Worlds Made Without Hands

I. Procession

“A map is not the territory it represents, but if correct, it has a similar structure to the territory, which accounts for its usefulness.” — Alfred Korzybski ¹

I see cameras popping up on every surface around me. Lidar cameras stuck onto self driving vehicles. Time-of-Flight sensor arrays in the ceiling of cashierless grocery stores. Automatic License Plate Readers perched on bridge toll booths. Further out, into the horizon, I can see infrared sensors mounted on drones as they fly over cropland, cameras mounted on factory lines to check for defects, and multispectral telescopes rigged onto satellites to snap photos of the planet’s landscape. From the telescopic to the microscopic, the cameras have it covered.

Cameras are the canary in the coal mine for new modes of being in the future. In a rough translation of the history of image making, cameras have led the way, like a torch at the end of an outstretched arm, into new worlds. Since their inception cameras have intercepted and conceived emerging notions of reality. The images produced by cameras have enshrined, indexed, demeaned, abstracted, documented, archived, poeticized, analyzed, and engendered a single world into many. It has transformed the linear into the non linear. The image, and by extension the camera, is a predictive and a prescriptive force. The non linearity of the image came long before the non linear mode of being we have today. Our mode of being now is defined by a nonlinear patchwork of internet communities, international proxy politics, and interrelated ecosystem collapse.

The present day camera is shedding light on a new reality of emerging data worlds. Not only are cameras capturing this world of data but are themselves becoming embedded in it. They have shed their glass optical lenses and dispersed themselves as an ambient sensing layer of pure data. The cameras have become bigger than their subject matter. They have enveloped the world. If cameras predated our nonlinear existence, do they also predate our virtual existence?

Virtual existence is a mode of displaced and replaced being. It’s a vaguery that can only be experienced. Flow states, ascendence, engrossment—these fall under the umbrella of virtuality. It is simultaneous destruction and resurrection of the self. This is the effect that images have had on viewers since the age of painting. When you look at an image you, the viewer, are first destroyed. If the painting has linear perspective, the painting then reassembles you to be at the center of its universe. A cubist painting on the other hand will barely reassemble you at all. The images today—coming from cameras distributed across the landscape and dispersed inside databases—like paintings first destroy the viewer. Unlike painting, these images reassemble us into the world of data. They amplify virtuality to a degree to which we must understand them as beyond “image”, they are something more encompassing.

The camera of today are data cameras, which we can call “seeing machines”. These cameras are at the forefront in a historical procession of image-making techniques leading away from material reality deeper into data worlds and virtuality. Before there were seeing machines there were the techniques of painting, photography, and coding. Just as a painter makes paintings by painting, and photographers make photographs by photographing, and coders make programs by coding, the new seeing machines make worlds by what I call “worlding.” Worlding is a word to describe the encompassing nature of new images. Worlding is the handless process of making images in and of the data world.

Worlding is a progression from previous techniques. Painting and Coding are handed productions, in that a skilled artisan uses their hands to manually create images. These handed processes are backed with intention, subconscious or not, and as such the resulting painting or program is imbued with a bespoke vibe of the auteur. On the other hand, photography is a predominantly handless process. Early conceptions of photography described it as an “imprinting” process. Images were seen as an imprint of the material world. Thinkers would later re-coin imprinting as indexing or archiving, but effectively it means the same thing. To this day images are still seen as an index of reality: they are used in the courtroom as evidence and in medical procedures to diagnose illness. Worlding is like photography in its ability to imprint the world, but the world it imprints is the world of data. Worlding images are made inside the data realm by seeing machines. These are software-based cameras—lensless—which can scrape and algorithmically congeal images from raw bits and bytes. More so than making imprints of the material world, seeing machines make imprints of the ecology of digital media. GANs, photogrammetry models, object detection systems—these applications are heavily software based and use as their base reference other digital material like user-generated selfies, food pics, and landscape photos. As such these images are best seen as imprints of that material, rather than an imprint of what the material depicts.

Images made by worlding are called worlds. Worlds should fundamentally be understood as data visualization rather than reality visualization or as an index. While a departure from the previous methods, worlding draws upon the common qualities of handlessness and world-imprintedness of painting, photography, coding. It is not an alien black-box system but a historically grounded concept. These worlds made without hands can be understood as a progression in image making techniques. The effect of this progression of technique is also a progression of being. This new being is virtuality.

In the following essay I will explore this new mode of being. I will trace the lineage of previous image making techniques and illustrate how worlding retrieves precedent techniques. By investigating worlding we can see what being in the future could be like. Overarching my argument, or rather underlying it, is a straddling of two minds. One mind has sympathy for the viewpoint of what some sects of philosophy call “informational realism”, which is part of a greater study of the “nonhuman turn”. Essentially these two movements point out an underlying logic to the world that is extra-human. Namely, that the universe is fundamentally composed of information, and matter itself derives from this. It is not averse to human subjectivity, but zoomed all the way out the movements understand information to be the more heavily weighted compass of change. Without getting too far into the weeds the gist of this movement is the intention of shifting perspectives toward ambience. It is to understand the world in a non-human-object antagonism but to understand the world as a whole in which people are a part of it. While informational realism is useful, I also ascribe to the belief that the experience of this foundationally information world is itself not foundationally informational. More specifically the experience of virtuality is disparate but not contradictory to material reality. Seeing images and using images are epistemically different that the methodology of making images. Being in worlds of information does not make the being itself informational. While images can be boiled down to pure information, people’s experience of being cannot. The experience of being confers agency, and agency seems to ride against the tide of a predeterministic universe of information. Virtuality and reality are complementary, in my mind, which makes the exploration of worlding all the more useful, as it perfectly straddles the distance between the data and material reality. It is with these two disparate minds that I approach the concept of a world made without hands.

Worlding should be understood as a method of distinguishing between real and the fake images. It’s a device to understand images beyond their face value and to see them as assemblages of data. My goal with this essay is to dissolve the “immersive magic” of worlds made without hands and emphasize the legibility of it. At the same time, I believe the long term purpose of fluency in seeing machine is one of cyboric “companionship” as Donna Haraway might say. Which is to say media literacy is a tool for finding seams and reveling in them. I want to reinstate a cosmic fascination with being in these worlds made without hands by offering ways of existing in it that still confer agency. A world made without hands is something to teach, regulate, play with, rather than ignore or shut down. It is paramount that we know how to speak seeing machine, because seeing machine already speaks us.

II. New Acheiropoieta

Worlding is an accumulation of the mediums that preceded it and feed into it. For instance, worlding retrieves an ability to “imprint” the world in the same way photography is a medium that takes imprints of the world. Beyond retrieval worlding also inverts old understandings of control, obfuscates the artists hand, and enhances the novel notion of data worlds. Worlding is not a straightforward development of image making technique but is instead a complex remix of all that came before it. To understand worlding then requires a knowledge of the paths that run through painting, photography, and coding.

A painting is a reflection of the artist who made it. Like the carving and etching techniques that came before it, painting requires an adept and trained hand to configure an image. With painting one could depart from the basic linearity of text and speech. Painting demonstrated a perceptive simultaneity that text and speech could only achieve over time. Speaking and writing occur in sequence, and while the worlds language can generate can be non-linear, the fundamental delivery is one word after another. Painting is the ability to directly extend the hand of the artist and with one brief flash make in the viewers eye something that language could not.

After painting came photography. A photograph still reflected the photographer that made it, but to a lesser and different degree than painting. It represented the trajectory of image making techniques towards the increased extension of the artists hand away from the body. Photography marked the beginning of the departure of the artist’s the hand by automating a process of imprinting reality into a new image.

I should note that while I describe these phase changes as hard lines, their transformation over time is soft and blurred. For instance the line between painting and photography is blurred and grayed. Media historian Lev Manovich notes that famed technical photographer Edward Muybridge sometimes utilized painting instead of pure photography. In some of his lectures on Zoopraxiscopes, a kind of early motion picture projector, Muybridge would use manually painted images of slides to represent the photographs he would later take.² We can also observe that early “color” photographs and “color”films were made by overpainting black and white film with colorful translucent pigments. This is to say that developments between media are hardly linear. The development of image making techniques is more accurately like a cell dividing into other cells; mitosis on blurry edged mediums.

Photography as a “handless” medium was itself a blur of change. The iconography of the hand in photography was present in its earliest days and today it in fact a prescient way of conceiving of images made by seeing machines.

In 1888 George Eastman released the Kodak #1 camera. It was a hand-held sized device designed for amateurs to take casual photos. It was a revolutionary camera for its ease-of-use, and quickly gained the name “detective camera” for how discrete and invisible they could be—people hid them in derby hats, binoculars, and canes. The slogan caught on too: “You press the button, and we do the rest.” To many working fine art photographers the Kodak camera represented low brow art for non serious work. Serious photographers turned their nose up at it, calling the people who used such devices “Button Pressers.” Alfred Steiglitz, a prominent fine art photographer at the time, actually embraced the point-and-shoot style cameras, which he called “Hand Cameras.” In an 1897 article he wrote that the “hand camera was in very bad repute with all the champions of the tripod.” He continued, adamant on the usefulness of such a device,“the hand camera has come to stay—its importance is acknowledged.”³

The notion of the hand in photography has since evolved. In 2006 the artist Sascha Pohflepp refashioned the hand camera and reframed what it means to be a “Button Presser.” His interactive work titled Buttons is a camera without any optical lens. On top is a big red button which when pressed will cause an image to pop up on a small LED screen on the back of the camera. The image that pops up is one that was grabbed from a random place the internet which happened at the exact same time as when the button was pressed but uploaded by another person. The user effectively plucks an image from the internet database. In the context of the historical Button Presser, this art work succinctly conveys the phase change in image-production. With the Hand Camera, Steiglitz notes, the general public would shoot off photos “helter-skelter, taking their chances as to the ultimate result.” The notion of chance which emerged from the hand camera is also foundational to Buttons. Ironically, within the Hand Camera-esque work is the mode of handlessness, while the magnitude of chance has been globalized. The phase change then is that photographer does not choose what photo to make, they instead get a photo, and the photo they get was not made by their hand.

Buttons, Sascha Pohflepp, 2006 ↗️

The Hand Camera of the present today is like that of Pohflepp’s. Buttons taps into the trend of the convergence of images into databases, but also speaks to the individual experience of it. The experience is a profound inversion. This inversion flips the process of image making. It is not taking a photo then sharing it, it is finding a shared photo and then taking it. An image is not a text to be read but is actually text which happens to be an image. The image is not made, it is spawned. The inversion also retrieves the early debates of photography: Is the camera taking the image or is the photographer? Who is control and does anyone have a hand in it? What do we believe images to even be? Through the automative property of databases and algorithms coded to make images from these databases, this inversion is amplified and scaled up. In it we find a mirror house of infinite and unexpected meaning.

An early deliberation of photography framed it as practice without human intervention. In 1844 Henry Fox Talbot stated in the introduction to his photo book The Pencil of Nature, one of the first photo books ever, **that his images were not in fact made by him the artist, but were instead “impressed by Nature’s hand.”⁴ Talbot’s framing of photography was device-centered. For him, in a triangle between nature, camera, and human, the photograph existed as a sequence of power emanating from nature directly into the camera. Tablot’s successors, like Edward Muybridge and Harold Edgerton, would extend this Enlightenment-inspired thinking to a brand of photography centered around scientific and objective capture of reality. Reality became a window which opened itself exclusively to cameras.

Even before Talbot, and before the “official” invention of photography in 1826, some argue that image-production was always about impressions—in the sense of imprints, but also in the sense of interpretation and reception. David Levi Strauss argues in Photography and Belief that the notion of photography has biblical roots. The core example is The Shroud of Turin, a famous Christian relic. The shroud is a four by one meter sized piece of fabric with a faint dual-image stain of a male figure. The impression appears to depict a crucified Jesus Christ laid to rest. Strauss writes, “Believers claim that the shroud is acheiropoetic… it is a direct emanation of the Son of God, unmediated, without human intervention or subjectivity: an objective trace.”⁵ “Acheiropoieta” is Medieval Greek for “icons made without hands”. Acheiropoieta are objects observed by people which seem to have just appeared from nothing, as if it were a transmission from an other, usually spiritual, realm. Christ’s image is acheiropoetic in the sense that it is an uninterrupted succession of influence from Christ’s body to the fabric. Contextually, Christ himself is described as “the image of the invisible God” (Colossians 1:15). The Shroud is thus an image of an image of God himself, made without hands.

At once a double impression, like enlarging a negative in the darkroom, the Shroud is also a surface of divergent interpretations. Strauss cites examples: Researcher Nicholas Allen proposed an alternative history to the Shroud, based on carbon dating analysis of the fabric which places it somewhere between 1260 and 1390 A.D.—much later after Christ’s death and when image-making capabilities had increased. Allen’s theory suggests that early photographic techniques were used to make the Shroud’s image, including the use of light-sensitive chemicals, a camera obscura, and optics. Other historical theorists tagged onto Allen’s theory and think that the shroud could have been made by Leonardo Da Vinci, who would have been one of the only figures at the time to have been able to pull off this technical feat—on top of that, the theorists claim that the impression could actually be of Da Vinci himself! Of course simultaneously there are those who do believe the provenance of the image as an imprint of Jesus. Others still, take a more neutral approach, like the Catholic Church, who neither denounce nor confirm its authenticity.

The Shroud of Turin embodies the complexity of narrative of all acheiropoieta. Acheiropoieta gestures to the lore and myth surrounding an image’s origins and suggests a multiplicity of impressions embedded within itself. The effect of this becomes a discursive focus on what “original object” made the impression and through what processes. Was it Christ? Da Vinci? An image of God?

Today, there is a New Acheiropoieta. Seeing machines can make images without hands. Through a massive network of automation they generate magical imprints that feel like new photographs. A “seeing machine” is a technological device that extends the human eye to see further, smaller, and in more detail to augment our understanding of the world.⁶ We all have seen a photo from a seeing machine. We feel like we know our planet because we’ve seen a selfie of the whole Earth taken by a satellite. It situates us. But a seeing machine is an also a framework of thinking about the ways cameras extend their own eyes. Defined by artist and photographer Trevor Paglen, and originally coined by Vilém Flusser, the term “encompass[es] the myriad ways that not only humans use technology to ‘see’ the world, but the ways machines see the world for other machines.”

Seeing machines make worlds without our hands but with the electron-sized hands of a network of computers. This networking of edges, connections, and links can be understood, collectively, as composing and pulling the veil back to reveal the world of data. Google Earth and the Black Hole image, which are analyzed later in this essay, are examples of present day worlds made by seeing machines. Just as Nature was to Henry Fox Talbot’s camera, and as God was to the Shroud of Turin, so is data to worlds made by seeing machines. In my view, the original object of the New Acheiropoieta is data. Data is that cohesive, discernible figure which is impressed into seeing machines to make worlds. The world that seeing machines capture, make images of, “imprint”, is the world of data. Seeing machines don’t make images of the material world.

This might be contentious. Just like the actual Shroud, there are conflicting theories around what the original object of New Acheiropoieta is. Some see the processes as a black box of technical jargon. They see the the true originators of images—even those made by networks of machines—as direct emanations of people and culture. These thinkers understand image making as fundamentally an anthropomorphic concept and as such images are not images without human intervention. For them, the original object of images made by seeing machines is the human figure. This bias towards cultural determinism can muddle the fact of what author Joanna Zylinska sees as “the inherent nonhumanity of all vision”.⁷ Human vision for Zylinksa, is a “part of a complex assemblage of perception in which various organic and machinic agents come together—and apart—for functional, political, or aesthetic reasons.” Indeed the consequence of seeing machines is that they open up an entirely expanded definition of “seeing”. Seeing, and I agree with Zylinska, has today been exploded to be an “assemblage” of human-made artifacts and the nonhuman. This grayer and more blurred take more accurately encapsulates the thrust of these new images, these new imprints. It suggests that the notion of the hand, emblematic of the human, once at the center of sight and image-making, through painting, and in part photography, is distancing itself from seeing. The eye and the hand can be symbolically delinked in the this new nonhuman taxonomy of seeing. For these kind of thinkers, whom I more closely align, the original objects becomes that which is outside anthropomorphic definitions of seeing. The original object is data.

The New Acheiropoieta points to inherent mythologizing of these worlds made without hands. Worlds made without hands are made by seeing machines, which are not a unique ahistorical abberation in a suddenly new method of conjuring images, but as I argue, retrieves some of the basic notions of imprinting from photography. This mythology is up for interpretation; the two large camps of which I have just outlined. As such these seeing machines should not be understood as ultra modern devices but as a retriever of a much older method of making images: shrouded in mystery, and made without hands. These image worlds elicit discussions about autonomy—did the computer really do it, or is the computer just doing what someone told it to? Through a discussion of the the image making technique of coding, the next in the procession after painting and photography, I think we can explore these questions, and find the ways that worlding revives these same questions.

III. Information Inversion

There is a sentiment rooted in computational thinking which is that machines only do what they’re told. This was first uttered by Ada Lovelace in 1843 in her notes about the Analytical Engine, an early concept of a computational device: “[It] has no pretensions whatever to originate anything. It can do whatever we know how to order it to perform.”⁸ The situation of seeing machines today is in complete alignment with this sentiment, still, almost 200 years later. It’s not magic. They do what they are told. However the twist is that more and more machines are told what to do, not by people, but by other machines. For instance, the internet, which we conceive as a primarily network of human activity, is in fact mostly bots talking to one another.⁹ In a network a person may originate a message, and hardcode its destination into a system, but the message still takes on an unexpected grammar as it is passed from node to node. Machines use a patchwork tapestry of maps and models to “see”. The sampling from this tapestry forms a web of noise of such rich complexity that the emergent grammar must be distinct from “machines just do what they’re told.” In a game of telephone with a billion participants, did the first person really **tell the last machine what to say?

Seeing machines make images with this new grammar that goes beyond what they were told to do. This property can be traced to the image making technique of coding. Coding images often relies on generative processes—utilizing chance, randomness, and iteration. Worlds take on the same quality of chance, randomness, and iteration, and take on the effect of that quality as well: bottomless and profound interpretation. We have always assigned generated things, things which just seem to appear, the things that are born, a sublime meaning. They persist on their own, with their own strange loop, and insist on their own “I”—a blooming perennial, a patch of grass bursting through the sidewalk, a newborn kicking and screaming in the car seat, the sun rising and setting, and a ticking clock.

There is also a quality of the sublime within computer programs. A chatroom, an open world video game, a digital canvas, a looping screensaver, and the desert of an empty spreadsheet. These programs feel sublime, but in a different way than the feeling of walking under giant trees. I think they feel sublime because they invert something. I think they invert the power of their aesthetic, their outward image, to point inwards, toward the magic that these programs can exist at all. The power of the image is not in its form as a final imprint, but instead the power is in how in the image was made. The focus becomes on the processing of the image. This focus results in an inversion in value: suddenly the part of the images that matters most is the information inside of it. The “image” of the chat stream, the “image” of the open world video game, and the “image” of the digital canvas fades away and instead the vastness of these tools overwhelms the senses. The tools become portals and vessels that hold and open up into a world of data.

John Conway’s Game of Life and Craig Reynold’s Boids are two computer programs, from the 70’s and 80’s respectively, which simulated life-like behavior based on very simple rules or parameters. These programs summon a sense of lore on their surface. It is like watching alchemy occur when you watch the cells and boids undulate in the screen. If you experience these pieces without prior knowledge you would have a hard time deducing the rules. Same with the other way around—read just the rules and you would have a hard time envisioning what the program would look like when executed. This surprising emergence of complexity from simplicity is a quality of many new acheiropoieta. Indeed these aptly named “artificial life” simulations point to the genetic quality of digital code: discrete, binary building blocks folded into infinitely complicated and unexpected aesthetics.

There is a tension between what one can call the “materiality” of these simulations, i.e. the visualization of cells and boids moving on the screen, and the underlying “information” which structures it— the code, the rules. It’s clear that one controls the other, but which way? In The Condition of Virtuality, N. Katherine Hayles compares the dichotomy of materiality and information with developments in molecular biology. She notes a turn in 1945 in scientific discourse which proposed the idea that organism reproduction might be governed by informational code. At a rhetorical level, an “impossible inversion” occurred, which flipped the then-logical idea that the gene was contained in the body into: the body is contained within the gene.¹⁰

In other words, information “becomes the site of mastery” of control over material. She writes, “the body’s materiality articulates a preexisting semantic structure. Control resides in the pattern, which is regarded as bringing the material object into being.” While the Game of Life and Boids are not explicitly images, their construction of a dichotomous material-information demonstrate a control of the pattern. They are not operating at database scale, but are to a degree autonomously collaging data; the data is in this case code, which animates the simulation into being. One can understand them as primitive, inverted digital cameras. In the same way CMOS sensors translate captured light to produce coherent photographs, these programs translate capture their code to produce coherent animations. The data is out there, already in existence, and is plucked and summoned into motion.

Source↗️

Source↗️

Perhaps this is the original model of image making. The Shroud of Turin is an example in which the story about the body impresses itself on the cloth more so than the actual body. Information about the body overrides the actual event in the matter of importance. The dominance of narrative, patterns of speculation, theories of original process, outweigh the actual body imprinted into the shroud. What matters is the story.

A distinctive symbol of New Acheiropoieta is that of GANs (Generative Adversial Networks). GANs are the combined effort of two competing seeing machines. Both deriving from a singular data set, one machine, the ‘generator’, tries to invent images to fool the other. The other machine, the ‘discriminator’, tries to discern whether the data was faked or part of the original data set. This process automatically loops until the discriminator is adequately fooled into believing the image is authentic. The new image then is presented as authentic to the GAN, which is how GANs can produce images of face of people that don’t exist based on a dataset of faces of actual people.¹¹ It is this loop which is handless, and generates, through poiesis, a bringing-forth and birthing, images made without hands.

The output images of GANs today are goopy and dream like. Familiar patterns appear and disappear. Likewise, we experience the illusion of pareidolia and see familiarities and faces where there are none. Like other examples of worlding, the liquid texture provokes picnolepsy; an unconscious lapse in our own discrimination. The image is permanent in-betweenness and perpetual copying in motion. But behind the fluidity is the legible rigidity of the worlding process. Casey Reas, an artist who works with code, has an illuminating framing of GANs: “I think of a GAN model as a complex camera. Like a camera, a GAN is an apparatus that can be used by an artist to make pictures… they assist with creating unexpected images, unlike any that have been created before. They can be unlike photographs and paintings—they are truly something new.”¹² Reas emphasizes the idea that GANs are deliberate extension of humans, like an artist’s brush, but also points to the unexpectedness of them as well. This duality between intentional and unintentional carves out and reveals, as if through a seam, a third space between painting and photography. They are a camera, in that you can choose what environment to set it in. But you can’t choose where it points. The dataset of images is under the control of the artist—they decide what goes into the system. Landscapes, portraits, still lifes. But the GAN becomes its own painter and uses the provided images as a palette to illustrate its own becoming. Contrary to Ada Lovelace in some ways the GAN does not appear to just do what it is told and is writing its own story.

A modern day Shroud of Turin would have an impression of a cyborg on it. A cyborg, emblematic of the space between machine and man, and between photography and painting, and a messiah figure for some, stamped into a length of fabric. GANs are the cyborg of modern cameras. They take from both the oldest methods of image production and the most contemporary. Like the Shroud of Turin GANs are an imprint of data onto its surface made without hands. Like the The Game of Life and Boids GANs can generate images that feel alive as if birthed through a genetic code. Like Pohflepp’s Buttons, GANs are a camera that plucks from a dataset of images. They encapsulate a variety of processes which itself becomes the grounds for it own a mythology.

It is the reinstatement of mythology into the process that breeds a divergence of meaning. New Acheiropoieta becomes a surface of speculation and conspiracy. It is precisely because the processes create emergent complexity from simple rules that there can be infinite meaning extracted or projected onto a world made without hands. The original object of data is hidden by the image surface. If the mainstream belief is that the camera captures imprints of reality, I hope now is that it is more readily apparent that the images produced by seeing machines are making imprints of the data world.

IV. Worlding

“The mass image agglutinates every image, still and moving, into a single vast, inter-connected artifice.” —Sean Cubbitt¹³

The ways machines see for other machines is increasingly engendering a new meaning to images. To a seeing machine an image is fundamentally hypertext. Images are not singular visual descriptions of decisive moments. They are nodes in a network of other images. They are non-static files flying around, intermingling, interchanging, and interoperating with other files in relational databases. An image is more like a conglomerate of strings of text and numbers— like a BLOB, computer science speak for Binary Large OBject, or “a collection of binary data stored as a single entity.”¹⁴

Worlding reveals to us the true nature of digital images: data visualization. This truth has been hidden behind a cloak of convenience. The obstruction is akin to Paul Virilio’s concept of picnolepsy, which is a condition of frequent lapses in consciousness, gaps in perception and being, caused by a high speed of living.¹⁵ The artificially seamless design of image making, saving, and sharing has few affordances that cue our consciousness to the underside of images.

The underside of images is text, mangled and raw. The seemingly simple process of opening an image of the computer obscures a violent reconstruction of raw data into a human-readable JPEG. In the same way our own minds reconstruct a memory anew every time the brain retrieves it, machines reassemble an image anew every time the network retrieves it. Digital images, as we tend to know them, are in fact highly synthetic visual ephemera of otherwise machine-native strings of text. Text becomes the foundation of images rather than a footnote, caption, or description of them. As media philosopher Vilém Flusser puts it, “Text, originally a metacode for images, can themselves have images as metacode.”¹⁶

This idea that images are the effect of interoperating bits of text is a development in the grammar of seeing machines. I call this grammar “worlding.” Worlding is the process of collaging done at a database-scale by autonomous machines. It’s a montaging of highly specific computational photography styles. Worlding is what the Black Hole image, Generative Adversarial Networks, and Google Earth have in common—unstructured data from sensors and inputs automatically acquiescing into a contiguous and virtual sheath.

We can begin to see images as data visualizations with some present day examples. Open Google Earth and you will be able fly a virtual camera smoothly over oceans, savannas, and cities. The entire Earth’s surface is rendered as a seamless 3D image. This seamlessness is made possible by a technique called the Universal Texture, which was patented by Google in 2000.¹⁷ The Universal Texture is a software technique that allows for the efficient loading and rendering of a digital representation of Earth. It is able to do with by storing the image of Earth into a singular, couple-dozen-terabyte virtual image texture of varying resolution density. When a user flies through the world they only experience the world at their subjective scale. So, in real-time the Google Earth system takes a slice of this unified and expansive texture and gives it to the user to look at. One of the creators of Google Earth adjacent to the patent, Avi Bar-Zeev, describes the tech like “feeding an entire planet piecewise through a straw.”¹⁸

Google Earth images are captured from a variety of devices. Satellites, aircraft, balloons, and kites take images at different angular perspectives across different decades. These at once discorrelated images merge together into a singule texture. This is the idea at the core of worlding: collaging, or the gluing of heterogenous material. (“Action of gluing” is the etymological meaning of collage, which comes through French from Greek.) Every patch of the Universal Texture is “glued” with other images, stitched and patched to make sure clouds, inconsistent daylight, and unwanted edges between photos are omitted. Moreover, the sheer amount of raw images used to generate a whole Earth (petabytes, plural) is more than can be stitched manually, so the texture is necessarily generated with automated computational photography processes.

The Universal Texture is an example of the product of worlding: a seeing-machine-generated terracollage. But the Google Earth example also points to a crucial property of worlding, which is its delivery of an entire planet through a “straw.” The straw is a useful tool to describe the experience of Google Earth because it is based on the fact that the user-controlled camera can only be in one place at one time. The individual experience of something derived from worlding is narrow; the worlded environment is dynamic and foveated relative to each of its inhabitants. As if wearing horse blinds each viewer has a fixed render distance in a bubble around them, resulting in each inhabitant being fed their own proprietary copy of the same world. This “strawness” is an integral caveat in any individual’s study of worlding.

Strawness tells us that despite one’s god-like view of Google’s Earth, you cannot truly see the whole picture. Only in moments of breakage do the seams become apparent and reveal what is underneath. “Google Earth is essentially a database disguised as a photographic representation,” remarked Clement Valla, an artist who has investigated Google’s Universal Texture.¹⁹ In a 2010 series titled Postcards from Google Earth Valla took and printed screenshots of areas of Google Earth where terrain height data and satellite texture data misaligned. These two competing inputs were generated by separate parties: agencies scanning natural topology, and satellite companies scanning manufactured infrastructure. The resulting image encapsulates a misalignment of natural geology and manufacture landscape, and in so doing tears a hole in the veil to reveal a naked seeing machine. This “mistake” subsequently makes more real the database beneath the virtual soil. “These images are not glitches. They are the absolute logical result of the system,” Valla emphasizes.

Valla’s remark is a tenet of worlding. Images are the outer shell of data bases. Worlding is about how databases use their own logic to materialize, visually aligned or not, into a massive surface. Google Earth’s materialization of their database is particularly skeuomorphic. It’s not the only project that does this either: DARPA’s Transparent Earth project²⁰ and Niantic’s (Pokemon Go’s parent company) Planet-Scale AR Alliance project seek to do similar things.²¹ These organizations want to map a 1:1 digital twin of the world and make a contact print of the planet.

This is important to point out because not all worlding is skeuomorphic. Non-skeuomorphic also occurs, albeit the output is not colloquially understood as an image. For instance, the 31 GPS satellites pointing at the Earth right now can triangulate and identify earthquakes. They can even sense changes in soil moisture by small movements in the ground.²² Images produced from this data are heatmaps, a sort of nonhuman x-ray vision of the world. This is to say that even devices which not usually understood to be cameras are in fact capable of worlding. Rather than measuring light these devices infer data through other extrasensory measurements. If it worlds, it’s an image. Worlding without use of light perhaps more clearly illustrates that all images are data visualizations.

Take the famous image of the Black Hole at the center of our neighboring galaxy, M87. At first glance it looks like one seamless albeit blurry image. In reality, this image is a collage of fractured noise. Its processing was anything but coherent. The image data was collected with an array of eight different telescopes from around the earth collectively called the Event Horizon Telescope (EHT). Each was precisely oriented and timed with an atomic clock. The sensors picked up millimeter length radio waves. In total five petabytes (or five million gigabytes) of image data were captured, aligned, denoised, and interpolated with software to output the famous orange glow we know today.²³ Design researcher Benjamin Bratton notes: “To make this image, our planet itself became the camera… The mechanism is less a camera than a vast sensing surface: a different kind of difference engine.”²⁴ This new sensing surface is a data visualizer. It’s a “difference engine” that compares what it is told to compare. Through comparison it can see.

The image is the result of an engine that reconstructs, reassemblages, and refactors contrasting perspectives. The algorithms used to denoise the data were heavily vetted by scientists who pre-ran them on synthetic data, and compared their results to what was expected, to ensure they were not generating biased results. The phase of discerning the Black Hole signal from the petabytes of captured noise was conducted in four independent groups, each using their own vetted algorithm, to produce their own “fiducial images.” These fiducial, or reference, images were subsequently blurred and averaged together to form the singular final image which we know today. Without a true reference image of a Black Hole to compare the results to, the team behind the EHT had to rely on a long process of comparative vetting to ensure their data-based interpretation of the image was as honest as it could be. Perhaps it is fair to say that the Black Hole image is made more out of relational edges than out of solid pixels.

Understanding that images are data visualizations speaks to the fact that images are processes rather than fixed things. Artist Hito Steyerl describes “poor images” as “copies in motion,” and that “the poor image is an illicit fifth-generation bastard of an original image.”²⁵ Steyerl intended the term poor images to describe a lower political class of images; over-compressed, over-shared crumbling thumbnails, proxies, metadata. Poor images are usually ignored are but extremely active for they derive their identity by being exchanged. Memes, mezzanine files, temporary caches: these files types are poor images. Thus in a way the final Black Hole image can be understood as a Frankenstein’s design of fifth-generation fiducial images. It’s not a poor image, but it is strangely the output of the same process of “copying in motion.”

When I look at the lone Black Hole image I actually see a grid of possible fiducial black hole images vibrating within their grid of cells. The single image I’m looking at is a consensus, a surface agreement, a verb, of a comparative process. The always-in-process of the image indicates that there could be potential images in later succession. Following the momentum of image, could a Black Hole Image 2.0 emerge? Without an aberration in the surface that one can sometimes find in Google Earth, it can become more difficult to find the database of fiducial images underneath the Black Hole image. But it’s there, shimmering under the surface.

The worlding process is not bracketed to the explicitly large scale. Smartphones employ a complex method of worlding whenever you use it to take a photo. “Simple” captures are in fact an elaborate layering of different machine learning models to sharpen details, retouch and recognize faces, computationally blur backgrounds, generate depth maps, and apply pinpointed color balancing. It taps into database-scale image making. These machine learning models are themselves trained on huge datasets of image and captions pairs—like the ImageNet dataset, a collection of millions of manually curated image-label pairs, which fuels object detection models. The model is myopic to what it is fed, and only knows what is given to know. It creates its own maps of meaning, a kind of Universal Texture extracted from its given dataset. The model as scaffolding made of images becomes the basis of production for more images. An image taken on your phone can thus be understood as an collaged expression of precisely these underlying models, maps, and atlases.

In a dramatic sense, smartphone photography is the same as Black Hole photography, and vice versa. At both the planet and at the pocket scale, database-scale image production is occurring in the fashion of collage. They both are an act of worlding, meaning, the coalescing of a database to make images. The ways machines see for other machines is inverting the meaning of images to be about how its data is processed. In other words, the content of the image is superseded by its information. Worlding points to a reframing of images as database, which may sound extreme, but I don’t think this is to our detriment. We aren’t locked in on a one-track future of a world made without rans ruled by databases. There are many paths that this world can take.

V. Worlds made without hands

In a world made without hands one walks on the roots of images. Data entanglements form a sheath, convergent and hardened, sprouting with codec flowers, dirt and detritus of poor images fall to decompose in the soil, under which is a network of mycelium pushing stories and myths around to other plants as if they were nutrients. The roots form a living bridge, to where? What kind of world are we headed towards?

Julia Watson, an architect and anthropologist, documented a series of “Lo-TEK” solutions used by indigenous peoples around the world, from the deep past to the present day. Lo-TEK is part portmanteau and part acronym of “low-tech” and Traditional Ecological Knowledge. One group she highlights are the Khasi, an ethnic group in India who have developed bridges and ladders formed from the roots of the endemic rubber fig trees. Watson notes the necessity of this bridge design to the landscape as “the only bridges able to withstand the force of the monsoonal rains.”²⁶ The Khasi’s bridge design utilizes a latticework pattern, interweaving bark and boulders, and allow the roots to wrap around it and continue to grow through it over time. In such a wet environment an artificial bridge made of dead wood would quickly rot, while the living roots thrive. The bridge is more than utility, it has a symbolic purpose too. Watson writes, “the rubber fig tree is entwined in the Khasi sacred origin story as the tree which forms a mythical bridge called the jingkieng ksiar or ‘the golden ladder to heaven’ that grows to a place in the clouds called ’bneng’ meaning heaven.” In such a way, a bridge made without hands can ascend to another world.

Watson celebrates these designs as a way to “redefine the mythology of technology to include indigenous innovation.” This bridge is one of many examples of highly innovative systems used by indigenous populations. It is at once perfectly suited for the job—allowing people to cross a river—and also has embedded with it an ecologically harmonious and long term relationship with the landscape. Lo-TEK should not be understood as “primitive” design, writes Watson, but as a more intensely future-conscious design system than what is employed today.

In the media landscape we must be inspired and learn from the ethos of Lo-TEK. Namely, the exercise of harmoniously handless creation, in which the landscape flows fluidly with Lo-TEK infrastructure, has lessons for us. The urgency is in the fact that worlding is around us. It’s not an inevitable, it’s already here. Media theorist John Durham Peters puts forth in his book The Marvelous Clouds, “Media are not only about the world; … they are the world.”²⁷ Media ecology, once a distant concept informed by the likes of Marshall McLuhan, has over time manifested into an actual ecology. There is a whole literature of understanding media through the lens geological mining, data center energy use, and other environmental extraction. However for the purposes of this essay the focus is at the semantic level that the sign-signfier link between media and material has shrunk to such a degree that we can fluidly speak of the environment as an entangled mixture of media and material.

As inhabitants on this mass image landscape we have a responsibility to make the world work. To do this we need mythologies that define the systems of worlds made without hands, as well as mythologies that define modes of being in it. It’s no coincidence that a root bridge is both useful and spiritual; the two are mutually supportive. By mapping of The World, The Earth, and The Planet, concepts by philosopher Eugene Thacker, to three disparate systemic concepts of worlds made without hands, I want to help define the varying mythologies of the New Acheiropoieta. These three systemic concepts are the Metaverse, the Noosphere, and the Singularity. Additionally, I want to lay out a mythology of being in these worlds made without hands by diving into the etymology and alternate meanings of worlding. In worlding we find that the process of becoming and being are the same actions. Tugging on this string can serve as a mode of reaping agency in an automated landscape.

In his book In the Dust of This Planet Eugene Thacker explores environmental catastrophe through the genre of horror. Despite an overtone of dread, at the start of the book Thacker lays out a very useful device for thinking about the mythology of worlds. He distinguishes the terms “World”, “Earth”, and “Planet,” to convey different interpretations of the same human-landscape relationship.²⁸

The idea of “The World” is the another way of saying “the world-for-us”. Meaning, that when people say The World, they mean the hermetic relationship between human civilization and the landscape. The World is way of looking at this physical landscape where humans are placed at the center of it. The World describes a human solipsism—the world as itself strawness.

“The Earth”, on the other hand is another way of saying “the world-in-itself.” The Earth broadens the circle of empathy of The World and expands it to all life, not just humans. The Earth concerns the biome, the weather, animals, evolution, disasters, seismic events—all entities not necessarily a contingent human presence. The colloquial “Nature” or wilderness is most closely synonymous with The Earth.

Finally, “The Planet”, or, “the world-without-us”. In Thacker’s use of it, he conceives of the Planet as something without life, and in particular, without human life. It is post-human, yet exists entirely in the wake of us—there is no planet before The Earth. It is an object only in and of itself, and finds no identity in relation to any life form. The Planet is a wasteland, an “imprint”or scar of the life that came before.

The descriptions of the World, Earth, and Planet map linearly to the hopeful and fearful visions of a future technological virtual reality “world” or images. I hope to complement Thacker’s descriptions by mapping them to notions of planetary-scale media apparatuses. This works because the way we interface with world-earth-planet environment parallels the way we interface with a media environment. More specially, by doing this comparison we can outline the possibility space where worlding manifests.

The World is analogous to the base philosophy of the metaverse. The term ‘metaverse’ was coined in 1992 by Neal Stephenson in the sci-fi novel Snow Crash to describe a virtual reality world. In the book the metaverse is the virtual space in which people don digital avatars to work, play, and gamble. Stephenson writes the metaverse as a space privately owned by a chain of mega conglomerates, which as it happens is not too far off base of the term’s recent re-emergence. To Meta founder Mark Zuckerberg’s the metaverse is the new platform on which to construct a proprietary virtual reality. In both fictional and non-fictional conceptions of the metaverse has a dominating and human-centric relation to territory. The World is a linguistic device to reflect this solipsistic and directional understanding of a world made without hands: without humans, the metaverse would not exist. The metaverse represents a methodology of thinking in which humans control and are the ultimate vanishing point of a media landscape.

On the opposite end of the spectrum, and skipping over The Earth for a moment, is the Planet. The Planet represents an ambient re-centering of landscape in the absence of human influence. Running parallel to this idea are the fantastical dystopic visions of a planet run by machines. This is the kind of planet depicted in The Matrix trilogy: a universe in which machines farm human bodies for energy in order to sustain their tyrannical rule over civilization. The simulated virtual reality is a social red herring to quell any uprising. In other words, this universe is the realization of Ray Kurzweil’s vision of a post-Singularity. In the Singularity, the-world-without-us is in the custody by machines. Machines dominate the landscape of which people are a subcomponent. It’s a nightmarish take on the Marshal McLuhan phrase, “Man becomes, as it were, the sex organs of the machine world, as the bee of the plant world.”²⁹ The Planet is thus a marker for the post-subjective media landscape in which nothing exists apart from images. There are only objects. It describes a desert of inbred JPGs, PNGs, and MP4s copying and worlding each other to no other end than to multiply. In this media landscape “we no longer look at images—images look at us.”²⁹

Orthogonal to both the metaverse and the singularity myths, is another myth, closely aligned with Thacker’s device of The Earth. This myth is that of the Noosphere. The Noosphere is a concept conceived in 1922 by the philosopher Pierre Teilhard de Chardin. The term translates literally to “mind sphere,” and was described as a sort of consciousness-sheath for the planet. The Noosphere hovers above the atmosphere of the Earth in the form of an ethereal mesh of pure thinking, communication, and ideas. It is media incarnate, materialized as a bubble, a sort of virtual twin of the Earth overlaid upon it. The world-in-itself describes precisely this self-determinant view of the planet. It implies an evolution of the planet to extend itself into a state of higher being. This idea has been reflected throughout history. In the 1970’s James Lovelock and Lynn Margulis’s wrote about Gaia, and framed the Earth as self-governing biological superorganism, and in the 2010’s with Benjamin Bratton’s notion of The Stack, which uses the metaphor of a software stack to describe the interacting layers of global human activity.³⁰ All of these concepts deal with a layering system in which the planet grows and emanates more complex renditions of life the more layers that are added. In the case of Teilhard this inevitable advancement follows a natural progression from a biosphere to a technosphere then finally to the Noosphere.

While worlding is already occurring at a macro scale, the ability to guide the microdirection, like that of guiding growing roots to form a bridge, is still within the realm of control. If we consider each of these scenarios as points laid out in a triangle. the space between them collectively forms the possibility space for worlding. Reality is likely to fall somewhere in the middle of these three points. To that end, I don’t wish to do a crystal ball reading of which pole holds the future, but simply propose the possibility space as the realm of possible mythologies of the New Acheiropoieta. But, my hunch tells me that a reality of a world made without hands will contain aspects of all of these. Because within each are relationships between people and media that we can already see and will likely persist to varying degrees: human owns machines, human subject to machine, human companion to machine, human revere machine, human becomes machine, and everything in between. Worlds that are of purity in any extreme would be abhorrently bad. A virtual landscape that is but a hall of mirrors facing back at ourselves will drive us mad, a desert of the real in which we become servants to media machines will ruin the species, and an ethereal layer in which bodies become burdens and consciousness must be uploaded will kill the human spirit. The tremendous task now is how to figure out a way of being in this world so that we can have the wherewithal to know how to guide, tenderly, the way the roots grow.

VI. Being in the Worlding

“The point is that you know this is an illusion. So I hope [virtual reality headsets] never get small and convenient. I think they should be ugly and a little awkward.” — Jaron Lanier³¹

Understanding the mode of being in worlds made without hands can confer in us an ability to guide it. Being in these worlds is one of virtuality—that is, one of simultaneous destruction and recombination. So, how can we be, in the existential sense, in a world made by seeing machines? Making art can be a way of taming virtuality and using it as a means to guide worlds. Making art can be a way to bring worlding beyond its seams and see within it beauty for all its seamfullness. Worlding itself not only is a way to describe what seeing machines are doing to make images. Worlding is also a method for us collectively to wrangle, describe, and enhance our virtuality.

Worlding has always been about how people exist together. After all the term “worlding” has earlier definitions than my usage of it. Worlding likely originates from Martin Heidegger who used term “being in the world”, often translated as “worlding”, to describe the existential mode being in the world. Heidegger was concerned with breaking down the duality of being. He did not like the discernment of the subjective and objective, between mind and body. He wanted to break the duality down into the much more essential phenomenological experience of the present. For him this was possible by engaging with the world, using tools, breaking tools, which yielded the essence of being: experience. Since Heidegger other thinkers have slowly taken the frame around individual existence of worlding and scaled it up to become wider and wider. Donna Haraway for instance takes worlding to mean the relationship and conflict with other beings in the world. In another use of it Gayatri Spivak takes worlding to mean the forced imposition of a lens of the world onto a subaltern. Many others have taken the term and slightly modified it for different moments in history. The term is adaptable, but the trajectory appears constant: worlding as a term has evolved to be less about the individual and more about a grand system of being with others.

Continuing this trend in its most contemporary manifestation is artist Ian Cheng. His conception of “worlding” is understood from a systemic perspective, that is, from the perspective of an artist making a media universe. In Cheng’s definition, “worlding is the artistic activity of an individual artist conceiving, incubating, triggering, and nurturing a World towards aliveness.”³² Cheng is a video game artist and uses video game engines to create simulative stories that run in-real time for the viewer. No two simulations are alike. Cheng’s description of worlding is one of maintenance and constant adjustment. He continues, “worlding is firstly the act of creating a life, then secondly letting that life live itself.” Worlding is the act of imbuing life. The thing which has been worlded needs to have a way to perpetuate itself. Nature and Life are perhaps the ur-example of worlding, as they both have a tendency to grow and persist, even against the odds.

Cheng’s definition of worlding describes virtuality perfectly. Virtuality is not the feeling of being an avatar, standing still in an open cybernetic field of digital grass and trees. Virtuality is this persistence of self—it’s the stubborn reconfiguration in the face of destruction. It is about the cycles of maintenance. Virtuality is about remaking, remixing, and becoming oneself ”towards aliveness”, as Cheng writes. Virtuality is less a state than a direction. Virtuality is a way of seeing images as not a state of reality, a portion of captured spacetime tucked into a small glass vial, but as a direction of a data world. When I see images I don’t see pixels, I see vectors, rays, and arrows pointing as if a compass toward a new reconfiguration of myself.

In this same way we can approach the question of control and agency in a world of seeing machines. Control can be had in fully automated environments, but to assert it individually one must know that control is a direction. Looking back to the example of painting we can see in this period of image making that was about a state of control. The image was controlled in totality by the painter. What the painter did with their hands, what gestures and strokes they conducted, directly effected as a 1:1 relationship how the image would appear. Painters are a state dictator of sorts. As noted, the trajectory of image making has become less about the hand, and more about the handlessness. This I think can be a somber and at least sobering fact for many people. Photography and coding have extended the hand further away from our bodies. But the conception that the procession of image making is an unstoppable march destined for the Noosphere, the Singularity, or the Metaverse is not the point. The point is not that the hand should detach from the body. The point is that this procession exists and that within it there are gaps. As Clement Valla showed in Postcards from Google Earth there are moments of reverie to be had in the ruptures and tears in the otherwise monolithic fabric of data worlds. In the ruptures it is possible to find this new way of seeing, which can be refreshing, as it peels back the surface tension of magic and mythology to reveal plain, nonmagical legibility.

Within Muybridge’s Zoopraxiscopes, and how early on he would paint the images as a proxy for photographs, there is also lesson about finding agency in the automated. Retrieving older methods of making images and mixing them with the very new yields a heightened appreciation for both mediums and allows for an emergent direction of control. Painting on photographs can be a cathartic realization of the abundance of seams in the world. Grabbing the hand to put back into handlessness might be the point of handlessness after all. The combination of the two is 1 + 1 = 3; the handed images and the handless images are companions and hold each other, interlocking fingers.

In my work I have found painting on images made through worlding has made me more comfortable with virtuality. Making art is how to world, as Cheng notes. It is also possible to practice worlding within a larger system of worlding. Even by collecting and organizing your own databases, then pruning, tending, and gardening your patch of data, fenced off from the wilderness of the rest of the data world I have discovered it is possible to find the same sense I get from painting. In my series Centaurs I collect my own garden of images and run them through seeing machines—uploading them to cloud services, or running them through a local program—to conjure from thousands of images a high resolution 3D model. Onto this model I paint over with a virtual paintbrush, in 3D dimensions, part of a virtual reality kit. I use my hand to follow the contours of the photogrammetry model and trace it, in part accurately in part not. What I come away with is a confluence, a merged biomorphic entity, that combines the hand and the handless. Working in companionship with the seeing machine I’m capable of creating a dramatic confluence, and my own world, of my paint and a machine’s image. Using my hand to make in a world made without hands is the sublime—it’s the assertion of “I”.

A world made without hands is not a technologically determined world. Seeing machines are around us, and the way to speak their language, and guide them like the roots on a living bridge, is to play with them.