Scroll a platform’s “For You” page for an hour and a particular fatigue sets in. It’s not just the numbing repetition of formats—the unboxing, the dance, the reaction shot. It’s the slow, creeping realization that the feed isn’t showing you what’s good, or even what’s popular in any broad cultural sense. It’s showing you whatever is most efficiently like the thing you paused on for three seconds yesterday. The algorithm, in its silent, relentless processing, has become a de facto curator, a tastemaker that doesn’t taste. It sorts, it predicts, and in doing so, it quietly reshapes the very definition of visual art for a generation that encounters images primarily through a glass rectangle. This isn’t a story about technology corrupting a pure, pre-digital art world. It’s about how the material conditions of distribution—the compression, the endless scroll, the shadowban—are now the primary forces sculpting the aesthetic object itself.
The Algorithm as a Patron of Predictability
Historically, the shape of art was dictated by the patron. A Medici prince wanted a fresco that flattered his dynasty; the church wanted an altarpiece that instructed the illiterate. Today’s patron is a black-box machine-learning model, and its demands are far more opaque. The algorithm’s currency is engagement, a metric that flattens the complex spectrum of human response—awe, discomfort, melancholy, critical thought—into a binary of “dwell time” and “interaction.” The result is a visual culture optimized for legibility at a glance, for the quick hit of recognition before the thumb flicks upward. A painting that requires slow looking, a photograph that sits in uncomfortable ambiguity, a digital collage that refuses a clear focal point: these are not just ignored, they are actively penalized by a system that cannot categorize them for the next user in the queue.
This isn’t a new dynamic, but its scale and precision are unprecedented. The gatekeepers of the 20th century—critics, curators, dealers—were flawed, biased, and often exclusionary, but their logic was at least nominally discursive. An artist could argue with a critic; a movement could form against a gallery’s aesthetic. The algorithm, by contrast, offers no dialogue. It presents a frictionless, personalized feed that feels like a natural reflection of the world, yet it is a world engineered to maximize a single, non-artistic variable: time on platform. The art that thrives here is not necessarily the most challenging or profound, but the most frictionless. It is art that looks like content.
The Aesthetics of the Scroll
Walk through any contemporary art fair and you’ll see the influence bleeding back into the physical world. The paintings are high-contrast, the sculptures are designed to photograph well from a single frontal angle, the installations feature a “hero piece” that functions as a selfie backdrop. This is not a coincidence. Artists, like all producers in a platform economy, are learning to internalize the demands of the distribution channel. The materiality of the digital image—its compression artifacts, its color gamut optimized for OLED screens, its square or vertical crop—is no longer just a container for the artwork. It is a collaborator, and often a domineering one.
Consider the rise of the ultra-saturated, digitally native aesthetic that dominates platforms like Instagram. Colors are pushed to the edge of the screen’s capability because a muted palette reads as a dull post. Compositions are centered and symmetrical because the algorithm’s object recognition models, trained on commercial photography, are more likely to classify and recommend a “well-composed” image. The very tools of creation, from the smartphone camera’s default HDR processing to the one-tap filters in editing apps, are built to optimize for this pipeline. The artist’s hand, once a guarantee of a certain material authenticity, is now often indistinguishable from the computational smoothing of a beauty filter applied to a still life.
The Glitch as a Site of Resistance
If the algorithm demands smoothness, then friction becomes a political and aesthetic stance. A small but significant strand of contemporary digital art deliberately courts the glitch, the artifact, the compression error. This isn’t the nostalgic vaporwave of the last decade, which romanticized the visual noise of dead formats. This is a harder-edged practice that weaponizes the platform’s own materiality against it. Artists are creating images that break the auto-enhancement tools, videos that confuse the content-ID system, and performances that exploit the unpredictable latency of a live stream. The work is not just about the final image, but about the labor of the machine in failing to process it. The glitch becomes a record of a struggle between the artist’s intent and the platform’s smoothing logic, a visible scar of the computational process.
This is a form of critique that doesn’t require a wall text. It’s embedded in the file itself. When an image is repeatedly screenshotted, re-uploaded, and compressed, the accumulation of artifacts tells a story of its circulation. The “material” of the work is no longer just paint or pixels, but the data trail of its own distribution. A photograph that has traveled from a group chat to a Twitter feed to a Pinterest board carries the scars of that journey: the JPEG blocks, the cropped edges, the color shifts. These are not flaws to be cleaned up; they are the patina of the network, a record of the image’s social life. The algorithm, in its attempt to standardize and optimize, inadvertently creates a new kind of materiality, one that artists are beginning to treat as a primary medium.
The Labor of Visibility
Behind every post that “makes it” is a hidden architecture of labor. The myth of the algorithm as a neutral meritocracy obscures the sheer amount of work required to be seen. Artists are now expected to be their own publicists, social media managers, and content strategists. They must master the rhythm of posting times, the syntax of hashtags, the visual language of the thumbnail. This is not a secondary activity that happens after the “real” work of the studio; for many, it is the primary work. The studio becomes a content farm, and the artwork is just one asset in a stream of stories, reels, and carousel posts. The pressure to maintain “visibility” forces a pace of production that is fundamentally at odds with the slow, uncertain process of making meaningful art. The feed demands novelty, but it rewards familiarity. The result is a churn of minor variations on a proven theme, a visual stutter that the algorithm reads as consistency and the human eye reads as exhaustion.
This labor is also deeply gendered and racialized. The expectation of constant, emotionally transparent self-presentation falls disproportionately on artists from marginalized communities, who are often required to perform their identity as a form of “authenticity” to gain traction. The algorithm, trained on biased data, amplifies these dynamics. A Black artist’s work about trauma may be boosted because it fits a recognizable pattern of engagement, while their abstract geometric paintings are ignored. The platform doesn’t see art; it sees content categories, and it sorts them according to historical patterns of attention that are anything but neutral. The result is a soft but pervasive pressure to make work that fits the identity slot the machine has assigned.
The Platform as a Medium
To think clearly about this, we need to stop treating social media as a neutral distribution channel for art. It is a medium with its own material properties, and those properties shape the work as surely as the grain of a woodblock or the chemistry of a photographic emulsion. The infinite scroll is a format; the algorithmic feed is a format; the 1:1 square crop is a format. These are not transparent windows. They are constraints, and like any constraint, they can be used generatively or they can simply deform the work to fit. The most interesting artists working today are not those who ignore the algorithm, nor those who pander to it, but those who treat it as a material to be sculpted, hacked, and interrogated.
This means understanding the platform’s logic at a granular level. Why does a video with a human face in the first frame perform better? Because the object recognition model is trained on faces, and the recommendation system prioritizes content with high-confidence object matches. Why do text-heavy images often get suppressed? Because the system is designed to penalize “meme” formats in favor of “native” visual content. These are not neutral technical decisions; they are aesthetic judgments embedded in code. When an artist uses these rules to make a piece that is simultaneously a critique and a successful post, they are working in the tradition of the readymade, selecting a pre-existing object—in this case, a behavioral prediction—and re-contextualizing it as art.
The New Gatekeepers
If the algorithm is the new curator, then the content moderator is the new conservator. Their decisions—what is “sensitive content,” what violates “community standards”—are the contemporary equivalent of deciding what hangs on the gallery wall. But unlike a gallery, these decisions are made at an industrial scale, by a workforce often outsourced to countries with lower labor costs, following policy documents that are living, breathing, and frequently contradictory. A classical nude sculpture might be flagged and removed while a hyper-sexualized, fully clothed image is promoted because it drives engagement. The distinction is not based on art-historical context or even a coherent moral framework, but on what the platform’s models predict will keep users scrolling and advertisers comfortable.
This creates a chilling effect that is more subtle than outright censorship. Artists begin to self-censor, avoiding certain subjects or formal strategies not because they are irrelevant to their practice, but because they are algorithmically risky. The fear of being “shadowbanned”—a phenomenon platforms deny exists in the way artists experience it, yet is empirically observable through sudden, sustained drops in engagement—leads to a flattening of visual language. The algorithm doesn’t need to remove work; it simply needs to make it invisible. Invisibility is the ultimate form of gatekeeping in a networked world, and it leaves no paper trail.
The Materiality of the Digital Image
We often talk about digital images as if they are immaterial, weightless, infinitely reproducible. This is a convenient fiction. Every digital image has a material substrate: the server farms that store it, the undersea cables that transmit it, the rare earth minerals in the screen that displays it, and the electrical grid that powers it all. The resolution, color profile, and file format are not abstract properties; they are material constraints that shape what we see. A heavily compressed JPEG is not just a “lower quality” version of a TIFF; it is a different image, with its own characteristic artifacts and a specific visual texture born of the discrete cosine transform. The algorithm’s preference for certain file sizes and aspect ratios is, in effect, a preference for certain material conditions of the image.
This materiality extends to the labor of viewing. The infinite scroll is designed to minimize decision-making, to keep the thumb moving. It is a format that discourages contemplation. The art that thrives in this environment is often the art that can be consumed in the time it takes for the next piece of content to load. This is not a judgment on the quality of that art, but an observation about the conditions of its reception. A profound, slow-moving video work that requires a viewer’s sustained attention is not just poorly suited to the platform; it is actively hostile to the platform’s business model. The platform, in turn, is hostile to it.
FAQ: The Algorithm and Artistic Value
Does the algorithm actually change what art looks like, or just what becomes popular?
It does both, and the distinction is blurrier than it seems. The algorithm certainly acts as a filter, promoting some visual styles over others. But because artists are responsive to the system that distributes their work, the filter becomes a feedback loop. Artists begin to produce work that is more likely to pass the filter, and the aesthetic of the platform seeps into the aesthetic of the studio. The “look” of Instagrammable art—high contrast, centered composition, a single clear subject—is now visible in galleries and art schools, even in work that was never intended for the platform. The algorithm doesn’t just select; it breeds.
Is the algorithm’s influence any different from the old gallery system’s influence?
Yes, in two key ways: scale and opacity. A gallery owner’s taste might be narrow, but it is at least a single, identifiable taste. An artist could choose to seek out a different gallerist, or to work outside the gallery system entirely. The algorithm is a monolithic, planetary-scale system that governs visibility for a significant portion of the art-viewing public. Its “taste” is not a coherent aesthetic position but a statistical model optimized for engagement, and its operations are a trade secret. You cannot argue with it, and you cannot easily opt out of it if you want your work to be seen. The power asymmetry is total.
What does “materiality of the digital image” actually mean in practice?
It means paying attention to the physical and computational infrastructure that makes an image visible. A JPEG viewed on a phone is not the same object as a TIFF on a calibrated monitor. The compression algorithm, the screen’s color gamut, the platform’s resizing script—these are all part of the work’s material condition. For artists, this means considering file formats, export settings, and compression artifacts as aesthetic choices, not just technical afterthoughts. For viewers and critics, it means analyzing the image as it actually appears in its native habitat, not as an idealized abstraction. The “original” file on the artist’s hard drive is a ghost; the image that circulates is the real work.
How can artists resist the algorithm’s flattening effect?
Resistance doesn’t have to mean quitting social media. It can mean treating the platform as a hostile collaborator. Some artists deliberately create work that is difficult for the algorithm to parse—abstract, text-heavy, or temporally demanding—and accept the lower engagement as a cost of doing business. Others use the platform’s tools against it, creating work that critiques the very interface it’s viewed on. The most effective strategy is often to build a practice that is rooted in a community, not in a feed. When the primary audience is a group of peers who encounter the work through direct conversation, not algorithmic recommendation, the pressure to optimize for the machine is reduced. The algorithm hates a closed loop.
The Next Step for This Conversation
This is not a lament for a lost golden age. The pre-digital art world was hardly a utopia of pure aesthetic judgment. But the shift from human gatekeepers to algorithmic ones is a profound change in the material conditions of art-making, and it demands a critical language that is as comfortable with code as it is with canvas. The questions we need to be asking are not “Is this good art?” but “What are the conditions that allowed this to be seen? What labor was required to make it visible? What material forces shaped its final form?” These are the questions that will animate this column going forward. We’ll be looking at the specific tools—the compression algorithms, the content-ID systems, the recommendation models—and treating them as the aesthetic agents they are. Because the algorithm is not just a curator. It’s a collaborator, and it’s time we started critiquing its work.