People are right about exactly one thing when it comes to YouTube: the place is a hole. They talk about falling in, starting out with the surface-level cats and music videos, until the recommendation algorithm does its work and they suddenly discover that they’ve turned into a Nazi. They talk about the depths. They talk about a tunnel or a vortex. This is all true, in a very literal sense. Most of the internet is a membrane, and things skim across its flat surface. That’s what it means when something goes viral: it becomes lubricated, slippery; it rolls, fluid and unstoppable, between the corners of the world. But every location is indexed, searchable, on precise co-ordinates and open to a roving gaze. It’s true that some parts of this flat surface are fenced off – locked Twitter accounts, private forums, paywalled websites, academic journals – but there’s always a high degree of entropy at work, wearing down the levees. Someone takes a screenshot. Someone’s account gets hacked. Any sufficient quantity of liquid content will inevitably end up slopping over the walls. But YouTube is something else.
It’s taken the mainstream quite a while to notice exactly what’s been happening on the site. But, to be fair, how were they supposed to find it? Text and images are synchronic: they’re arranged, fixed and static, to be scanned over, harvested, and thrown away. Video is diachronic: its basic unit isn’t meaning, but time. And YouTube contains a lot of time. Five hundred hours are uploaded to the site every minute. An entire human lifespan goes up every day. Thirty thousand years’ worth of video is added every year, which is six times longer than the entirety of written history.
Forty thousand years ago, the first known piece of figurative art was created: a sculpture in mammoth ivory of a man with the head of a lion. We don’t know what the lion-man did, or why it was produced. We don’t know what worlds of meaning its distant creators inhabited. This was the beginning of human cultural history, and in total YouTube archives a span of time fifty times longer than that. If unbroken generations had lived their lives watching every single YouTube video uploaded up until today, the first in the chain would have been an australopithecine, a squatting ape whose only tools were sticks and stones. You’re paddling out over that same chasm every time you watch someone opening boxes or shoutily explaining their political opinions on YouTube. It’s an accretion of masses and masses of impacted time, heaped over itself, condensed down to a single point. It’s a hole torn through the fabric of the universe.
Down in these depths, there are celebrities you’ve never heard of. Millions of children are obsessed with other, perfectly ordinary children, who mostly just answer questions about their favourite colours. Millions of adult men watch other adult men playing videogames. There are nursery rhymes and cartoon skits for toddlers that feature injections, decapitations, and torture. There are flat earthers and anti-vaxxers and people who think ordinary geometrical coincidence is a vast system of Satanic symbolism. There are senseless centuries that seem to have been spun out by algorithms. And what a surprise: just like the dark side of the moon, just like the inaccessible plains of the Antarctic, this black hole is full of Nazis.
I’ve been aware of right-wing YouTube for years now, before it started piling up real-world bodies in the street, before one of its heroes ran a hilariously failed Ukip election bid, before it became something that ordinary parents felt they had to worry about. What always struck me, far more than the antisemitism and the conspiracy theories, was just how lonely it was. A man sits in front of a camera, alone, and talks at rambling length about how the Jews are ruining everything. They’re talking in an empty room. Nobody’s actually there. And unlike film or TV, internet video is almost structurally designed to be viewed alone. You don’t watch it with a friend or a partner, you just share it on other digital platforms, to other people in other empty rooms.
This isolation is there even for the right-wing YouTubers who made it big – but most people never make it big. So many of the channels I saw had viewerships in the low tens, and these people still churn out videos, day after day, hour after hour. I found one, a video titled My message to the radical left, which had been viewed exactly once: by me. The orator sways and wobbles and pokes the phone camera up his nostril. You created your enemy, he says, and that was your biggest mistake, because with the anti-discrimination and the affirmative action you never leave us alone. Did this person know that I was the one he’d be talking to? Could he know that his message would, at long last, reach its destination?
YouTube was always going to end up being ruled by the right, because right-wing politics are a politics of loneliness. The helpless, atomised individual, endlessly at war with the world around them and everyone in it, desperate to cling to some imagined national, cultural, racial, or political community, talking to an enemy who isn’t there. The new 14 words: Oh yeah, and by the way, please don’t forget to like, share, and subscribe. If the site’s algorithms seem to be sending people into a tunnel to reaction, it’s because that’s what’s there; if the site has become a fascist playground, it’s because fascism is the sickness of internet video as a medium.
As always, the symptom gets confused with the disease. A lot of people seem to know intuitively that there’s something very badly wrong with the grand system of online video, but that complaint ends up turning into a very limited demand: YouTube just needs to kick the Nazis off its platform, and then it’ll be fine. It will not be fine. The platform itself, the whole complex of platform capitalism, is a machine for making Nazis, and the Nazis are only the most visible of its products.
This is why the mostly well-intentioned attempt to foster a leftist YouTube community is doomed to fail. It feels (with one or two honourable exceptions) deeply awkward, and not just because of the gangliness of the people who make these videos, or the way their jokes tend to fall flat. It’s the wrong content being shoehorned into the wrong form. Mass participatory politics can’t be fully expressed by one person talking to a camera in an empty room, in the same way that the unknown shamanism that surrounded the lion-man figurine couldn’t be expressed in a Papal bull, and mathematical proofs make poor protest chants. The left that takes shape on YouTube and the various other social media platforms tends to be a gloss over something that remains fundamentally reactionary: bickering and resentment, cringiness and vituperation, a bitter identification with imagined national, cultural, racial, or political communities, a subject at war with the world around them and everything in it. You can make a video or a blog post against neoliberal atomisation if you want, but it’s still in you, baked in to your every word.
It can’t be drowned out and it can’t be switched off. The only way to shut down the fascist creep on YouTube is to shut down YouTube itself.
This is, for obvious reasons, not Google’s preferred approach. For a long time, they simply did nothing, which is at the very least a coherent approach. The line goes that the site is a platform rather than a publisher, anyone can use it, and the firm isn’t responsible for what people throw into the hole it’s given them. And in any case, something like YouTube is impossible to effectively moderate. If you wanted full scrutiny, you’d need 90,000 moderators, watching every single newly uploaded video on eight-hour shifts. (YouTube presently has about 2,000 employees.) Pay them $10 an hour, and that’s $7.9bn a year. The complaints of liberal internet users are not worth $7.9bn a year. But they are worth something. And while the vast tunnel of YouTube can’t be effectively explored by humans, it can be processed by machines.
YouTube already used artificial intelligence to clamp down on nudity and copyright infringement in videos; now, in the wake of some stupid scandal, it’s doing the same for politics. The problems with this approach (even putting aside the inevitable free-speech squabble, or questions over whether we really want to give giant capitalist tech firms the power to determine what is and isn’t politically acceptable) are obvious. Almost as soon as the new anti-Nazi robot was installed, a slew of antifascist videos were taken offline, often for using Nazi imagery such as the swastika. If they’re smart, actual Nazis tend not to brandish swastikas in people’s faces, because the symbol has a well-known off-putting effect. Antifascists, who want people to know exactly what it is they’re fighting against, will deploy the swastika, for precisely the same reason.
Maybe these issues will be ironed out. More likely, people will end up learning what to do and what not to do to avoid being flagged up by the algorithms. Computers tend to have a hard time processing irony, ambiguity, and the use-mention distinction, so these things will vanish from our discourse and our two-million-year archive. (Of course, this was already happening.) We will start to think a little more like machines, sorting everything into clear, cold categories. We will start to speak less and less in language, and more and more in code.
This isn’t new; every technology turns human thought, to some degree, into an extension of itself. No animism without pigment, no Enlightenment materialism without clockwork gears, no fascism without radio. The difference is that previous technologies only left their stamp on relations between human subjects, while digital communications interposes itself entirely. This is why, even if it works perfectly, with no discursive collateral damage, algorithmic moderation is still a nightmare – not because it closes down the flows of speech, but because it creates a concentrated torrent of non-communication. I might have been the only person to view My message to the radical left, but despite the title it wasn’t to me. The intersubjective dimension is draining out of the world; what’s left is empty talk, psychotic mumbling, externalised monologue – not addressed to another human, but pouring itself into the void. All those people in front of cameras in empty rooms: they’re talking to and for the machine. They put their lips to the black hole and speak, and no echo meets them out of its infinite and lonely depths.
PS: The problem is that a lot of what I’ve said about internet video here also applies to writing. Literature is also solitary, composed in silence, read in silence; it’s a fundamentally pathetic and asocial activity. In writing, we also talk to the inscription-machine more than we do to any actual reader: as Derrida argues in his commentary on Lacan’s seminar on The Purloined Letter, a letter never reaches its destination. One property of writing is its capacity to go unread. That said, three points in defence of my practice: writing is not embeddable within a concentrated technical platform; the materials of writing are not (necessarily) a global communications infrastructure but an emergent and mutually agreed-upon system of words; writing is removed from its object, and therefore involves a properly significatory aspect that video – which can only enframe, capture, and replicate – lacks. As such, it’s intersubjective in a way that video can not be, because words are not an exterior technology but the foundational stuff of subjectivity. But maybe we’ll have to get rid of all this as well.