Giana Festa is a digital producer and writer based in Sydney, Australia. She is a 2025 DNR Bootcamp alumni.
Mark Leckey’s 2009 thesis, “Mark Leckey in a Long Tail World”, is a futurist theory on the revolutionary potential of the internet as a vehicle for artists to meaningfully shape culture. In the years since his lecture, many of the ideas he presented have, in various ways, been realized. The internet has removed many of the barriers of cultural production, allowing for the formation of micro-centers of culture, politics and art (including, arguably, this very blog).
Various types of heretofore undocumented or suppressed information has been distributed and popularized, bringing new ways of understanding the world, providing people with access to functionally life-saving knowledge and shaping the political desires of a whole generation. But in the 17 years since Leckey’s lecture, platform capture and monopolization, big data surveillance, reactionary lawmaking and weak regulation have decisively foiled a total realization of the revolutionary potential of p2p and the open web 2.0. In 2026, we stand at a critical juncture for Leckey’s vision of the web, as the long tail faces a new, corrupting, and potentially overwhelming enemy in the form of generative AI.
I first watched Mark Leckey’s lecture “Mark Leckey in a Long Tail World” in October this year, perched in front of my laptop at my kitchen table. Online there exists a recording of his performance at the Institute of Contemporary Arts in London in 2009, split into seven parts and uploaded to YouTube. In it, Leckey sets out his conceptualization and personal experience with “The Long Tail of the Internet”. He goes beyond the conventions of a regular lecture, making use of props, stage lighting, music, and vocal effects. I watched and listened as his saturated, 240p form roamed around the stage, his excited British voice beaming from his headset mic to the auditorium sound system to the video camera microphone to my laptop speakers 17 years into the future. As the lecture unfolded, I begun to feel agitated. By the time it ended, I was paranoid. Watching alone in my kitchen in the middle of the day it felt like, through Leckey, the internet itself was talking to me. Like this extremely niche video with 750 views, linked to me via a Discord server in the backwaters of the internet where it had dramatically shaped the life of approximately 20 post-internet art pol losers, was foretelling it’s own arrival to my screen.'
The idea of the long tail itself is not new; the use of the term in statistics dates as far back as the 1950s, and journalist Chris Anderson wrote a whole book on the concept as applied to the online media market in 2004. But Leckey’s specific take as it pertains to desire, art, and the internet, as well as his very deliberate presentation of that take, is pretty original. He starts by laying out how new technology has changed the nature of cultural production and transmission. In the 20th century we had broadcasting, a distribution system of one to many, like radio or television. Audience demand took the form of a Pareto curve, with producers making media for the 80% majority. However, in the 21st century, as media became digital, the cost of production (and reproduction) became so marginal it was materially insignificant. Consequentially, Leckey argues, we saw a huge upsurge in the volume of mass media, specifically content catering to niche audiences. This is what makes up the long tail; an unlimited choice of perpetual and imperishable content catering to an ever-subdividing market of individual interests. As time passes the tail can only grow longer and longer, the web’s functionally infinite storage space laying out the contents of your timeline to fossilise behind you as you scroll. Time is converted into an arrangement of space along the length of the tail and thus the online world becomes an everywhen, a sedimented accumulation of everything that has ever been uploaded since the web was first made public in 1993. This idea is a compelling one, particularly in the way that it makes a sensible topography out of the seemingly endless data stream of the internet.
Having (literally) roughly chalked out what the long tail is, Leckey then examines it’s potential implications. Back in 2009 his outlook for the future of the long tail phenomenon is hopeful and radically humanist. He sees the ossified cultural production along the tail as an endless list of options from which we can choose from for realising our deepest libidinal desires. What’s more, if we understand the internet in its ideal form as a decentralised network of users, each datapoint along the length of the tail represents not only a piece of content but also a person, or group of people, who made that content. Because each user is interacting with each other in real time, we can create a peer-to-peer feedback loop of desire and production; online content is created to realise the needs, beliefs, and desires of the interconnected group of peers who make it.
At this point Leckey steps things up a notch. He tells the crowd that, through the long tail machine, we can arrive at the future of desire. Smoke is flooding the stage, IDM is pumping through the speakers. Leckey speaks in foreboading monotone about how all of humanity is going to be united in ecstatic cybernetic equilibrium, a sort of Human Instrumentality Project-like singularity. The long tail will bend into a ouroboros and we “become one with the swarm”. This long tail machine would have a feedback loop so immediate and effectively self-regulating that it becomes almost prescient; our desires are informed by our content / our content is informed by our desires. The long tail know the contents of our dreams before we dream them.
“It seems that I can actually just wish things into existence, especially pictures. Whatever image I imagine, or hope for, seems to be already there (on the internet), waiting.”
– Mark Leckey, The Long Tail, 2009
It was this line in particular that really threw me, because it reminded me immediately of a contemporary advertisement I had read just the other day:
In 2009, Leckey calls his activated cybernetic long tail machine an “an anagram engine” that collates all of human culture and knowledge into it’s system. Such a description is not far from how we might understand large language models in 2026. Fed with monumental amounts of data non-consensually scraped from almost the entire internet, generative AI is a vertical integration of the long tail, instantaneously parsable via nothing more than a prompt typed into a textbox. LLMs are a crude perversion of the long tail manifest, bottom-trawling the repository of humanity’s collective consciousness, consuming a vast amount of data that no single human mind could hold, and in return offering up synthetic, reconstituted, profoundly disturbing franken-content.
When viewed through this lens, we gain a new insight on the potential impacts of generative AI on online culture and cultural transmission. The first and most immediate consequence of the advent of LLMs has been the pollution of the internet. The Facebook algorithm has been shovelling AI images into people’s feeds for years, encouraging users to waste their days interacting with fake artist babies and shrimp Jesus. OpenAI has made a TikTok clone with a timeline solely made up of AI generated videos, and it has 4.5 million daily users. In January last year, AI-generated articles officially overtook human-made ones. AI slop is everywhere. At the same time, Google Search AI summaries are siphoning viewership and consequentially ad revenue from individual sites, reducing the economic incentive for people to create original content in the first place.
Using the long tail model, we can predict that as the proportion of real, human-made content continues to rapidly decrease, the feedback loop that sustains the long tail ouroboros will become contaminated. Internet culture will become derivative in the most literal sense, as human posters are replaced by AI models that cannot do anything more than recycle their past output. Everything on your timeline becomes another reconstitution of a reconstitution of a reconstitution of a reconstitution of a viral twitter meme from five year ago.
Simultaneous to the pollution of the long tail will be its individualisation. Instead of creating and consuming online culture in tandem with our peers, our timelines will become hyper-individualised. Peer-to-peer is replaced with data-to-corporation-to-consumer; user profiling and predictive algorithms become so strong they can literally “foresee” (create) your “needs” (drive to consume) before you do. Like when you get an ad on YouTube for a product that claims to meet a need you have, but have never spoken about. Every user will have their own personal micro-ouroboros, their desires and needs mediated and met by a machine.
“Soon you will open Netflix, you open up Instagram, and it’s gonna be content that only you see. It’s gonna be that specific.”
– Connor O’Malley, Stand Up Solutions, 2024
This new feedback cycle leaves room for unilateral prompt-injection by the owners of the algorithms. Leckey’s engine was supposed to include humanity: his long tail machine was decentralised, dispersed across the desktops and minds of billions of internet users and thus mediated through the mechanics of real, profoundly empathetic and inherently mass democratic interactions between people. Generative AI cedes the collective power of the long tail in guiding culture to the owners of the algorithms. It’s common knowledge that algorithms, including LLMs, are instilled with various beliefs and bias explicitly or implicitly predetermined by the sociopaths in Silicon Valley. If you peel back much of contemporary discourse/ marketing/ propaganda about AI, you can find at it’s core a normative, technocratic belief in the importance and benefits of ‘efficiency’, which is itself effectively a codeword for ever-greater margins of capitalist value extraction.
“A lot of people saw that Peter Thiel interview and they were like, oh that’s funny that a guy thinks like that. You have no fucking clue how bad it is. You have no idea. You have no idea how the most powerful people, who are in control of the tech, don’t give a fuck about you. They hate you. They hate your flesh, they hate your bone, they hate your brain. Every part of you that is inefficient they want to replace with efficiency. Guess what’s left when you do that? Nothing. You are what is inefficient.”
– Jreg, Expanding The Tech Spectrum, 2025
Through these dual processes of pollution and individualisation, eventually the p2p network of human interaction holding the long tail together will dissolve. Each of us will become lost and isolated netizens, surrounded by ever-multiplying phantoms in our own subconscious-scraped, prompt-engineered slopidises where all “new” culture we consume is nothing more than the personalised bastardisation of content past. The future online is a special shade of grey picked just for you.
It’s ironic really: the very fact that I even watched Leckey’s lecture in the first place is a sign that, at least right now, the long tail persists. The fact that I was inspired to write this article, and you are reading it on the Do Not Research Substack, proves peer-2-peer discussion of stupid niche bullshit isn’t dead yet. Hell, maybe reading this will inspire something in you, and then you will create a wonderful piece of art that inspires others - a perfect instance of Leckey’s long tail machine still in action.








Your essay immediately made me think of something Jak Ritger touches on in “The Great Lock-In of 2026,” in the new PPP. He footnotes Aatish Bhatia’s 2024 piece on model collapse, which is what really sent me down this path. Jak invokes the concept as a way of thinking about how AI systems might narrow not only cultural production, but our expectations of what culture can be.
Bhatia’s account of the technical phenomenon feels almost uncannily pertinent to your framing of the Long Tail: as models recursively train on synthetic output, the statistical distribution narrows, and the rare, unusual and surprising material in the tails begins to disappear.
What I really like about your essay is that it gives that technical problem a cultural topology. The Long Tail doesn’t simply become clogged with reconstitutions of reconstitutions; the human heterogeneity that allowed the tail to keep extending in the first place begins to disappear. Reading your piece made the parallel suddenly feel very literal.
🫡
I love Leckey’s essay!