hits a lot closer to the heart when he’s analyzing the existing and near-future state of affairs. But the chamber music metaphor has a lot of appeal. Through cooperation, leadership becomes superfluous. Through improvisation, authorship becomes superfluous. In other words, the consensus that forms is a hyperstition, an idea which infects reality and reproduces, brought to life by the skillful and willful application of technique.

And hyperstition should never be underestimated. In Don Quixote, whether by Cervantes or Menard, it’s Quixote’s erstwhile squire, Sancho, who ends up most transformed by the high-minded dreams of chivalry and adventure.