You Peanut: How a One-Liner Becomes a Pile-On
He made a small point about free speech and got back a clown, a nut, and a lecture on American gun deaths. Not one person argued with the point.
He made a small point about free speech and got back a clown, a nut, and a lecture on American gun deaths. Not one person argued with the point.
He made a small point about free speech and got back a clown, a nut, and a lecture on American gun deaths. Not one person argued with the point.
A bloke I know, or really a few blokes I know rolled into one, posted a dry line on a Tuesday night. Something to the effect that you can tell how free a country's speech is by how nervous its government gets about jokes. Nothing clever. The kind of thing you'd mutter at the telly. By Wednesday morning he had a stack of replies. A cartoon clown. A slow-clap gif. "Galah." "Mate, are you OK?" "Did a chatbot write this for you?" "Have a look at the States before you start." And a neat table of firearm deaths, which nobody had mentioned.
I counted for him, because he didn't want to. Plenty of replies. None agreed, which is fine. What struck me is that almost none disagreed either. They didn't touch speech at all. They reacted to who they thought he was and which team they thought he played for. That's the whole essay, really: a pile-on is mostly a reaction to the messenger, not the message. I'm not taking a side on his point. It wasn't a great joke. I'm interested in the machine.
The cleanest evidence for this is a 2003 study by Geoffrey Cohen, titled, helpfully, "Party over policy". He gave people a policy to judge and varied one thing: which party was said to back it. People tended to back the policy their own side supposedly supported, even when its content cut against their usual views, and they mostly didn't think the label had moved them. The content mattered less than the badge.
Dan Kahan calls a related habit identity-protective cognition: we process claims in ways that protect our standing in the groups we belong to. His 2013 study found people across the spectrum doing it, and the more reflective thinkers were, if anything, better at it. It's not a stupidity problem. It's a belonging problem.
These are lenses, not mind readers. I can't tell you what any one replier was thinking. But they fit the shape of the replies. Nobody read the post. They read the badge they'd pinned on it.
Once you've decided the messenger is from the other mob, a second reflex kicks in. John Jost and Mahzarin Banaji proposed system justification theory in 1994, and Jost, Banaji and Brian Nosek reviewed a decade of evidence in 2004. People are motivated to see the system they live under as basically fair, and some of that work suggests the defence firms up when the system is poked by an outsider.
None of the repliers seemed to love the government. Most would have said worse on another day. But a criticism from the wrong person isn't heard as criticism. It's heard as an attack on us.
Then the bit we all know. A bloke who makes a point in public is, in a small way, claiming to see further. Norman Feather spent years on this. His 1989 paper looked at how Australians feel about high achievers and their fall, and a 2008 study looked at how people's own status shapes their resentment, schadenfreude and sympathy when a high achiever fails. The short version: whether we enjoy a fall depends a lot on whether we think it was earned, and on how we see the person.
That last part is the messenger again. The same line from a mate reads as fair comment. From someone who seems up himself, it begs to be cut down. Most cultures have a version of this. Ours just has a name and a good metaphor.
"Look at the States" and the gun numbers weren't rebuttals. The post was about speech here; the replies were about guns somewhere else. They answered the tribe they'd assigned him, not anything he wrote. And they restored the ranking. We're fine, because over there is worse. A comparison doesn't have to win the argument. It only has to make the discomfort go away.
Now the machinery. In 2017 William Brady and colleagues found that moral and emotional language made political posts spread further, mostly among people who already agreed. In 2021 Brady, Molly Crockett and others found that people who got likes for outrage tended to post more outrage later. Crockett's 2017 essay sets out why: online, outrage is cheap, the audience is huge, and the reward arrives in seconds. Note "among people who already agreed". The feed rewards you for performing for your own side, which is a messenger game, not a message game.
There's a twist. Takuya Sawaoka and Benoît Monin found that once many people have condemned someone, onlookers start to see more condemnation as excessive and feel for the target. Past a point, the pile works against itself. Nobody in it can see where that point is.
And it's common. Pew's 2021 survey found 41 per cent of American adults had been harassed online, and half of those said it was over their politics. That's US data, so treat it as a rough guide.
Ridicule isn't mainly aimed at its target. In 2000 Leslie Janes and James Olson ran experiments they called jeer pressure: people who just watched someone get mocked became more cautious and more likely to conform afterwards.
That's a lab finding, and a feed isn't a lab. But it suggests the clown gif works best on the lurkers, and what it teaches them is about identity. Not "that argument is wrong" but "people who say that sort of thing are clowns".
The newest move is the AI jibe, and it's the purest messenger attack of the lot. It doesn't even need a tribe. It says there is no person here worth answering.
Maurice Jakesch, Jeffrey Hancock and Mor Naaman ran six experiments with 4,600 people and found they couldn't reliably spot short self-descriptions written by AI. They leaned on cues that feel right and don't work, like first-person pronouns or mentions of family. Weixin Liang and colleagues found AI detectors flagged non-native English writers as machine-made far more often than native speakers. If people and tools both misfire, the accusation says little about the text. What it does is remove the speaker, so the point never has to be met.
Nothing heroic. The bloke I know closed the app, and that was the right call.
A few things I'd take from it. When the replies are about you and not your point, the point hasn't been answered; don't mistake volume for a verdict. Reply once if you must, to the argument, then leave. And the harder one: before you reach for the clown, ask whether you're answering what was said or who you think said it. Most of us have been in the pile. The machine doesn't care which side you're on.
He put the kettle on. By Thursday nobody remembered, including the loudest. Pile-ons feel like a verdict, and they're mostly weather.
If this was worth your time, you can leave a tip.
New essays by email: ricoholt.com/list
Rico Holt · ricoholt.com
If this was worth your time, you can leave a tip. Rico Holt runs on readers, not big tech or big media money, and the essays stay free.
Leave a tip · A$5Get the next essay and the book launch price first.
Privacy: your email gets used for one thing, sending you Rico Holt essays and book news. It's stored with Netlify, our form host. Never sold, never shared. Want off? Reply to any email and you're gone.