Interview Series: Computational Media Theory and Digital Transformation | Interviewer: Gökhan Çolak, Semay Buket Şahin, Tuğba Bahar
Theoretical Framework
Every interface embeds assumptions about what communication is and how it should work. The social media feed — a single column of content, ordered by publication dates or algorithmic relevance (this became standard after 2016) — is not a neutral communication channel which has no effect on the message. It encourages particular types of content and publishing strategies. For example, Instagram currently privileges attention-grabbing and immediately understandable images, carousel post with a number of slides, and short videos. Of course, people also post in different formats and styles, and some become influential — but a big part of the millions of people who post regularly follow the format and strategies encouraged by the interface.
Can we call this ideology? No. Traditionally, ideologies were created intentionally, by identifiable actors with political agendas. Interfaces are designed by engineers and product teams who optimize them for ease of use, engagement metrics, and other practical criteria. This is why we should think of interfaces as ideology. It is less a question of false consciousness than of a design monoculture: billions of people communicating through a handful of interface templates that share a number of core principles, and use the same interaction “language.”
Platform Culture and Continuity
Both, depending on where you look. At the level of media forms, there is strong continuity. The social media feed is still a data stream — the same logic that governs cinema, television, and even the telegraph. Posts replace shots, algorithms replace editors, but the underlying structure is familiar. (In 2011 I published an essay called *Data stream, database, timeline: New forms of social media* which discussed a social media feed as a new media form. This essay continues my analysis of forms of new media such as a database and a navigable space that I analyzed in the late 1990s.
But scale changes everything. Here it is appropriate to recall Hegel’s famous idea: **quantitative change eventually produces qualitative change**. Quantity turns into quality. When 1.5 billion people are producing social media, and 5 billions are consuming, the cultural ecology transforms. What gets amplified, how collective memory forms, how trends emerge and die within days — none of this works the way it did when a small number of TV broadcasters spoke to a large passive audience. That is a qualitative break, even if the interface looks like a slightly updated newspaper.
Cultural Analytics and Methodology
Both approaches offer unique advantages — and both have limitations. The qualitative approach involves close, careful reading that can produce nuanced interpretations which situate each work in its historical and cultural context. But you can only do this for a limited number of cultural objects. So if we only use this approach, the long tail of culture — the vast majority of what gets made in any period or region — remains invisible. As Franco Moretti expresed this in relation to literature scholarship and literary canons, “the majority of books disappear forever.” (Moretti, The Slaughterhouse of Literature,” 2000.)
Quantitative and data visualization methods combined with large cultural datasets help us to analyze this long tail. The key idea is to consider not only a small number of masterpieces or other works that are important according to some criteria, but *to consider ideally everything created by everybody in a given medium, period, and place.* This for me is the key idea of what I called Cultural Analytics (2005-). Note that my idea is different from the more normal quantitative studies (i.e., Digital Humanities, abbreviated as DH). In DH, researchers are using quantitative techniques, which include statistics, machine learning, and AI, but they never try to analyze everything. The most ambitious such study is *Quantitative Analysis of Culture Using Millions of Digitized Books* (2011) which used a corpus of over 5 million books digitized by Google Books – but it only books as opposed to also cultural artifacts in other media. (5 million books is about 5% of all books ever published.)
Why do I think that ideally we want to include and analyze everything together? In my view, such an approach would produce better understanding and interpretation of any particular theme or cultural technique. We will be able to take into account all other uses of this theme or techniques – and also place it in the full landscape of cultural history. This approach can be seen as the extension of Erwin Panofsky’s iconology (identifying cultural symbols across art history, this was introduced by Panofsky in 1932 and codified in 1939) and Warburg’s *Bilderatlas Mnemosyne* (placing hundreds of artworks’ photos together in order to study interactions between common visual motifs; 1924-1929).
Can we expect that qualitative and quantitative methods in media studies and the humanities can be successfully used together? Recall what happened in the history of social sciences. They separated into distinct qualitative and quantitative approaches in 1920s-1930s, and the separation and suspicion of each other continues today.
And what about the most influential social scientists of the last 50 years? Pierre Bourdieu was one of the few who managed to combine rich theoretical contributions with quantitative methods, and also data visualization. In contrast, Bruno Latour relied on ethnographic methods fused with deeply theoretical ideas, and deliberately stayedaway from quantitative data collection.
Big Data and Cultural Hierarchies
Any method can reproduce hierarchies if applied uncritically — and “big data” is no exception. If you only analyze what has already been deemed worth digitizing, you end up amplifying the canon rather than challenging it. Such dataset reflects existing decisions about what a cultural institution (and society at large) considers worth collecting or preserving.
In contrast to quantitative methods in humanities, Cultural Analytics has a potential to surface previously overlooked and marginalized work – recovering for us “the majority” of cultural artifacts that normally “disappear forever” (Moretti) from our cultural memories and scholarly studies. This becomes possible, because it can operate at a scale where the long tail – or at least parts of it – becomes visible.
In relation to history of images, there is a difference between two disciplines: the much older art history and the newer visual culture studies that emerged only in the late 1990s. I didn’t want to do art history – and when the first-ever visual culture program was established, I entered it, earning PhD in visual culture in 1993. From the beginning, my first publications (1991–) focused on digital graphics and animation as new forms of media, as opposed to historical or contemporary art. After 2007, when our Cultural Analytics Lab was established ([lab.culturalanalytics.info](https://lab.culturalanalytics.info/)), I had an opportunity to finally start analyzing visual culture at scale. In the next few years, we created and analyzed 43 different datasets covering as many possible types of imagery and video across visual culture, as we could. They included covers of magazines over decades of their publication, 1 million pages from manga books, music videos, and tens of thousands of screenshots from a long video game play.
A few weeks ago, with the help of an AI assistant, I was able to develop and release a new version of the main tool we were using in our lab since 2009 — ImagePlot. So now I’m thinking about new large visual culture datasets which I can analyze and visualize using both this tool — and other tools I can now quickly make using AI for planning and coding such tools.
Interviewer: Tuğba Bahar
Speed-Driven Aesthetics and Platform Culture
I would resist the label “algorithmic superficiality.” Lo-fi is not the absence of aesthetic thinking — it is a deliberate aesthetic choice, with a long history. It connects to ambient music, minimalism in architecture and design, and other art movements that valued reduction over accumulation. Minimalism has been perhaps the most dominant global styles in design and architecture since the 1990s, and lo-fi aesthetics can be seen as it’s another manifestation — long before TikTok or scroll culture. The easy interpretation is that minimalism is a response to our information-saturated and busy lives. But perhaps that explanation is too convenient.
Minimalism may simply be what a genuinely *global* aesthetic looks like — one that travels across platforms, languages, and cultures precisely because it requires so little to produce and is easy to receive. Historically, it’s not hard to see the connection between minimalism and the broad *Modern Movement* in architecture (1920s-1930s), and the later *International Style* (dominant in 1950s-1960s). (Modern Movement was focused on smaller buildings – typically just a few floors. International Style shifted focus to skyscrapers filling major city centers. While both share the same overall aesthetic principles, the change in scale and construction technology also created visual differences.
The current minimalism is more close to the earlier Modern Movement than the later International Style. Similar to the Modern Movement, minimalism today often uses tactile and natural materials with rich textures. Another similarity is how spaces incorporate and exploit effects of natural lights and shadows. So in reality, good minimalism is not at all minimal – although we find plenty of simplistic and formulaic examples.
Historically, we already saw two reactions to dominance of first International Style and later minimalism. The first started in the 1960s, when some architects and designers started to use ornaments, saturated neon-like colors, layers of textures, and playful geometric motifs. The second reaction, named “maximalism,” begins in the 2010s, and it adds more dense combinations of various details and clashing patterns.
Since generative AI models are trained on billions of images from the web, in principle they absorb all styles, as long as they are represented sufficiently well. So can we expect that this new image, video and 3D generation technology lead to a aesthetics – a new global visual style or language, the way it already happened in architecture with the shift to new constriction materials and technology?
This remains to be seen. In architecture, cinema, music, or fashion histories , the new materials and technologies had a big effect on emerging aesthetics. For example, to take cinema, Technicolor changes the visual aesthetics of cinema;lightweight 16mm cameras enabled the handheld look of the 1960s New Waves films; CGI led to the hyper-real worlds in sci-fi cinema. In fashion, synthetic nylon revolutionized the silhouette and accessibility of mid-century clothing.” Given such examples, it is logical to assume that we will see new aesthetics because of new materials and methods of manufacturing, construction, and interaction (new UIs) – rather than the current use of AI to mostly make creating in already existing aesthetics faster.
AI, Aesthetics, and Creative Agency
Let’s consider carefully these two categories: *creative subject* and *interface operator.* Is there a clear boundary between them? All creative work involves selecting among possibilities. A film editor selects among takes, a curator selects among artworks, and a designer selects among layout possibilities. *Selection* has always been part of creation.
While generative AI continues this principle, it adds a very important new dimension. The possibility space is now infinitely large — and it is also *generative.* This means that you are not selecting from a number of variations already made by you or by others. Instead, you are making new objects and their variations by giving AI instructions and feedback. In other words, the *possibility space does not exist beforehand, but is created by you* during your creative process using AI tools. So the process of creating and selecting becomes one. This is very important.
Another key characteristic of possibility space as defined by AI tools and their interfaces is that its structure and its limits are not visible to the user. For example, let’s say you are using an AI image tool. Using your input (a text prompt, reference images, other inputs), AI model will keep making new versions of the same output. (Technically, this works as follows: The model maps your input’s distinct attributes—such as subject, medium, and lighting—to find their intersection within its latent space, and then uses probabilistic sampling to ensure that each generated output is a unique, non-identical variation of that concepts’ overlap.) Often, simply changing one word or even a word’s position in a text prompt can lead to different outputs – thus making visible to us a new direction of the infinite possibility space.
The creative industry professionals — photographers, video makers, graphic and UI designers, architects, fashion designers, composers and others — usually need significant control over the outputs generated through these explorations. While the input such as a prompt can of course already specify lots of details which the output will have, often a professional needs finer control. This desire to combine the openness and the speed of explorations in possibility space with the tight control over results is what drives the evolution of generative AI tools today. This is what we see being gradually added in new versions of the tools, or the new tools developed by companies or individuals.
For example, when the AI image tools first came out (2022), there were no control mechanisms — besides what is written in a text prompt. But already one or two years later, companies added a number of controls, such as being able to add image references to a prompt to control a subject appearance, style, and composition. The same happened in the history of AI video generation tools. Today, many many such tools include a long list of control options which are borrowed from professional cinematography. They include (as of 5/2026) shot type/framing, camera angle, camera movement path, movement speed/intensity, focal behavior (focus/depth of field), motion style (handheld vs locked-off), scene transitions, and so on.
Flow Culture and Memory
All traditional cultures had mechanisms to preserve social memory. It was transmitted via oral forms, images, sculptures, and later texts. The Bible, the Quran, and the Torah are examples of such important texts. As societies keep developing, special institutions and new technologies for preserving social memory have emerged – libraries, archives, cataloging systems, and canons of important works. In short, *remembering always requires special effort, energy, and time* dedicated to this.
As societies kept growing in size and complexity, the amount of generated information and also the numbers of events in our daily life (checking news, going to work, meeting a friend, attending 10 art exhibition openings in one evening) kept growing. New media technologies have become essential to capture, saved, and provide access to these memory captures: from photography to audio recordings and film, and later digital photography, digital video, and 3D capture. And in the beginning of the 21st century, social networks emerged, which became new massive collective memory and behavior archives. (However, they don’t provide free public access to their full historical data.)
Media memory technologies are not passive recorders of independent reality — instead, they often actively generate this reality. Many events in our lives would not happen if we had had no way to record them. For example, lots of people traveling today spend time searching for particularly interesting locations to take photos and video for their social media posts — or follow the guides which already describe these locations. Here the ability to turn an experience into a recorded memory is what “generates” this experience in the first place.
Let’s now turn to the question of preservation — how to archive the constantly growing flow of online content? According to some estimates, about 14 billion images are now shared online daily; on TikTok alone, people upload tens of millions of new videos per day. In 1996, Brewster Kahle founded Internet Archive and began large-scale web crawling that same year. Today this archive contains 1 trillion web pages which includes multiple captures of the same websites as they changed over many years ([archive.org/about](https://archive.org/about/)). During the 2000s, national and institutional web archiving programs emerged, such as the U.S. Library of Congress and national libraries’ .gov and domain crawls. Also in this decade, digital humanities researchers and archives started developing dedicated procedures and tools to preserve websites and social media accounts such as Facebook, Twitter, and YouTube.
One of the important questions that motivates me and in particular my Cultural Analytics projects is how to create appropriate interfaces for these massive archives. Many of our lab projects have been about such new interfaces designed for exploration of large visual collections.
Interviewer: Semay Buket Şahin
Digital Archives and Stability
No archive has ever been truly stable. Papyrus decays, libraries burn, data and media formats become obsolete. The Library of Alexandria was destroyed. Medieval manuscripts were lost. Entire languages have disappeared along with everything written in them.
Digital archives face the same fundamental problem — but at greater speed. File formats become unreadable, storage media deteriorate and can’t be read using new devices, platforms shut down, companies go bankrupt. The question is not stability but maintenance: who is responsible for keeping the archive alive, and who pays for this? These are institutional and political questions, not technical ones.
Convergence of Media Fields
I started making still and moving images with computers in 1984, and begun publishing my articles about digital art and culture 1991. And what I saw in these fields during the 1990s was the same situation that already happened earlier with photography and film. When a new media begins to be used, at first you only have practitioners, so some of them start writing about it. And some become important theorists of a given new media (e.g., Eisenstein). Other practitioners start organizing communities of artists working in the new medium, writing about it, publishing journals, organizing group exhibitions, and so on (e.g., Alfred Stieglitz).
The academic world is always responding to new cultural phenomena moreslower, and this is why it takes a while before academic scholars start writing about each new medium. Film studies as a formal field only emerged in the 1960s. In the 1990s, I could count on one hand the people who were systematically writing about “new media art” (and I was one of them.) But I think in the 21st century, this gap is becoming smaller. The number of academic papers from scholars in humanities and media studies about AI is already massive. However, this gap may always remain because before academics can write about some cultural artifacts, somebody has to create them first.
prcarnetworld.com 
Leave a comment