The Slow Unmaking of the Web: From Short-Form Video to AI
“Kids these days” are alien to the optimistic origins of the Web. Exploring it was one of the highlights of my own teenagehood. Some of my fondest memories are of playing Ragnarok Online and serving as an admin for Brasil GraphicX, a now-defunct design community. These adventures sometimes included my “IRL” connections; for one, the other admin for the forum was a colleague from school, so my adventures in Prontera and Payon often involved my brother and our friends.
Like any place worth exploring, it was not without danger. I vividly remember being scammed out of an item worth half a million “zeny” through some strange exploit. I was furious and shouted “Nãooo!” so loudly that I scared my mom, who could not understand how something inside a computer screen had upset me so much. I was also occasionally the culprit. Once, I stole a rare item dropped by a powerful monster I had not killed, an unforgivable sin for the uninitiated. The aggrieved player followed me around for hours afterward, cursing my family line five generations deep.
Even social media was so much fun. The first I experienced was Orkut. Most popular in Brazil and India, it led to a flurry of social activity in the country; I remember my older cousins rekindling friendships from their youth, and seeing which girls from school had peeked at my profile (a feature that would then be used to try to sell LinkedIn Pro subscriptions). When Facebook came to town, more or less alongside the screen picture, I remember being able to schedule days out in the pool club by simply posting on my “wall:” anyone doing something today? It’s so sunny.
In a way, peak Facebook was a social technology like no other. In the glory days, you could browse roommates and flats, find nearby events and check if your friends were attending, and buy pretty much everything. There was also a true sense of it being “the public square” for regular people, and not only for influencers. I still get the occasional “memory” reminder of a discussion I was having in public with my friends around some political topic, about a band, or a book I was being forced to read for high school.
Other people found communities with nobler goals: creating open-source software or building an online encyclopedia. But even in the absence of a community, there was this sense of something being built collectively mostly because it was cool. I remember being awed by extremely complicated Warcraft III custom maps, from DoTA to “Ninja Slide,” and later, in college, by the Linux kernel. This idealistic notion of the Web as something that unites humanity still informs how many of us talk about technology, even as the experience of being online has moved far from it.
And I am not alone in feeling that this aura of naïve optimism around the Web has disappeared. In an excellent essay, Yancey Strickler pinpoints the turning point in the mid-2010s, when “the public and semi-public spaces we created to develop our identities, cultivate communities, and gain in knowledge were overtaken by forces using them to gain power of various kinds (market, political, social, and so on).” Most important, he draws an analogy between what happened to the Web and Liu Cixin’s “Remembrance of Earth’s Past”. In Liu’s universe, intelligent species avoid broadcasting their existence because to be seen is to become a target. Strickler argues that something similar happened online. As the open Web became dominated by platforms, advertisers, mobs, bots, and algorithmic incentives, public visibility became too costly to ordinary users
Another framing is Doctorow’s “enshitification:” that platforms gradually degraded as they turned from serving users to serving business. In his book of the same name, he argues that once users were “locked in” to platforms that genuinely made life easier and more fun, they flipped the switch to become “extraction machines.” This is perhaps the obvious capitalist evolution of a laissez-faire Web. When each individual contribution is small, users are loosely coordinated, and there is no shared institutional project, the actors with the clearest incentives and deepest pockets eventually set the terms.
These two framings are, in my view, complementary as one describes the institutional logic of the vibe shift, while the other describes the subjective experience of the ordinary user. And they culminate in short-form video platforms. Absent social graphs, any notion of a community, and no sense of building something bigger than the parts, all that’s left are wildly entertaining pieces of 5-second video content. Although many may dabble in their making, the typical short-form video people consume will be created by an elite class of influencers whose rise was shaped by the relentless, selective pressure of the algorithm. If, in previous iterations, platforms rewarded “subscriptions” and loyalty to influencers, on TikTok this is stripped away to almost its fullest extent. Sure, people still form parasocial relationships with Web sub-celebrities, but this is incidental to “the feed,” which will very happily replace your taste in a heartbeat.
Generative AI, then, arrives not in Rome at its height, but in a city already sacked by barbarians. As influencers quickly noticed, it represents an opportunity to replace them altogether, if the public is willing to go along. Yet defending the existing influencer economy as “the Web as it was” is a misguided cause. That world was already downstream of the Web’s unmaking. A more ambitious agenda is to reimagine the Web for a world in which many of the technical challenges of creating and maintaining online communities are far simpler than they were in the idealistic phase of Web 1.0. At the same time, this brand new world brings its own challenges, especially the loss of signal. If coherently produced text once implied attention, effort, and some minimal human investment, in the age of ChatGPT it no longer does.
A sane agenda for Web or whatever, thus, shouldn’t be reactionary—because the stable status quo that was reached prior to ChatGPT and Veo kind of sucked. I think people should take strong stances against the use of AI in ways that diminish human connection and erode the digital commons, while acknowledging that these technologies make it easier than ever to build weird, idealistic projects: small communities, personal tools, and other spaces that need not be optimized for the feed. The project should aim to preserve or rebuild the parts of the web worth saving under fundamentally different conditions.
Three specific directions seem promising. First, what I’d call “homesteading,” we should just rely less on platforms altogether. As Drew Dimmery puts it in his blog, platforms reduce complexity by constraining possibility, but especially in the age of vibe coding, we can prompt a Web of innumerable possibilities. Build your own website! Start a weird project! And do it outside of a platform. Second, decentralized platforms, where the incentive structure might preserve some of the early Web 2.0 things I grew up with. There’s no reason why platforms like Bluesky or Mastodon shouldn’t succeed! They reopen the design space, letting everyone chime in on how communities can and should organize, and on what values technological affordances should be built upon. Finally, I’m excited by possibilities to preserve humanity, effort, and intentionality in online interactions. If coherent writing no longer reliably signals attention, we need provenance, community reputation, slower norms, proof of participation, and rituals that distinguish contribution from content.
I’ve been described as “annoyingly optimistic,” and perhaps this essay proves the point. But the optimism I want to defend is the very same optimism that has made the Internet such a magical place. For what it’s worth, it’s a sunny day in Princeton. Anyone doing something today?