Picture a restaurant that refuses to serve anything anyone in the world might be allergic to. No gluten, no dairy, no nuts, no shellfish, no soy. What's left isn't a restaurant. It's self-denial on a plate — a menu engineered to offend nobody that, in the process, feeds nobody either.1 The analogy is blunt on purpose: you cannot communicate anything with texture, humor, or edge to an audience large enough to contain someone who'll object to any given texture, joke, or edge. The choice isn't between offending some people and offending no one. It's between saying something and saying nothing.
"Outrage" gets used for two entirely different phenomena, and collapsing them together is where brands and communicators get genuinely stuck. Genuine offense is costly to produce — it requires someone to actually be hurt, and hurt has a real, if hard-to-measure, cost to the person experiencing it. Confected outrage is nearly free to produce — it requires only that someone perform distress publicly, which costs them nothing but a moment's typing, especially anonymously.2
The two phenomena look identical from the outside — both present as public complaint about the same piece of content — but they're generated by structurally different processes, and the reliable way to tell them apart isn't to examine the complaint's content, it's to examine what the complaint actually cost the person making it.
Here's the sharpest version of the test the source offers: a well-known public intellectual solved the problem of hecklers and disruptive protesters at his talks not through security or moderation, but by scheduling talks at 8:30 in the morning. Attendance by disruptors collapsed. The people scandalized enough to protest weren't, it turned out, scandalized enough to set an alarm clock.3
That's the whole mechanism in one image. Genuine distress is costly-signaled by definition — a person who's actually been hurt by something doesn't need to perform the hurt, because the hurt itself is the cost already paid. Confected outrage needs cheap channels specifically because it has no underlying cost to draw on — it exists entirely as performance, and performance seeks the cheapest available stage. Raise the cost of the stage even slightly (an early morning, a stamped letter, a phone call instead of a tweet) and confected outrage evaporates at a rate genuine offense never does, because genuine offense was never dependent on the stage being cheap in the first place.
This isn't a permanent feature of human nature — it's a consequence of how cheap public complaint has become. Complaining to a newspaper used to require paper, a stamp, and days of waiting for a reply; under those conditions, most complainers really were experiencing something close to the distress they were reporting, because the friction filtered out anyone whose distress wasn't real enough to survive the delay.4 Digital complaint removed the friction entirely. You can complain about anything, interminably, from a chair, anonymously, with zero social cost — and the source notes a version of this that would be unacceptable in any face-to-face setting (repeatedly, obsessively complaining about one narrow topic gets you avoided at a pub) simply doesn't carry the same social cost online, where there's no room to physically avoid the person and no accumulated reputational drag from being visibly, tediously fixated.4
You publish something with a real edge to it — a joke that could land wrong, a claim that isn't hedged into meaninglessness. Within hours, a visible cluster of complaints appears. Your first instinct is to treat all of it as one undifferentiated wave of offense and to start drafting an apology.
Instead, you run the actual test. You look at the accounts complaining loudest — not at what they're saying, but at their broader pattern. Is this a normal account with a normal range of posts, who occasionally posts something critical, who happened to be genuinely bothered this time? Or is this an account whose entire visible output is complaint, about this and everything adjacent to it, on a loop? The first is real, and worth taking seriously even if you ultimately disagree. The second is a pattern, not a person reacting — and the source's own heuristic is exactly this: check what someone tweets about generally before weighting their specific complaint.5
You also notice what the complaint actually cost the complainer. A letter, a documented pattern of consistent principle, a willingness to be named and hold the position under their own identity — these carry weight. An anonymous account's single drive-by post, indistinguishable in form from a thousand others performing the same gesture that week, doesn't — not because the feeling behind it couldn't be real, but because the format gives you no way to tell, and you can't run a business or a public voice by treating every unverifiable signal as equally load-bearing.
The source's strongest evidence is a genuine natural experiment — a 1991 UK newspaper price-war ad that ran without complaint, that the source states plainly "would practically get you arrested" if run today, alongside a specific brother's real-time emotional reaction (crying with laughter, having missed the register while living abroad) as a corroborating data point that the underlying material genuinely was and is funny, not merely tasteless.6 The genuine tension: the source's own framework doesn't fully resolve how to handle the case where a large volume of cheap, low-cost complaints turns out to represent an emerging real shift in what a broad audience finds acceptable — cheap signals can be individually unreliable while still being collectively informative about a genuine, if unevenly distributed, change in sentiment. The costly-signal test tells you which individual complaints to weight less; it doesn't tell you when to update your overall read of an audience even while discounting most of the individual data points.
Sutherland's own analysis here converges tightly with his separate argument about expensive versus cheap advertising media — the same underlying principle (cost filters for sincerity, cheapness enables performance untethered from substance) explains both why an expensive ad campaign signals genuine confidence in a product and why cheap digital complaint signals nothing reliable about genuine offense. He doesn't explicitly name this as one unified principle applied twice, but the structural identity is exact: in both cases, what something costs to produce is doing more diagnostic work than what it actually says.
Psychology — Ingroup/Outgroup Tribalism Mechanism. Confected outrage frequently functions less as a genuine reaction to content and more as a tribal-coherence display — publicly performing distress signals loyalty to a group's values to that group, independent of whether the underlying content actually caused the performer any real distress. The virtue-signaling spiral the source separately documents (the AC Grayling/Brexit case, where sustained public outrage persisted for years over something the same person showed no comparable engagement with before it became identity-coded) is best explained as tribal signaling running on the outrage-performance channel rather than as accumulated genuine grievance. The insight the pairing produces: distinguishing genuine offense from confected outrage isn't just a costly-signal test, it's also a test of whether the reaction is actually about the content or about maintaining visible standing within a group that has adopted a position on the content — and the two explanations aren't mutually exclusive, they compound.
History — Hierarchy and Status Signaling. Costly vs. cheap signaling as a sincerity-filter isn't a digital-age discovery — it's a recurring structural feature of status and commitment signaling across historical contexts, wherever the cost of a signal is what makes it trustworthy rather than its content. Reading this page's mechanism against a longer historical record of costly-signaling-as-commitment-proof shows the internet didn't invent cheap, unreliable signaling — it just removed a cost floor that had previously existed almost everywhere, in almost every era, as an accidental side effect of communication being expensive by default. The insight neither domain states alone: what looks like a new problem of digital-age performative outrage is actually an old signaling-reliability problem, newly exposed at scale because the historical accident that used to suppress it (expensive communication) has been removed.
Sharpest implication: the reliable test for whether outrage is real was never in the content of the complaint — it was always in what the complaint cost the person making it, and cheap media didn't create insincere people, it just removed the cost floor that used to filter them out automatically.
Generative questions: