The Sorting Problem

This is the first in (at least!) a three-part series on the mechanisms of open ecosystems and human behavior A month or two ago, I got into it on Bluesky with @jacky.wtf. Up until that moment, I generally considered Jacky a positive acquaintance; someone I didn't know well, but with whom I was mostly directionally aligned. I started with a statement that I knew was a bit provocative, but was an observation I felt pretty sure about: you can't be a leftist (generally progressive) and hold a fundamentally conservative posture toward AI, especially right now. Not when political polarization has swallowed everything whole. That landed badly. So badly, in fact, he blocked me, and then I blocked him, but then we both unblocked after calming down. So let me try again, with more room to breathe. --- Let me say something very clearly before I go any further: I am not calling anyone a Republican. I am not saying your concerns about AI are categorically wrong. I am not saying Big Tech is altruistic or that commercial LLMs are nothing but sunshine. I hold all of that mess. I also hold what I'm about to say. Both things are true and I'm not going to pretend one cancels out the other. So! Let's get into it. What I mean by "conservative" This is where the thread blew up. Jacky hit me with "words mean things!" when I called the anti-AI position conservative. And yes, they do! That's actually my whole point. There are two ways to understand what a word means. The first is how people use it right now. In 2026 in American political discourse, "conservative" has been so thoroughly captured by the Republican right that using it to describe anyone on the left feels like an accusation. I get that. If you hear "conservative" and your brain goes straight to MAGA, the idea that I'm applying that word to you, a leftist, sounds like an insult. That reaction makes total sense inside the current political moment. But there's another way to understand a word, and it's the one I was using: where the word comes from. What it actually means at the root. "Conservative" comes from conservare — to preserve, to keep, to hold onto what exists. And when I look at the anti-AI left, that's what I see. They want to conserve work for humans. They want to conserve property rights to intellectual property. They want to conserve the current relationship between people and labor. They want to preserve what we have against what's coming. I'm not saying that's wrong. I'm saying that's the direction the position faces. Backward. Toward preservation. And in a hyperpolarized moment where everything gets sorted into "progressive" or "conservative" — where those aren't just political labels anymore but identities — facing backward is a problem. Not a moral problem, though - a strategic one. The sorting machine Ezra Klein laid this out in his 2022 book, Why We're Polarized, and I think it's the piece most people are missing when they talk about AI and the left. Here's the short version: Americans used to hold messy, contradictory bundles of beliefs that didn't all line up with one party. You could be pro-union and anti-immigration. Pro-military and pro-welfare. The parties were big, sloppy coalitions. Over time, through a process Klein calls sorting, all of those identities - your race, your religion, your geography, your cultural tastes, your media diet - collapsed like two black holes into the current partisan binary. Everything lines up now. Everything predicts. And here's the part that matters: the causal arrow flipped. It used to be that you had your beliefs and then found the party that roughly matched. Now you pick the team — or get sorted into it — and the team's positions become your beliefs. Identity drives belief, not the other way around. But not only that, this wasn't what anyone expected. Sociologists had generally though we'd avoid polarization because we had cross-cutting identities. They did not expect politics to eat those things. Now, apply that to AI. AI hasn't fully sorted yet. There are people on the left who use it enthusiastically, people on the left who think it's an existential threat, people holding both positions at once and feeling weird about it. That's an unstable state, but Klein's whole framework says mixed-alignment positions don't survive the sorting process. Eventually one stance will become the "correct" left position and the other will feel like betrayal. The anti-AI left's problem isn't that their arguments are bad, their problem is that their posture reads as conservative in a world where identity organizes everything. The energy of "stop," "go back," "preserve what we had" — that triggers an identity alarm in people who've built their entire self-concept around being progressive. It doesn't matter that the underlying concerns about labor & IP might be legitimate left concerns. The vibe is wrong for the team. Meanwhile, the pro-AI position has an easier sorting path because it feels progressive — new, forward, transformative, future-oriented. Even when the actual corporate structures behind AI are deeply capitalist and exploitative, the directional energy reads as progress. And in a sorted world, directional energy beats structural analysis every time. So the anti-AI left is stuck in a trap Klein's framework predicts perfectly: they can't persuade through argument because belief isn't being driven by argument — it's being driven by identity alignment. And they can't win on identity because their position's preservationist energy puts it on the wrong side of the progressive/conservative feeling. Which means they need a different approach entirely. The machine eats everyone Here's the part that I need y'all to be aware of: this isn't just an anti-AI problem. The sorting machine doesn't have a team. It eats everyone. I saw this play out on my own timeline a couple weeks after I posted that thread. Someone posted a provocative pro-AI take — basically, "if you're so confident Claude isn't conscious, explain how human consciousness works." He got LIT UP for it. Another person who positioned themself as "pro-AI" quote-posted it, positioning himself as the reasonable middle: the loudest anti-AI voices are "simply grandstanding for internet points," and here comes this AI enthusiast making the polarization worse. See how reasonable I am? I can criticize both sides. Except look at what he actually did! In the same post where he's calling out another pro-AI person for inflaming things, he's dismissing the entire anti-AI position as clout-chasing. That's not the reasonable middle. That's someone who wants credit for being moderate while still flattening one whole side into grandstanders. He's doing the polarizing thing and pointing at someone else as the polarizer. I quote-posted it and said, basically, "I love when we blame systemic problems on individuals."I even took the time to clarify directly: this isn't an attack on you, it's an attempt to illustrate a larger systemic pattern. The response? He went right past what I said and pointed back at the original poster he'd quoted: the individual OP "illustrates, in microcosm, some of the systemic issues around AI promotion." But sure, he gets the larger point. Read that again. I pointed at the system. He said "yeah, but look at that guy." The systemic critique just slid right off. He couldn't sit with the possibility that his framing — not just Geering's provocation, but his own "reasonable middle" act — was the machine running. So he acknowledged the structural point just enough to seem like he heard it, and then redirected back to the individual he was already comfortable criticizing. The one person who got closest to a useful read was someone who said, essentially: it's inflammatory and it's pointing at something real. Both things. Both things. That's the move the sorting machine doesn't know what to do with — holding two truths at the same time without collapsing one into the other. The anti-AI left does real, important work identifying harms. The pro-AI left does real, important work identifying benefits. And both sides, when the machine gets its hooks in, reduce the other to caricature — grandstanders or techno-utopians, Luddites or shills — because the machine demands you pick one and flatten the other. I'm trying not to do that. I don't always succeed. But the attempt matters. And this isn't just about AI. I've watched this same machine operate across race, across transness, across every axis where identity gets sorted into teams. I've been eaten by it myself — called a bigot by people I share marginalizations with for the crime of holding two truths at once(that's a story for another day). The point isthis: this is not an AI problem. This is a polarization problem, and AI is just the latest thing it's digesting. And here's the thing that keeps me up at night. The sorting machine doesn't just sort opinions. It sorts people. Once you've been sorted into a position — anti-AI, pro-AI, whatever — you stop being a person who holds a view and start being a representative of that view. A symbol. And symbols can be targeted in ways that people can't survive. When the machine runs hot enough, it stops being a discourse problem and becomes a governance problem. What happens when sorted communities collide inside a space with no structure to hold the collision? When there's no institution absorbing the impact, just individuals standing where the institution should be? That's a different essay. But it starts here, with the machine that makes it possible. The progressive case (that we're trying to make) Here's where I get frustrated: The benefits of AI are real! They're documented in more and more papers! For everyday people — not just tech workers, not just the already-privileged. Accessibility. Translation. Education. Creative assistance for people who never had access to the tools before! So when the anti-AI left denies those benefits exist, when they offer no competing vision except "we will take this away," they lose the ability to persuade. You can't beat something with nothing, baby. What I want to see — what I think the left needs — is a progressive framework for AI that does three things at once: acknowledges the real harms, acknowledges the real benefits, and offers a vision for where this goes that isn't just "back to before." Because "back to before" wasn't great either. Let's not romanticize it, fam. Pro-labor is a stepping stone (and that's not an insult) Back to the initial story, because this is the other place the thread got heated. Jacky does labor organizing work; he sits on a cross-union steering committee. He pushed back hard when I said that being only pro-labor means you're ultimately pro-work. He called it a nothing burger and said I was being one-dimensional. I think he was hearing me say his work doesn't matter. That's not what I said, but I do understand why it landed that way, and I want to be more precise about what I do mean. Here's what he said that was particularly correct, and I didn't really even grok it at the moment: > "The concept of work governance affects housing, healthcare, family dynamics. Work governs life, even in absence of it." That's exactly right. And it's exactly why pro-labor isn't enough. Let me break this down, because I think there are two different things happening that look like a disagreement but are actually two people describing different parts of the same system. The labor organizer is mapping the structure — how work functions right now as the governance mechanism for all of life. Housing, healthcare, family stability — all of it runs through work. That's not a defense of work. That's an indictment of the system built around it. And he's doing the hard, immediate, right now work of making that system less brutal for people trapped inside it. I'm describing the destination — a world where work is optional, where most people don't have to do it, where those who choose to do the work others don't like are ridiculously well-compensated, where abundance is the baseline and not the reward for selling your body through wage labor. Jacky is saying "you can't just yank work out — everything else collapses with it." AGREED! I'm saying "which is why we need a vision for what replaces those structures, not just better conditions within them." We need both; the structural map and the destination, the person navigating the current terrain and the person pointing toward the horizon. Neither one alone is sufficient, they either get lost or stuck. But here's what I won't do: I won't limit my imagination to a world where unwanted work is mandatory in order to live well. That's not the ceiling. It can't be. That's a way station, at best, a treadmill at worst. I am pro-labor as long as we exist in a capitalist economy. That's why I support unionization, worker-owned cooperatives, universal healthcare, all of it. But I am also pro-freedom from work, which means I'm ultimately pro-getting-past-the-need-for-labor-as-a-survival-mechanism entirely. Sitting only at pro-labor, as your end state, means you still believe we must work at things we don't want to do, that don't make us happy, in order to live. That is a fundamentally puritanical perspective. And I think we can do better. And of course we end up at Marx (yeah, Marx) Marx believed automation could be the thing that broke the back of capitalism, even as it would be terrible when in the hands of capitalists.I think he was right about both parts. Personally, I believe the capitalists have severely overestimated their ability to stay in control of this. Capitalism doesn't work without a populace that has money to spend, and if AI and automation eliminate the jobs that give people that money, the whole engine stalls. That's not a theory — that's the internal logic of the system eating itself. I don't know the time frame. I don't know how ugly the middle part gets(and y'all, it's gonna get ugly). I do think we have to keep going and break through, because the alternative — trying to hold back the tide — has never worked and isn't going to start working now. So what am I actually saying? I'm saying the anti-AI left has a sorting problem, not an argument problem. The concerns are valid. The posture is strategically fatal. I'm saying "conservative" isn't an insult — it's a diagnosis, and if you can't hear the diagnosis because the word has been captured by partisan politics, that's actually proof of the very polarization dynamic I'm describing. I'm saying pro-labor is necessary and insufficient. The structure matters and the destination matters. The people doing the work of the present as well as the people imagining the future need each other, and dismissing either one is how we all lose. I'm saying the left needs a progressive vision for AI that doesn't pretend the harms aren't real and doesn't pretend the benefits aren't real. One that faces forward. One that imagines better than what came befre. One that the sorting machine cannot eat. I don't have all the pieces of that vision, but I know what the first step is: stop trying to hold back the tide, and start building the boat. Now, get comfy, because there's a whole hell of a lot more to come. This post is AIL 3.0