The Internet’s Lonely Urban Design
The Information Superhighway paved over Virtual Villages; can we bring them back?
In the beginning, the internet resembled a quaint village where everybody knew your name. It was largely self-governing and self-policing, with its own norms and customs. Like many insular communities it struggled with, and feared, immigration. Every September a new crop of college freshmen would get their first computers and briefly overrun the village before eventually getting assimilated into its culture.
As computers became more prevalent, Septembers got longer. Finally, in 1993, September never ended. People from all over the world were joining all the time, ceaselessly, bringing their own tastes and expectations to the nascent community. The internet became a diverse city. It developed a seedy underbelly, a hipster neighborhood, contentious politics, content moderators, culture wars.
Some early internet users bemoaned the loss of their nerdy subculture. Others were more optimistic. It seemed likely that the internet would make the world smaller and kick off an era of global cooperation and innovation.
The reality has been much more of a mixed bag. The early utopian dreams of the internet — that it would be a virtual Library of Alexandria, a Melting Pot, a conduit for World Peace — have only sort of come to pass.
There is no technical reason for these shortcomings, the computer network that underlies the internet is fully capable of connecting any of over 5 billion people together. The human network, on the other hand, hasn’t scaled so well.
Instead of subdividing into smaller communities and fiefdoms, the way real cities do, the internet has actually coalesced since its early days. A few sites (Facebook, YouTube, WhatsApp, Instagram, WeChat, TikTok, etc) each serve over a billion users. Smaller communities like forums and blogs have largely been absorbed into platforms like Reddit, Discord, and Medium. In this way the internet has expanded vertically instead of horizontally. As more people arrived it did not sprout new human-scale neighborhoods — instead it became a megalopolis where billions of people spend their lives within a small handful of corporate mega-towers.
Mall•E
Inside, it’s not so bad. Rent is free, our favorite brands are there, and our friends are right next-door! It’s a place tailor-made to maximize a bland sort of human happiness — more Wall•E than Blade Runner.
Yet Wall•E is still a cautionary tale; fascism is still fascism even when its denizens are fat and happy. I don’t mean to ascribe evil motivations to companies like Meta or TikTok (and this might be a good time to disclose that I work for Google, which is hardly any different). It’s just capitalism, baby! Everything it touches inevitably becomes a shopping mall. These places never promised to be free or democratic or progressive so it’s pointless to hold them to such a standard.
The problem isn’t the algorithm that dictates what you see. It’s not about shadow banning or free speech or wokeness. No, the problem is that the virtual world outside of Social Media’s mega-malls is so barren that it’s hardly worth planting one’s own flag in the sand. In a city without public spaces or public services, benign capitalism starts to look a whole lot like fascism.
Creating new spaces on the internet is actually quite easy, at least compared to the real world. A good web developer can single-handedly create an interesting website in a matter of weeks. But if they build it, will anyone come?
The Fediverse
The internet is fundamentally a pull-based network. If you want some piece of content you have to request (“pull”) it from a specific path on a specific server. The internet itself has no mechanism for recommending content, or preemptively sending (“pushing”) you things you might enjoy. Solving that problem has turned out to be extremely lucrative — some of the most successful companies in the world effectively make their money off content recommendation algorithms (eg. Google, Meta, Netflix, TikTok).
The Fediverse is an attempt (certainly not the first, probably not the last) to build a public push-based network. You pick a “node” — a server owned by someone you generally trust — tell it what you like, and content starts arriving in your feed just like any other social network. The difference is that it’s built on an open standard, ActivityPub, that allows both content and users to easily move between different nodes. To continue the dystopian city analogy, the Fediverse builds sky-bridges between corporate mega-towers and smaller buildings.
It’s a great idea, but so far the results have been a bit underwhelming. Even in the wake of Twitter’s death-spiral, Mastodon has barely topped 10 million users. Reddit alternative Lemmy has only 2 million even after Reddit’s tumultuous 2023. And even then, neither of these services have actually achieved much benefit from federation.
Perhaps the most interesting feature of Mastodon is its “local” tab, which shows content only from users connected to the same Mastodon node. I joined the Medium node so I could have my long-form blogs and short-form “toots” under one umbrella. In theory, the local tab should have content from other bloggers with similar interests to mine, or at least a similar hobby.
In practice, it’s basically just a bunch of randos. I say that without judgement — I’m something of a rando myself. It just turns out that a person’s choice of hosting provider doesn’t say much about their personality or interests. My favorite people on Mastodon are spread all across the Fediverse, not localized to any particular node. In this way, Mastodon is effectively just a new mega-tower. Things colloquially happen “on Mastodon”, not on a particular node or network. Federation, for all its merits, is reduced to an implementation detail.
To cut to the chase: the thesis of this post is that the internet needs fewer mega-malls and more cute little neighborhoods — but Mastodon tried exactly that and I haven’t found it compelling at all. So what gives? Maybe I should take a step back to figure out what I want and why.
Small fish in a big pond
Dear reader: You’re 1 in a million! You have so many attributes that make you special. Statistically, I bet there’s only maybe 5000 people on the whole internet like you. 2000 on Facebook, 800 on TikTok, 400 on Twitter… ok this isn’t the motivational speech I thought it’d be.
The internet has done for culture what globalism did for manufacturing: if you want to be successful you have to be the among the best in the world because you’re competing with the best in the world. American electronics manufacturers can’t build iPhones cheaper or better than Chinese companies can, so why should Apple buy from them? In the same way, I’m not as smart or interesting as Cory Doctorow, so why should you follow me? If I want to make it big on the internet I have to be both extremely talented and extremely lucky.
As another example, my hometown of Seattle used to be renown for its music scene. Bands like Nirvana, Pearl Jam, Death Cab for Cutie, and Modest Mouse cut their teeth in local venues before going national. Seattle still has a music scene, of course, but it’s less and less relevant over time. People increasingly discover music on Spotify or TikTok, not local radio stations or grungy dive-bars. The result is fewer opportunities for new bands to hone their craft and earn a living. Without a local group of music consumers to play for, every act must compete on the global market with established high-budget musicians like Taylor Swift and Miley Cyrus.
Indeed, the fact that both of those artists are still extremely relevant more than a decade into their careers indicates the kind of market entrenchment that comes from being a global brand. For 22 of the 52 weeks in 2023 the top song on the Billboard Hot 100 came from an artist who first topped that chart more than a decade before. The same can be said for only 3 weeks of 2013 and 0 weeks of 2003 and 1993. In other words, at least according to this admittedly small dataset, popular artists are staying popular for longer than ever before. Similarly, 7 of the top 10 highest grossing tours of all time happened within the last 10 years (even adjusted for inflation). And yet, a 2015 study found that the median American musician earned just $17,500 per year from their craft.
This is how a globalized world works. The rich get richer; the famous get famous-er. The middle class disappears. A few content creators get massive followings, sponsorship deals, world tours, magazine spreads. The rest of us — who really are quite a talented and handsome bunch — scrounge for the last few “likes” or “claps” or gigs or whatever. We’re not the best in the world so when the best in the world is just a few clicks away, we’re nothing.
Sorry, I didn’t mean that. You’re not nothing. You’re 1 in a million! So am I! We’re people who have the capacity to think, teach, entertain, and delight. We have a lot to offer, we just don’t seem to know how. The internet treats us with indifference, and we take that to heart.
Touch Grass
I know what the best modern social network is. It’s not one I expected to praise when I started writing this post, but the more I thought about it the more it became obvious. It’s Facebook.
Specifically, it’s the way my mom uses Facebook. Her feed is mostly other local moms talking about their day, asking for help, trading recipes, and planning events. Frankly, it’s kind of boring. It’s also genuinely helpful to people. On more than one occasion my mom has brought dinner to a neighbor she barely knows after hearing about some hardship from their Facebook page, and that favor has been returned in kind.
The secret to a healthy human-scale network of friends is, likewise, obvious. I went looking for friends in the “local” tab of my Mastodon node. I should’ve instead gone looking in my actual, physical location.
Exchanging ideas with far-away strangers is a wonderful thing. Yet as much as I love a good discussion I have to admit that exchanging dinner is even more wonderful. One day maybe we’ll all be brains in vats, a hive-mind free from all corporeal needs. Until then we can’t ignore our annoying physical bodies. For now, a good friend is one who can bring food. And all the other stuff too: care, affection, Christmas presents, sex, Mario Kart, etc.
So the internet cannot replace the physical world, but it can be a companion to it — a tool to add richness and context. It can bring people together. Except, it’s not doing that very well right now. America is in the midst of a loneliness epidemic and the internet appears to be more of a cause than a solution. The fediverse doesn’t seem to be helping either — but maybe it’s on the right track.
The Localverse
Facebook can help keep local friendships alive, which is a genuinely important service. That doesn’t excuse its many failures as a company but it does at least hint at what an actually positive social network might look like. One part that’s missing is a way to meet new people. It’s famously hard to make new friends as an adult in America and so far Facebook has done very little to help.
There are plenty of location-based social networks out there, but few focus on actual community-building. Nextdoor is about a neighborhood, not about the neighbors — its content focuses on taxes and utilities and “suspicious vehicles” rather than on the people. Similarly Meetup is about events, Tinder is about dating, and Craigslist is about buying stuff. All of these can be catalysts for community and friendship but that’s always more by happenstance than by design.
So what would a network designed for community-building look like? As is often the case, the answer can be found in Graph Theory. Most social networks have some sort of recommendation algorithm that finds the users creating the best content and helps them get more followers. Rational enough, but it creates a graph structure that I call a “mob” where a few strongly-connected “influencers” are surrounded by a sea of weakly-connected “randos”.
Real friend groups look like “cliques” which are strongly-connected, meaning that each person in a clique has relationships with most of the others in the clique. This is important because, looking at this through a clinical economic lens, you derive value not just from the relationships your friends have with you but also from the relationships they have with each-other. Those relationships are a source of community norms, ideas, and inside jokes.
Humans are only capable of maintaining around 150 stable relationships at a time, so the best kind of social clique is one with fewer than 150 people in it. Ideally far fewer, since most people are part of multiple communities (your neighbors, your coworkers, your college buddies, etc). If all of your friends are in the same clique, you might be in a cult.
A human-scale social networking app therefore needs an algorithm that recommends small groups of friends to connect with, rather than single accounts to follow. This might be akin to an algorithm that recommends Discord Servers or Subreddits, but with more of a focus on local community-building and friendship.
If I were a more driven entrepreneur this would be the part of the post where I link you to my new startup that does just that — but I’m not. And it’s not an easy problem to solve, not in the slightest. How can an algorithm predict real-world friendship? What would content moderation look like? And, how do you ensure that this clique structure doesn’t just re-enforce the algorithmic echo-chamber that’s driving cultural division?
Perhaps algorithms aren’t the right tool for the job at all. Can’t we all just go back to hanging out in local bars where everybody knows your name, like Cheers? Maybe, but algorithms have proven so remarkably effective at capturing our attention that the only option may be to fight fire with fire. TikTok seems to be fundamentally more entertaining than a dive bar, and that’s not going to change. Engineers got us into this mess, so engineers should help get us out.
Where the superhighway ends
Al Gore coined the term “Information Superhighway” in the early 90s as a nod to the then-recently-completed US Interstate Highway System. This was perhaps more apt than he knew. Both systems strengthened the bonds holding the country together but in doing so tore up local neighborhoods and, whether through traffic or through TikTok, increased the amount of time Americans spend alone.
I’m fascinated by Urban Planning. On the surface it’s just a pile of zoning codes and sewer maps, but behind all of that is a dastardly plot to improve the lives of residents by simply tweaking their surroundings. And it works! Well-designed walkable cities have been shown to make people healthier, wealthier, more independent, more social, more upwardly-mobile, and happier overall.
Good urban planning doesn’t always require fancy new tech (ie. gadgetbahn). Some of its best tools are decidedly primitive: concrete bollards for bike lanes, wooden shelters for outdoor dining, flashing lights at crosswalks, literal trees. The trick is to be strategic, to look broadly at the needs of a community and identify simple ways to strengthen it.
I believe it’s possible to do the same with the internet, but only if it’s treated like the public space it is. Only if individuals, non-profits, and governments step up to improve it.
I have a friend who moderates a welcoming local Discord server, complete with a home-grown bot that helps schedule watch parties and board game nights. We should all be more like him!
Of course, urban planning sometimes also benefits from big fancy mega-projects like public transit corridors or flood barriers. Likewise, the internet could benefit from new community-minded discovery algorithms. When facing a problem at the scale of a worldwide loneliness epidemic, no solution is either too big or too small.
