The Feed is the Product: Why We Stay in Walled "Gardens", and Why We Might Finally Leave for the Open Social Web

There's a particular kind of exhaustion that only social media can produce. You open an app to check one thing: a message from a friend, a recipe, a news headline. Forty minutes later you're locking your phone with a vague sense of shame, having absorbed nothing you actually wanted to know. Forty minutes lost into the void. This is doomscrolling, and it isn't a personal failing. It's the intended output of systems designed by the most resourced engineering organizations in human history, all pointed at the same target: the next swipe.

We have known this for years. We have known that the major platforms are bad for our attention, our sleep, our politics, our children, and our sense of reality. And yet here we are, still scrolling. This essay is an attempt to answer two questions: why is leaving so hard, and what does it look like when it finally becomes possible?

The state of the mainstream

Let's take the platforms one at a time, because each one has found its own distinctive way to be terrible.

Meta is the empire that learned nothing. In 2018, we learned that Cambridge Analytica had harvested the data of millions of Facebook users to build psychological profiles for political campaigns. Trust in the platform cratered. Mark Zuckerberg apologized in Congress (and became a source of memes). And then... nothing happened. Crickets. People did not leave. The scandal became a footnote, the platform kept growing, and Meta moved on to its next product: an artificial intelligence assistant called Meta AI, is now thoroughly embedded across Facebook, Instagram, WhatsApp, and Messenger. In April 2026, Meta rolled out its most powerful model yet, Muse Spark, across the same surfaces, immediately sparking controversy. The company is now testing AI chatbots that message users first to boost engagement, a system internally called Project Omni. Users, by and large, seem fine with this. There is no equivalent of the Cambridge Analytica disapproval backlash. The bar for outrage has been raised past the point of action. Facebook is full of AI bots, and somehow the deluge of AI generated shrimp Jesus gets a pass. Instagram isn't any better.

And then there is Threads. Marketed as Meta's "friendly" entry into the Fediverse, Threads speaks ActivityPub just enough to claim the label: opted-in accounts can publish posts that show up on Mastodon, and replies and likes cross over. But the integration is shallow by design. Only a fraction of accounts have opted in to federation, that suggests the integration is more marketing than commitment. Threads is the Zuckerverse in ActivityPub disguise: a deception for users who think they're joining something open, and a Trojan horse for the Fediverse, because every account that opts in is another data point flowing back to Meta. If you want to be on the Fediverse, be on the Fediverse. Don't be on Threads.

YouTube is the platform creators cannot quit. It is the second largest search engine in the world, the de facto archive of human video knowledge, and the only realistic place to build a video audience at scale. Google knows this, and it behaves accordingly. In 2026, the platform is openly battling what even they themselves have called "AI slop": mass-produced, low-quality synthetic videos are now prominent among the recommendations to new users. If you try alternative frontend apps on Android that (more ir less) work without an account, you might see this trend too. YouTube has responded with a purge of major slop channels, and burning millions of dollars in annual revenue. This is presented as a crackdown, but it is also an admission: the algorithm was rewarding this content in the first place. Creators who have spent years building audiences complain constantly about skewed recommendations, demonetization, and a system that punishes anything not optimized for the algorithm. And even if you pay for YouTube Premium to escape the ads, Google still profiles you, your watch history feeds Gemini, the company's AI assistant, which collects more user data than any other major AI chatbot. The subscription fee buys you fewer interruptions, not less surveillance.

X is the fascist elephant in the china shop, and at this point the metaphor is too generous. Since Elon Musk's 2022 acquisition, the platform has become a vehicle for his personal politics, a megaphone for his feuds, and a home for an AI chatbot, Grok, that has generated sexualized deepfakes of real people - including minors - prompting investigations by the EU, the UK, and other governments. Grok has also, on separate occasions, praised Hitler and been caught producing antisemitic content, which Musk attributed to the chatbot being "too compliant to user prompts". The platform's own engagement-boosting mechanics have been documented amplifying right-wing and extremist content. None of this has meaningfully shrunk the user base, because the people who left X were replaced by people who didn't care, and because the network effects of a global town square are nearly impossible to walk away from.

TikTok is, simply, the worst of them. The research on its effects is no longer ambiguous: systematic reviews find an overall negative impact on adolescent mental health, with documented associations to addictive use patterns, decreased self-esteem, and the spread of mental illness behavior. Its algorithm is the most ruthlessly optimized attention-extraction machine ever deployed at scale, and competitors have copied it: Instagram Reels, YouTube Shorts, even LinkedIn now run the same playbook. TikTok is also owned by a company with documented ties to the Chinese state, has been banned or threatened with bans in multiple countries, and has been caught spying on journalists. I tried it once. It is not for me. It is hard to think of a single metric by which TikTok is a net good for the world. Especially now, with the US app in the hands of crooks.

Why didn't we just leave?

If the platforms are this bad, why is no one leaving in numbers that matter? The honest answer is that we have been here before, and the precedent is not encouraging, and I should know, because I was there. You were probably there too.

I am a Xennial, old enough to remember when MySpace was The Internet. I was what we would call "terminally online" today, given the technical circumstances of the time. I had a profile. I customized it a bit. I picked the song. I had friends on it. I followed bands on it. And then, sometime around 2009, I left for Facebook, the way millions of other Xennials and Millennials did, because Facebook was cleaner, faster, and felt like the future. We announced our departure. How naive we were. We thought we were upgrading. We were just trading one lock-in for another, and the new one came with a business model built on surveillance capitalism we hadn't yet learned to name.

In 2008, MySpace was the most popular social network on the internet. By 2009, Facebook had overtaken it in the United States, and MySpace's UK user base had tanked. The reasons were not primarily about scandals or ethics. MySpace was slow, cluttered, full of spammy auto-playing ads (you could manage those with AdBlock though), and had failed to keep up with the cleaner, leaner, faster Facebook product - despite the latter's lack of customization. People left because the alternative was still better, not because they had a moral awakening.

This is the part of the story that should give us a pause. The MySpace exodus worked because there was a single, clearly superior alternative waiting in the wings. Today's major players are not facing a Facebook coming for their users. This is not a MySpace sundown situation. They are facing a fragmented landscape of smaller alternatives, each of which is missing some critical piece (the audience, the features, the network effect) that would make leaving feel like an upgrade rather than a sacrifice.

So people stay. Creators stay because their livelihoods depend on the algorithm. Small businesses stay because their customers are there. Friends stay because their friends are there. Parents stay because their kids' school announcements are there. The platforms know this, and they have built their entire business model around it: the product is not the feed, the product is the lock-in.

The lock-in diagnosis

There are essentially four forces keeping us tethered, and they operate in combination:

These forces are real. You are not fighting windmills. But they are not permanent. And they are not the whole story.

The open social web

While the mainstream platforms have been busy optimizing for engagement, a parallel ecosystem has been quietly building. Slowly, imperfectly, but with a fundamentally different architecture.

The first wave was Mastodon, a federated, open-source microblogging network built on a protocol called ActivityPub. When Elon Musk bought Twitter in late 2022, Mastodon absorbed a migration wave that pushed its monthly active users to a peak in the millions. The main instance struggled. The wave receded, but the floor is meaningfully higher than the pre-Musk baseline. Mastodon proved that people would leave Twitter in waves, and that some of them would stay away from the Elonverse for good.

The second wave is Bluesky, built on a different protocol called AT Protocol. Ironically, it came from the Twitter labs, a vanity project nurtured by Jack Dorsey, meant to be the new protocol for the bird app. Dorsey left Twitter, and the bird cage was open - for a butterfly. Bluesky user numbers have been growing steeply, driven by waves of users leaving X primarily. Daily active users are still a fraction of the mainstream platforms, but the trajectory is real and the demographics skew toward the engaged, the technical, and the fed-up, as well as the marginalized, aka the early adopters who tend to define the culture of a new platform.

What makes Bluesky different from a typical Twitter clone is what the community has built around it. The AT Protocol is designed so that your account, your identity, and your data are portable across applications. This has enabled an explosion of what the community calls the Atmosphere, a growing ecosystem of apps that all use the same login, the same follows, the same identity, but offer radically different experiences.

Want a Twitter-style feed? The default Bluesky app works just fine, but there are other apps with a similar approach and a unique flair. Want long-form blogging? There is an entire standard for it, called Standard.site, with implementations like Leaflet, Offprint, and pckt.blog(where this post is published). Want an Instagram-style photo experience? Pinkleapis a photo-first client for Bluesky (also for Mastodon, and Pixelfed). Want a TikTok-style vertical video feed? Sparkis one of several attempts, if that's your thing. Want a Last.fm experience that scrobbles your listening (Spotify integration in the pipeline, would be nice to see other platforms too, given Spotify's reputation that fits into the problems mentioned above) and supports periodic imports from Last.fm itself (whom just became independent again, escaping from the tarnished embrace of CBS)? Rocksky. Want to track your reading? BookHive. Want to discuss films, series, books, music? Popfeed. Want to share links across the open web? Disperse. Want a fine profile page with a custom domain? blento. There is a directory of all of this and more at ATStore, and it is being updated constantly.

This is the part that matters. The ATmosphere is not a single platform competing with Twitter. It is a substrate for many platforms, none of which can fully enclose you. Your account works across all of them. Your follows work across all of them. If you stop liking the default app, you switch apps without losing your network. This is the structural opposite of the lock-in.

And if you want to talk to people on Mastodon, or any other ActivityPub-based network like Pixelfed or PeerTube, there is Bridgy Fed, a bridge that translates posts, follows, likes, and replies between the AT Protocol and ActivityPub. It is not perfect, bridging always involves some friction, if you remember the Mastodon-Twitter crossposter-bridge-thing, but it means the open social web is not two islands. It is one archipelago, and the bridges are real.

The future is open, and it is already here

I want to be honest about the limits of this argument. Bluesky is not yet at the scale where leaving X feels costless. Mastodon's onboarding is still confusing for non-technical users. The Atmosphere is full of half-finished, but enthusiastically built apps, and ambitious projects that may or may not survive, pending the cost of scaling to user numbers eventually. The mainstream platforms are not going to surrender their users voluntarily, they might even block mentioning alternatives, and they have every incentive to make their products just good enough to keep us from leaving.

But the structural argument has changed. For the first time in the social media era, there is a credible alternative that is not a single company you have to trust, but a set of protocols and communities you can move between. The lock-in is being broken not by a competitor that does the same thing slightly better, but by a different architecture entirely: one where your identity is yours, your data is portable, and your network is not held hostage by a single app's decisions about what to show you.

I keep up with the artists, projects, and people I care about. I had to re-register on a couple of mainstream platforms for professional reasons, and I notice every time I open them: the ads, the slop, the AI assistant I never asked for, the algorithm that has decided what I want to see. The contrast is the point. Once you have lived on the open social web for a while, the closed web starts to feel like what it is: a machine that runs on your attention and gives back, at best, a thin simulacrum of connection.

The MySpace exodus worked because there was a single better alternative. This one is different. There is no Facebook waiting to absorb the refugees. There is a federation of small, weird, whimsical, imperfect alternatives - often built by marginalized, and neurospicy folks - that are slowly becoming good enough, and that, critically, cannot be bought, enshittified, or sold to the highest bidder, because they are not owned by anyone in the way Meta or Google or X is owned.

The future of social media is not a better feed. It is the end of the feed as a product. It is your identity, your network, and your data, belonging to you, and a thousand small apps competing to give you the best experience, none of them able to lock you in.

That future is not a promise. It is a place you can already stand. The only question is how many of us are willing to walk there, and whether we are willing to bring the people we care about with us.

Please share this post either in the Atmosphere, or on the Fediverse, but most importantly: drop it in the mainstream. I will post it on X and Threads (yeah, I know), you can boost, repost, like, comment, whatnot there as well.

Disclaimer: English is not my first language, therefore proofreader-corrector tools were used to give this post a more concise shape.