Author: cbcrenshaw

The future is local.

I mean local in several senses of the word. The future will be local, first, in the sense that the things you do there will be somewhere close to you instead of located on a computer somewhere in Atlanta or San Francisco or Dublin. It will also be local in the sense that the majority of things you will make and do there will likely be stored on your own computer, perched on your tabletop, stored on your bookshelf, built on your workbench, cooked in your kitchen, and so on, rather than somewhere else. You will own them. Related to this, the future will be local, finally, in the sense that you will share things there with local people whom you actually know, rather than digital representations of people in chat rooms or on headsets. You will likely post the things you make there on your own website, print them in your own zine, sell them in your own community. The internet is not dead, but its role as the primary force shaping our lives is coming to an end.

When I say “the internet,” I don’t mean the technical stack. I’m not referring to the network of networked computers communicating with one another using various protocols. Instead, I refer to the “phenomenological internet” of “the more familiar sites of daily use by billions of people” that Justin E.H. Smith defines in his book, The Internet is Not What You Think It Is. Smith writes,

“Animals are a tiny sliver of life on earth, yet they are preeminently what we mean when we talk about life on earth; social media are a tiny sliver of the internet, yet they are what we mean when we speak of the internet, as they are where the life is on the internet.”

To this definition I would add another category, however: the streaming media provider. When we speak of the internet, we also speak of Netflix, Hulu, Amazon, Disney Plus, and so on. These multi-billion dollar corporations draw on the rhetoric of “the internet” to position themselves as scrappy upstarts opposing the staid traditional media providers, such as film studios and television networks. Viewers have largely accepted this position and view these services as outposts of the internet on their television screens.

Prediction is a mug’s game, so think of this as a prescription instead of a prediction. There are several related trends converging over the next several years that are likely to drive people away from the comfy little burrows they’ve carved out of the internet by forking over $5 or $7.99 or $14.99 or a steady stream of personally identifiable data every month. Together, these trends map the contours of serious contradictions between abundance and desire, on the one hand, and humans and machines on the other, which strikes at the heart of the internet as we have understood it since around 2004. The dialectic emerging from these contradictions will drive new user behaviors in the next decade.

The first trend is the grinding ennui which has resulted from the relentless production of entertainment and cultural commodities for consumption on the internet. Reduced in the past several years to a sort of semi-nutritive paste called “content,” art and entertainment are losing their capacity to relieve and enrich us and now increasingly amplify the isolation and pessimism of life online.

A seemingly infinite stream of money dedicated to the production of entertainment on the internet has resulted in an ocean of unremarkable “content” that does little more than hold your attention long enough to satisfy the adware algorithm or build a platform big enough to stage the next bit of content in the franchise and queue up the next marketing event. Outside of their algorithmically contoured bubbles of fandom, there is little difference between Marvel and Star Wars or DC or YouTube creators or Twitch streamers or podcasts. Netflix shows and Amazon Prime shows and Hulu shows and HBO Max shows and Paramount Plus shows and Peacock shows and so on are indistinguishable blips in time, forgotten as quickly as they are consumed. Books scroll by on Kindle screens or drop serially onto shelves. Photographs and artwork slide past on instagram, meriting a second or perhaps a moment’s notice before disappearing into the infinite past. Pop music percolates through TikTok, moves week-by-week downward on officially curated playlists, radiates out into commercials, and then disappears, poof, as rapidly as it came, displaced by the next. Independent music on the internet–even on platforms nominally controlled by the artists, like Bandcamp or SoundCloud–exists in much the same sort of vacuum as it always has. The internet promised an efflorescence of color and creativity. What it gave us instead was a flat, white light that grows dimmer over time as the algorithms which shape it converge on a single point of optimization.

Because the vast majority of the “content” is indistinguishably boring, the second trend is tightly related to the first. Social media is dying. Many platforms, Facebook front and center, are already dead, gliding still on accumulated momentum but inevitably bound to stop. As recently as 2016, we believed that Facebook could change the world. In recent quarters, however, the most viewed content on the behemoth platform has either been a scam or originated somewhere else. The top 5 most-viewed links in the second quarter of this year, according to Facebook, consisted of TikTok, two spam pages, and two news stories from NBC and ABC on the Uvalde School Shooting. TikTok leads the second-place spam page by a huge margin. Facebook is not a healthy business. Ryan Broderick recently summed up the situation with Facebook admirably on his excellent “Garbage Day” Substack. “Facebook, as a product, is over,” Broderick writes. “Meta knows it. Facebook’s creators know it. Possibly even Facebook’s users. But no one has anywhere else to really go.”

People who rely on social media to promote and build businesses are beginning to note a general decline as well. According to a poll detailed in a recent article on “creatives” frustrated with social media, 82% believe that “engagement” has declined since they started using social media. “I’ve given up on Instagram,” one freelance artist noted. “I wasn’t even sure it was making a difference with getting more work. And I seem to be doing okay without it.”

Facebook and Instagram are in rapid decline, but what about TikTok, YouTube, Reddit, Twitter, and others? A third problem, more profound than the others, faces these: there are no more users to gain. Two decades into the social media era, the market is highly segmented. New platforms like TikTok will continue to emerge, but their surge will climb rapidly to a plateau. The decades-long push for growth that fueled platforms like Facebook and Twitter through the 2000s and 2010s dovetailed with the proliferation of smartphones. Now that the smartphone market is saturated, social media companies can no longer look forward to a constantly expanding frontier of new users to sign up.

Relying on content algorithms to retain existing users or coax those back who have already left, platforms accelerate the ennui of optimization. This leaves precious little room for new types of content or new talents to emerge. Still, people will entertain each other. Those who create art will seek approval and criticism. Others will seek out new and exciting art and entertainment to enjoy. When there is no room on social media for to put these groups of people together, they will find each other in new (old) ways: on the street.

You may have recently heard that machines are going to solve the problem of creating new and engaging content for people to consume on the internet. AI models like DALL-E, Stable Diffusion, GPT-3, various Deepfake models for video, and others use the oceans of existing images, text, audio, and video to create new content from scratch. Some of these models, such as Nvidia’s StyleGAN, are capable of producing content indistinguishable from reality. Artists are beginning to win prizes with AI-generated work. AI-generated actors are appearing in media speaking languages they don’t know, wearing bodies decades younger than the ones they inhabit in reality. GPT-3 is a “shockingly good” text generator which prompted the author of a breathless article in this month’s Atlantic to swoon. “Miracles can be perplexing,” Stephen Marche writes in the article, “and artificial intelligence is a very new miracle…. [An] encounter with the superhuman is at hand.”

Some critics of these AI models argue that they will prompt a crisis of misinformation. Deepfakes may convince people that the President of the United States declared war on an adversary, for example, or a deepfake porno video could ruin a young person’s life. These are valid concerns. More overheated critics suggest that AI may one day surpass human intelligence and may, therefore, power over its creators like masters to pets. Setting aside the Social Darwinist overtones of this argument—that ”intelligence,” exemplified by the mastery of texts, translates automatically to power—machine learning algorithms are limited by the same content challenges facing social media. AI may create absorbing new universes of art and sound and video, but it can only generate content based on the existing corpus, and it can only distribute that content on existing networks. People have to create new texts for AI to master. The willingness of a continuous army of new users to generate these texts and upload them to the phenomenological internet of social media and streaming video, where they can be easily aggregated and made accessible to machine learning models using APIs, is declining. The same types of algorithms that prompted Stephen Marche to proclaim a New Miracle in The Atlantic are driving the most successful corporations in history right off a cliff as I write this.

These critiques of AI-generated content assume that people will continue to scroll social media and engage with the things they see there in ways similar to their behavior over the past decade. In this model, to review, users scroll through an endless stream of content. When they see posts that inspire or provoke, impress or irritate, they are encouraged to like, comment, and share these posts with their friends and followers. The content may be endless, but the people on both sides of the transaction are the most important elements in the decision to like, comment, or share. Users are not impressed or provoked by the content itself, but because of the connection it represents with other people. They respond and share this content performatively, acting as a bridge or critic between the people who created the content–and what they represent–and their friends and followers. If you remove enough of the people, all of the content loses its value.

At a more fundamental level, people are the appeal of any creative work. Art without an artist is a bit like clouds or leaves: these may be beautifully or even suggestively arranged, but they offer no insight on what it means to be human. GPT-3 may tell a story, but it does so mimetically, arranging words in a pattern resembling something that should please a human reader. You may level the same criticism at your least-favorite author, but at least they would be insulted. GPT-3 will never feel anything.

AI-generated content will neither solve the content problem for platforms nor prompt a further crisis of misinformation and confusion for users. AI content will be the nail in social media’s coffin.

As a result of these interlocking trends–the crushing ennui of “content,” the decay of social media, the dearth of new smartphone users, and the incompatibility of AI-generated art with human needs–“culture” is likely to depart the algorithmic grooves of the internet, sprout new wings offline, and take flight for new territory. Perhaps, once it is established there, the internet will catch up again. Perhaps then software will try, once again, to eat the world. This time it has failed.