How a Lie Got Made This Weekend

A Jewish friend of mine got screamed at on the street last week. Just walking. A small town in the Midwest, of all places. Somewhere that hasn't historically been the kind of town where this happens.

That bothers me more than I can quite articulate.

Hatred isn't new, obviously. What's new is how ordinary it's become. How ambient. The person who screamed at him didn't know him, had no quarrel with him, no history between them. He just saw a Jewish man walking, and decided that was something worth yelling at.

Here's what I keep coming back to. Most of the discussion of all this treats it like an information problem, as if people simply don't know enough and the fix is more facts, sharper rebuttals, better fact-checks. I don't think that's right. The issue is interpretation, not information. Once the lens through which a person reads the world gets bent badly enough, more information actually makes the situation worse. Every fact that contradicts the worldview becomes more evidence for it. Every correction gets folded back in as further proof of the conspiracy.

I've been documenting how some of these networks actually operate for a while now, and the longer you look, the more obvious it becomes that what we're watching has structure to it. There's an architecture. And the genuinely new thing, the thing that's only become possible in the last few years, is that the architecture is finally something we can see.

Graph databases like Neo4j, paired with decent hardware (a Jetson rig will get you most of the way there), make it possible to ingest social media at scale and render it as what it actually is: a network of nodes and edges. You can map entire conversation trees. You can layer in funding records, sponsorships, organizational affiliations. You can run the timeline backwards and watch a single claim mutate as it spreads outward from wherever it started. Millions of data points, processed in seconds, rendered as a picture the eye can read.

What you get isn't a clean partisan story. It's pretty messy. But it's legible.

You can find what investigators sometimes call source zero, the small clusters of accounts where a narrative first surfaces. You can see the amplification clusters that pick it up: the high-follower accounts, the ideological hubs, the bridge accounts that move material between communities that otherwise wouldn't see each other. You can watch a claim mutate as it travels, and you can see which mutations survive. In a lot of cases you can trace the funding behind the channels doing the heaviest lifting.

This past weekend the machinery did exactly what it always does, and it's worth slowing down to look at how.

In Washington on Saturday, a man named Cole Tomas Allen charged the security checkpoint at the White House Correspondents' Dinner with a shotgun, a handgun, and several knives. He had left a manifesto. He had been posting his views online for some time. His own brother and sister had warned police, before and after, that something was deeply wrong with him. The manifesto, per CBS and the Washington Post, laid out an intent to target Trump administration officials and reportedly carried anti-Christian sentiment. A Secret Service officer was hit and survived because of his vest. The shooting itself was a real event.

By the next morning none of that mattered.

Within hours, an image was circulating across X showing someone who looked like Allen wearing a sweatshirt with the IDF emblem. The photo was traced to a social media account that has since been deleted. Snopes ran it through several AI-detection services like Hive, Sightengine, Truthscan, and a few others. The results came back anywhere from two percent to ninety-seven percent likelihood of being AI-generated. The forensic tools themselves couldn't agree. In this environment, that's all the daylight a fabrication needs.

An Instagram account claiming to belong to Allen got screenshotted and passed around as further confirmation. But that account spelled his middle name "Thomas." Every mainstream outlet had it as "Tomas," no h. Within a day that same account had switched its profile picture to John Hinckley Jr. and updated its bio to a line about doing it "for the lulz." Anyone looking carefully could see it was staged. Nobody looked carefully. The screenshot was already in the canon.

Then came the conclusions. On one track: Israel did this. The shooter was a Zionist. This proves what we've been saying. On another track, sometimes inside the same networks: the photo was a false flag designed to discredit Israel. Both narratives drew on the same fabrication. Neither needed the fabrication to actually be real. The closed loop doesn't require consistency. It requires suspicion, and suspicion is renewable.

Some of the high-profile accounts that initially helped propel the photo eventually came back with caveats: unconfirmed, possibly AI, treat with caution. By then it had been seen by millions, screenshotted into group chats, embedded in YouTube thumbnails, and stitched into the mental file a certain kind of viewer keeps on what is "really" going on. The correction always arrives later, smaller, and mostly ignored.

And running alongside all of that was the older reflex, applied to the shooting itself. Staged. Fake. Convenient. False flag. The same pattern that followed Butler in 2024, the same pattern that follows almost every public event of consequence now. This isn't skepticism. Skepticism asks a question and updates when an answer comes in. What's happening is the inverse of that. It's a closed loop that takes every new piece of evidence as more proof of whatever conclusion it had already reached.

I think the right word for this is contagion, and I mean it pretty literally. Not because the people caught up in it are stupid. That framing kills the argument before it starts and forecloses the readers who most need to hear it. It's because the mechanism doesn't actually run on reasoning. It runs underneath reasoning. Repetition does the work that verification used to do. Group identity stands in for evidence. Whether a thing is true gets quietly replaced with whether it confirms what you already suspected.

I should say, before going further: I notice some version of this in myself anytime I open the news. The pull toward an interpretation that lines up with what I already believe is real, and it's strong. None of us is fully outside this. The question isn't whether you've ever felt the gravity. It's whether you can still pull against it.

This isn't only a left phenomenon or a right one, which is part of what makes it hard to talk about cleanly. The vocabularies differ, the villains differ, the causes attributed to events differ. But the underlying mechanism is the same on both sides: closed interpretive systems that can't self-correct, only self-confirm. My friend getting screamed at on a sidewalk and a real shooting now being called staged are products of the same disorder, even if the people having those two reactions would imagine themselves as opposites.

The danger here isn't bad ideas. Bad ideas get corrected all the time; most of intellectual life is that exact process. The danger is losing the ability to recognize correction at all. Once a contradiction itself becomes proof of the conspiracy, once "they would say that, wouldn't they" is the answer to anything, you don't have disagreement anymore. You have people sharing the same physical world while occupying different ones.

And so my friend gets screamed at on a sidewalk by someone who doesn't know him, has no actual grievance with him, and whose internal map of the world has been rearranged so that every event traces back to a single culprit. Every assassination. Every election. Every market move. That's what a corrupted lens produces. That's what years of saturation does to an ordinary person who would have just waved at his neighbor a decade ago. And it's also what produced that IDF sweatshirt within hours of a shooting that, by every available account, had nothing to do with Israel. The frame is now so primed that when reality fails to provide a Jewish villain, the network manufactures one and broadcasts it to millions before lunch.

The question this leaves us with is harder than the political ones we keep arguing about. It isn't whether the right or the left has it figured out. It's whether we can still be people for whom evidence is allowed to mean something, even when it cuts against what we want to believe. Whether we can sit with the discomfort of being wrong long enough to actually update.

Because once that capacity goes, the categories of enemy and ally become almost arbitrary. Anyone can be made into anything. And meanwhile, my friend is still trying to walk through his own town without getting yelled at.