Facebook’s own numbers put the engagement value of anger at roughly six times that of other reactions. That’s a plain measurement of what people do, and once it’s true, nearly everything else about how these platforms behave follows from it automatically.
You don’t need a conspiracy. You need a system that maximizes engagement and a fact about human beings, and the rest arrives on its own.
I find that number more disturbing than any scandal the platforms have had. A scandal gets investigated and apologized for. A ratio built into the business model keeps working every hour on everyone who opens the app, and the people paying for it are users who thought they were just checking the news.
Does anyone have to intend engagement-driven anger?
No, and that’s the uncomfortable part.
An optimizer pointed at engagement doesn’t know what anger is.
It knows that certain content produces more of the measured thing, and it shows more of that content. Nobody sits in a room and decides to make the population furious. They decide to maximize time on the platform, a reasonable commercial objective, and outrage is what the optimization finds.
Indifference is worse than malice here, because someone changing their mind can stop malice. An optimizer doesn’t have a mind to change. It’s an objective function, and it’ll pursue that objective wherever it leads until somebody alters the function, which nobody has a commercial reason to do.
What did Twitter’s 280-character limit really do?
It removed the room in which qualification lives. A limit that short can’t hold a caveat, a concession, or an acknowledgement that the other position has a point. So people stopped writing them. Then, once nobody wrote them, nobody expected them. Then a version of public argument emerged in which including them reads as weakness.
The constraint did more than shape individual messages. It selected for the people who do well inside it, and those people became the visible ones, and the visible ones set the tone for everybody else. That’s a considerable amount of cultural damage from a limit that started out sized to fit inside a text message.
Each step follows from the last: a limit too short for caveats ends with a public argument that treats nuance as weakness.
I miss the long arguments, the ones where somebody conceded a point and nobody treated it as a defeat. I don’t blame the people who stopped making concessions. The format punished them for it, and the platform profited from the fight that replaced them.
Why do people keep using outrage-driven platforms?
The real center of the book is about us, and I didn’t want to write it. We saw the warning signs. Facebook was making us depressed and the research said so. Twitter was making us angry and we could feel it happening. TikTok was doing something to attention spans and nobody seriously disputed it. We kept using them anyway.
Knowing turns out to have almost no effect on behavior when the reward is immediate and the cost is spread thinly across months. That’s a fact about people, not about social media, and these products are extremely well designed to exploit it.
Why twelve internet sins instead of seven?
Because the traditional seven didn’t cover it and the structure was too useful to abandon.
Wrath is the rage machine. Greed is the attention economy. Envy is the comparison trap. Pride is the echo chamber. Gluttony is information overconsumption, and sloth is the death of effort.
Then there are the ones the internet added.
Deception, in an era where you can’t establish whether an image is real. Vanity, as the performance of a self. Tribalism, as digital division. Amnesia, because the memory hole swallows anything older than a fortnight. And apathy, the illusion that reacting to something constitutes doing something about it.
Handing the cleanup to the people you harmed is a cowardly way to run a business.
That last one is the most modern and the most corrosive. A great deal of what feels like civic participation is a reaction being harvested.
Can you use engagement research safely?
Partially, by changing what you’re optimizing for. I think of it as harm reduction, and I won’t pretend it’s a cure.
Posting is meaningfully different from scrolling. Choosing who you follow deliberately is different from accepting whatever the feed offers. Noticing the moment the feed extracts a reaction from you is a skill and you can practice it.
None of that fixes the system. It just means you’re running a slightly different experiment inside it. The fault here belongs to the companies. They measured what anger was worth, kept the number, built the feed around it, and then told users to manage their own screen time. Handing the cleanup to the people you harmed is a cowardly way to run a business.
| Habit | Why it helps |
|---|---|
| Notice anger a machine chose for you | Spotting the moment the feed extracts a reaction is a skill you can practice |
| Follow deliberately | An unmanaged feed is somebody else’s optimization running on your attention |
| Post more than you scroll | Production and consumption produce different outcomes on the same platform |
| Treat reacting as neutral, not action | It’s the least effective thing available and feels like the most |
| Watch for qualification disappearing | The format trains you to drop caveats and concessions without noticing |
The book on this: Sins of the Internet is twelve chapters, one per sin, on what the platforms were built to do and what they’ve done to the people using them.
My take
Put the blame on the companies. They measured what anger was worth, built the feed around it, and then told users to manage their own screen time. It’s a cowardly way to run a business. You can still cut the damage. Post more than you scroll, choose who you follow, and notice the moment the feed pulls a reaction out of you. It won’t fix the system, but it’s better than handing over your attention for free.
What can a writer do about engagement-driven anger?
Notice when you’re angry at something a machine chose for you.
Follow deliberately. An unmanaged feed is somebody else’s optimization running on your attention.
Post more than you scroll. Production and consumption produce completely different outcomes from the same platform.
Treat reacting as neutral instead of as action. It’s the least effective thing available and it feels like the most.
Watch for the qualification disappearing from your own writing. The format trains you and you won’t notice.
We ate the apple knowing perfectly well it was poisoned. That’s the sentence I built the book around, and I include myself in the we, because I was scrolling too. Willpower as the answer is a cop-out. We were up against products tuned by teams of engineers whose whole job was keeping us angry and scrolling, and blaming individual users for losing that fight lets the people who designed it off the hook.
Anger drives the engagement numbers, and somebody chose that. Sins of the Internet covers what the platforms did to us, from Facebook’s own research on angry reactions to the 280 characters that killed nuance.
Frequently Asked Questions
- Facebook knows Instagram is harmful to teen girls: WSJ, Al Jazeera
- Twitter expands 280-character limit to most languages, CNBC
- Twitter to test 280-character tweets, busting old limit, Fox Business (Reuters)
