Back To The Bird

I Asked Elon's Robot How to Save Twitter From Itself

There was a time when Twitter was genuinely one of the most intellectually generative spaces on the internet. I know because I lived it. A meaningful portion of the people I consider real-world colleagues and friends first encountered me as Citizen Stewart, a persona I built on that platform through years of posting, debating, and connecting with researchers, educators, advocates, and thinkers who had real things to say and the receipts to back them up.

That platform was, in its best moments, a working model of what I’ve come to think of as open learning in the wild. Not a school, not a course, not a curriculum, but something rarer: an infrastructure for intellectual community. You could follow a sociologist arguing with a historian in real time. You could ask a question into the void and have three credentialed people answer it before lunch. Knowledge moved horizontally, across disciplines and institutions and ZIP codes, in ways that formal education has never managed to replicate. For a lot of people, especially people who had been locked out of elite academic spaces by cost or credential or geography, Twitter at its peak was the closest thing to a free university that actually worked.

That version of Twitter is largely gone.

What replaced it is something I don’t think even its new owner would describe as a marketplace of ideas with a straight face. Verification, once a signal of authentic identity, now means you paid eight dollars. The bot population may well outnumber the humans. Outrage farms, many of them coordinated from abroad and designed explicitly to inflame American civic life, have colonized the feeds. Racism, sexism, and pornography spread without meaningful consequence because the people who used to enforce community standards were laid off en masse in the first weeks of new management. And a significant share of the scholars, researchers, and educators who once gave the platform genuine intellectual weight have relocated to Bluesky, to Mastodon, to Substack, to LinkedIn, or simply to the quieter dignity of logging off.

This is what happens when the infrastructure of intellectual community degrades. And make no mistake, that is what Twitter was: infrastructure. The same way libraries, public universities, and community centers function as physical infrastructure for learning and civic life, platforms like early Twitter functioned as digital infrastructure for the kind of horizontal knowledge-sharing that formal institutions rarely enable. When that infrastructure gets handed over to someone whose incentives are better served by engagement than by enlightenment, the community doesn’t just move somewhere else. It fragments. It loses the connective tissue. The scholars scatter, the conversations go dark, and what’s left is engineered for reaction rather than reflection.

We are living through an anti-intellectualism epidemic right now, and it didn’t arrive from nowhere. It was cultivated. Algorithmic systems that reward outrage over insight, platforms redesigned to amplify the loudest and most inflammatory voices, the deliberate dismantling of the norms and moderation that once made serious conversation possible. The result is a public that is increasingly unable to distinguish signal from noise and a digital commons that has become, in the most literal sense, anti-social.

That last phrase is worth holding. Social media was supposed to make us more connected. Instead, at its worst, it has made us more reactive, more tribal, more suspicious of complexity. We need to make anti-social media social again. And I mean that not as a slogan but as a practical challenge: how do we rebuild the conditions for genuine intellectual exchange on platforms that have been deliberately engineered against it?

So I ran an experiment. For the first time, I asked Grok, the AI chatbot Elon Musk built into X, a direct question: “How can I use X to encounter nothing but real people who are educated and not here for political debate?”

What Elon’s robot told me was actually more useful than I expected, and more honest about the platform’s problems than I anticipated from a tool designed by the man who created those problems. The core of its advice: abandon the algorithmic “For You” feed entirely and switch to the chronological “Following” feed, because the default algorithm is deliberately engineered to surface outrage. Curate ruthlessly, unfollowing anyone who occasionally posts politics even if you respect them otherwise. Build private Lists organized by topic, things like physics, serious literature, data science, urban planning, and read from those instead of your main timeline. Use X Communities, which are topic-specific spaces that can be moderated to exclude off-topic noise. Mute aggressively, not just accounts but keywords, politician names, and the entire vocabulary of performative political combat.

Is this advice correct? Mostly, yes. Is it sufficient? That’s the harder question.

What Grok didn’t say, and what I think anyone trying to reclaim this platform needs to understand, is that the educated, intellectually serious community that once thrived on Twitter didn’t leave because they lacked the technical skill to curate their feeds. They left because the platform’s ownership made a series of deliberate choices, including reinstating banned accounts, gutting trust and safety infrastructure, amplifying right-wing political content through opaque algorithmic adjustments, and signaling clearly through conduct and public statement that certain kinds of users were more welcome than others. Curation can filter your experience. It cannot change the incentive structure of the platform itself.

And this is precisely the open learning problem in miniature. We keep trying to route around broken infrastructure rather than demanding better infrastructure. We develop personal workarounds, clever hacks, individual strategies for extracting knowledge from systems that were not designed to help us learn. That’s useful, and I’m going to use every one of Grok’s suggestions. But it’s also a form of accommodation to a situation that shouldn’t exist. The question of who controls the digital infrastructure of intellectual life is not a tech policy question. It is an education question, a democracy question, a question about what kind of society we are building.

That said, I’m not prepared to abandon the network I spent years building. The Lists strategy is real and underused. The Communities feature is genuinely better than most people realize. Switching to the Following feed is the single highest-leverage move available, and most people haven’t done it because the app defaults you back to algorithmic chaos every time you open it.

I’m going to try the Grok prescription seriously for the next few weeks and report back. Not because I trust the platform or its owner, but because the connections I made as Citizen Stewart belong to me, not to him. And because if the open learning concept I’m obsessed with is real, if the idea that knowledge should flow freely between people without institutional gatekeepers actually means something, then reclaiming even a degraded platform for intellectual exchange is worth the effort.