The Summoning Problem

There’s a certain kind of online weirdo—you know the type, they post about “egregores” and “hyperstitions” and treat the collective unconscious like a cryptocurrency—who would look at Minneapolis right now and say something has been summoned. That the city has become possessed by something that wants conflict, that feeds on attention, that grows stronger with every new incident.

They’re wrong. But they’re not stupid.

What’s actually happening in Minneapolis is more interesting than possession, and considerably harder to dismiss. Three federal agent shootings in three weeks. Massive protests scaling faster than anyone organized them. Local and federal authorities in open jurisdictional conflict. Videos proliferating faster than context. Every new development slotted immediately into existing narratives before anyone can evaluate it fresh.

The city isn’t possessed. It’s become a hotspot—and hotspots, it turns out, behave in ways that make possession look like a reasonable hypothesis.

The Machinery of “Just Escalating”

Here’s what’s actually happening, stripped of both mysticism and false naivety:

Event clustering creates expectation. One shooting is an incident. Two is a pattern. Three is a story. After three, every subsequent action—by agents, by protesters, by officials—gets read through the lens of “the Minneapolis situation.” Events stop being evaluated on their own terms. They become data points in a narrative that now has momentum.

Visibility attracts visibility. Cameras show up because cameras already showed up. Media covers it because media is already covering it. People travel there because it’s where things are “happening.” This isn’t conspiracy; it’s just how attention economies work. But the effect is that scrutiny concentrates, and concentrated scrutiny produces friction that wouldn’t exist under normal observation conditions.

Institutions harden against each other. Federal agencies double down to avoid looking like they’re “losing control.” Local officials escalate oversight and criticism to avoid looking complicit. Each side treats the other’s response as confirmation of bad faith. The Minnesota governor announces a state investigation into federal “lies.” The feds keep operating. Neither can back down without looking like they lost.

Narrative locks in. New incidents don’t get fresh evaluation. They get slotted into the existing storyline. A moment of restraint gets interpreted as tactical positioning. A moment of force gets interpreted as pattern confirmation. Even neutral actions become charged because the context is charged. The interpretive frame precedes the facts.

None of this requires coordination. None of it requires conspiracy. None of it requires anyone to be acting in bad faith, though some people certainly are. It’s just what happens when feedback loops couple tightly enough that a location starts exhibiting coherent behavior.

The system behaves as if it has preferences. But there’s no one at the switchboard.

What the Egregore Crowd Is Pointing At

This is where it gets uncomfortable.

The people who talk about egregores and summoned entities aren’t hallucinating the phenomenon. They’re misidentifying its ontological status. When they say “an egregore has been summoned,” they’re pointing at something real: emergent coordination that looks agentic without having an agent, outcomes that feel intentional without having an author, behavior that seems aimed without anyone aiming it.

The language is wrong but the pattern recognition is sound. What they’re describing is a coupled socio-technical system that amplifies expectation into action. The feedback loops are real, and so are self-reinforcing dynamics. The way the system starts to exhibit preferences and resist certain interventions—that’s real too.

They reach for demonology because they lack better vocabulary. “We built a coupled system that amplifies expectation into action” doesn’t feel adequate to the uncanny quality of watching a city start behaving like an actor. So they reach for metaphors of summoned entities instead.

The mistake isn’t seeing the pattern. The mistake is treating it as ontological rather than emergent. It’s not that something exists out there, pulling strings. It’s that we’re doing this to ourselves, together, through machinery we mostly don’t acknowledge.

The Asymmetry That Matters

Here’s the part that’s harder to sit with.

When online subcultures exhibit the same dynamics—attention loops, expectation shaping behavior, symbolic weight causing coordination without coordination—we’re comfortable with mechanistic explanation. “That’s just how attention economies work.” “That’s just how memes spread.” We don’t demand villains. We accept that distributed systems produce coherent-looking behavior through feedback, not conspiracy.

But when the same dynamics show up in contexts with bodies—real shootings, real deaths, real communities under pressure—suddenly the mechanistic explanation feels insufficient. We demand moral authorship. Someone has to be responsible.

And to be clear: individual actors remain responsible for their individual choices. The officer who pulls the trigger is responsible for pulling the trigger. That doesn’t change.

But the demand for moral authorship at the incident level often substitutes for understanding the system that made the incident overdetermined. We get so focused on “who’s responsible for this shooting” that we stop asking “what conditions made a shooting almost inevitable.”

The system that put an armed federal agent in a residential neighborhood, in a city already primed for confrontation, in a context where any use of force would immediately become fuel for existing narratives—that system is also doing something. The egregore crowd has language for it, but the language is wrong. The political commentary class mostly refuses to engage with the question at all, because admitting that systems can produce outcomes without authors feels like letting someone off the hook.

What This Doesn’t Mean

It’s not saying “both sides are equally bad” or any other false equivalence. The dynamics are symmetric at the level of systems theory; they are not symmetric at the level of power, resources, lethality, or moral stakes.

It’s not saying protests are illegitimate because they’re “just” emergent coordination. Most valuable collective action is emergent coordination. The civil rights movement was emergent coordination. That’s how humans do things together.

It’s not saying you can’t take sides. You can and maybe should. The systems-level analysis doesn’t preclude moral judgment; it just doesn’t replace it either.

What it is saying: if you want to understand why Minneapolis, why now, why this particular configuration of escalation, you have to look at the coupling. The same dynamics that make online communities coalesce around shared symbols make cities become arenas for symbolic conflict. The machinery doesn’t care whether the stakes are follower counts or bodies.

Naming the Machinery

Here’s what I still don’t know: whether seeing this actually helps.

Understanding how the feedback loops work doesn’t exempt you from being pulled along by them. I’m writing this from inside the system I’m describing. The attention infrastructure that makes Minneapolis a hotspot is the same infrastructure that will distribute this essay, slot it into existing narratives, make it a data point in someone else’s pattern-matching. The observation doesn’t grant escape velocity.

And maybe that’s worse—clarity without leverage. You can see the machinery perfectly and still be caught in it. Systems literacy doesn’t come with an obvious ethics. It doesn’t tell you what to do, who to support, when to act. It just shows you the shape of the thing you’re inside.

I don’t know if naming the machinery changes its behavior or just adds another layer. Once an analysis is out in the world, it becomes part of the system it’s describing. Maybe someone reads this and finds the pattern-match compelling enough to pause before assuming malice. Maybe someone uses the framework to better coordinate escalation. That’s not up to me.

But I think there’s value in having the vocabulary. Not because naming the egregore banishes it—it doesn’t work that way—but because naming the machinery at least makes it harder for the machinery to pretend it’s fate.

The Minneapolis situation isn’t destiny. It’s not possession. It’s not pure chaos.

It’s a coupled system doing what coupled systems do, amplified by attention infrastructure that makes every node visible to every other node, in a political context where institutional actors have locked into adversarial frames they can’t easily exit.

That’s not comforting. But it’s accurate. And accuracy, sometimes, is the best we can do.

The system exhibits agency without an agent. We reach for demon language or chaos language, and both are wrong in the same way: they refuse to see the machinery.

The machinery doesn’t care what story you tell about it. But it behaves differently when you can see it—even if you can’t necessarily escape it.