Answer The Hate

SOC · Answer online

Making something capable of spreading

Virality cannot be manufactured. What can be controlled is whether a thing is built to travel, and almost everything that travels shares a small number of properties.

Any countryAn evening8 min read

A dancer sweeps two large silk fans through the air in the middle of a London street parade, flags and onlookers on both sides.
Amanda Rose

The short version

  1. Nobody can make a thing go viral on demand. You can make it capable of spreading, which is a different and learnable job.
  2. People do not share things because of the thing. They share because of what passing it on says about them.
  3. Anger travels fastest and costs you the room, because it recruits people who already agree and repels everybody else.
  4. Joining a conversation that already has attention is far cheaper than starting one cold, and hardly anybody does it deliberately.
  5. Something posted to no audience almost never travels. Ask eight to fifteen specific people directly instead of posting and hoping.
What you need3 items
  • A note listing eight to fifteen people you could message directly, written before you need it
  • Two or three finished posts held back and ready, for a week when the subject is already moving
  • One specific thing you would ask people to do if something of yours travelled

Anybody selling you virality on demand is selling you something. What can actually be controlled is whether a thing is built to travel, and that is a learnable job needing nobody’s permission and no existing audience to start.

Before you aim at reach. A thing that travels carries your name to people who disagree with you, in numbers. That section is near the bottom, and it is the one to read first if this has already started.

The four properties

Almost everything shared has these four. Everything that dies lacks one.

One idea. One thing somebody could repeat accurately to another person the next day. Anything carrying two ideas is usually two posts, and travels as neither.

Legible without effort. Somebody moving fast can tell what it is without clicking and without reading to the end. The point lives at the front.

A reason to pass it on. Being true is not a reason, and neither is being important. The sharer needs something out of it. This explains what I kept failing to explain. This is what I wanted to send my brother and could not find.

It makes the sharer look like the person they want to be. This does more work than the other three.

Sharing is about the sharer

Every share is a small public statement about the person making it: this is what I care about, this is the standard I keep. So the question when you write is what passing it on says about whoever passes it on.

Weak: “These people are disgusting and anybody defending them is complicit.”

Strong: “This has been reported wrongly in four places today. Here is the original document, and here is the line everybody is missing.”

Sharing the first makes somebody look angry. Sharing the second makes them look careful. Plenty of people who agree with the first will not be seen holding it in front of an employer, so they read it, agree, and scroll past.

Anger, and what works better

Anger travels, and the people who first measured how far have since revised it down. The original figure was about 20 per cent for every moral-emotional word. The same group’s meta-analysis, across 27 studies and 4.8 million messages, now puts it at 13 per cent, and one of their own datasets came out the other way.

What held up better is uglier. Across 2.7 million posts, the strongest predictor of sharing was whether you were talking about them. Posts about the political out-group were shared roughly twice as often, and how strongly the writer felt came a distant second.

So the lift is real, small, and bought in the worst currency available to us. It buys reach among people who already agree, who were never the audience capable of changing, and it does not cross. The number goes up while the room goes.

Surprise travels, and it travels because it is new rather than because it is hot. A checkable fact most people had wrong goes further than any opinion about it. So does dignity, somebody behaving well while being treated badly. So does ordinary experience outsiders have never been shown.

The first line does nearly all the work

Most of what you lose, you lose at the opening. On video, the one platform that publishes a figure asks for something interesting inside five seconds. On text it is the first line. Assume the reader is moving, distracted, and looking for a reason to keep moving. The first line’s job is to stop them, not to summarise and not to set up context.

What stops people: a specific detail before any explanation of it, or a flat contradiction of something they thought they knew. What does not: “I have been thinking a lot about.” A wind-up is where readers leave.

Write the post, then delete your first sentence. It is usually a run-up, and the second is the real opening.

Whatever travels is complete where it sits. The passage goes in the post and the link underneath, for verification rather than comprehension.

One person, one specific thing

Scale does not travel. A particular does.

Nobody can hold a six-digit figure, and everybody can hold one person, on one day, doing one thing. Find the single case inside the number, lead with it, put the number underneath. For an institution, that is the one email unanswered for eleven weeks, quoted, with the date on it.

Joining beats starting

Starting cold is rare and mostly luck. Joining something already moving is achievable, and hardly anybody does it on purpose. The audience is already there and hunting for material, so supplying the missing piece costs a fraction of manufacturing attention. Keep two or three finished things ready for it.

On a moving story, early reports are wrong in both directions, and being early and wrong with your name on it is the version people keep. Know what the hour costs, though. Across 126,000 stories and 4.5 million tweets, the truth took roughly six times as long as a falsehood to reach 1,500 people. A later reanalysis puts that down to false stories having bigger audiences rather than a different route, which changes the explanation and not the head start. Have the thing already written. Do not type faster.

The seeding problem

Something posted to no audience almost never travels, however good it is. Posting and hoping is where most of this work is wasted. Ask specific people instead, privately, one at a time. Eight to fifteen, chosen because they reach people you do not.

Building the account that gives a post somewhere to start is the other half of this, and it is a separate job: how a small account gets seen.

Say what it is in one line, say why you thought of them, and say that no is completely fine. People agree far more often than campaigners expect.

The asks, word for word

“I’ve written something about the meeting on Tuesday. You reach people I don’t, and I thought of you.”

“One line and then I’ll leave you alone: this is the bit everyone’s got wrong, and here’s the document it’s in.”

“No is completely fine, and I’d rather have a no than have you share it and feel awkward.”

One firm boundary, and it is on you. This is asking people to amplify something you made. It never means organising a group to descend on one person, report one account together, or fill one inbox. That is harassment whichever way it runs.

“I’m not organising anything at a person. If that’s what this is turning into, count me out.”

If it does spread

Read this before you post.

Much of the new attention will be hostile, because the post has left the people who agree with you. Read the guide on answering hostile replies now, rather than at midnight with your chest tight.

Decide beforehand what you want people to do, and the gap here has been measured. Thousands of people were followed for four months while they followed political accounts. What they believed moved. What they did did not: registering, turning out, attending anything, none of it shifted at all.

So the ask goes underneath, small, dated and doable from a phone. That is the difference between a loud week and a result.

Deleting, and going again

Take a copy first: what you delete is already in other people’s screenshots, and yours is the copy that disappears. Delete when the post is factually wrong, when it identifies a private person, or when it carries a sentence you cannot defend alone. None of that is a failure of nerve.

A post that did nothing is often a good post opened with the wrong sentence. Something you believe in is worth two or three more attempts across a month, with a new first line each time. Reposting identical text reads as spam.

The honest measure

Treating a small number as failure is how people talk themselves out of the work. Two hundred of the right people beats fifty thousand of the wrong ones.

Count these instead. Somebody you never met used your explanation in their own argument. A journalist or a student officer got in touch. One person changed what they were going to do next.

Before your next post

Fill in the blanks106 words

Two questions from memory. What are the four properties, and which does your last post lack? And what do you do to a finished post before publishing it?

Then set up. List eight to fifteen people by name, and write the one thing you would ask them to do if something of yours travelled.

If I have written something I think matters, then before posting I will delete my first sentence and check the point still lands with no link.

If something of mine starts moving and the replies turn hostile, then I will stop replying, screenshot everything, and open the hostile replies guide before typing.

Sources7 cited

Sources

Where this page relies on somebody else, here they are, so you can check rather than take our word for it.

  1. Emotion shapes the diffusion of moralized content in social networks, 563,312 messages on gun control, same-sex marriage and climate change; each moral-emotional word raised diffusion by about 20 per cent, more within liberal and conservative networks than between them. The same authors' 2025 meta-analysis across 27 studies and 4,821,006 messages revises the figure to 13 per cent, 95 per cent confidence interval 6 to 20 per cent
    Proceedings of the National Academy of Sciences · 26 June 2017
  2. The spread of true and false news online, About 126,000 stories tweeted 4.5 million times: "It took the truth about six times as long as falsehood to reach 1500 people"
    Science · 9 March 2018
  3. Estimating the effect size of moral contagion in online networks, a pre-registered replication and meta-analysis, Meta-analysis of 27 studies across 5 laboratories, 4,821,006 observations, incidence rate ratio 1.13, 95 per cent confidence interval 1.06 to 1.20; in the authors' own five-topic replication one dataset returned 0.97
    PNAS Nexus · 14 October 2025
  4. Out-group animosity drives engagement on social media, 2,730,215 posts; posts about the political out-group were shared or retweeted about twice as often, and out-group words were the strongest predictor, with moral-emotional words comparatively weak
    Proceedings of the National Academy of Sciences · 23 June 2021
  5. How social media creators shape mass politics, a field experiment during the 2024 US elections, Pre-registered; Americans aged 18 to 45 following creators August to December 2024. Policy positions and vote choice moved by 0.1 to 0.2 standard deviations, and there was "no evidence that SMCs cultivated offline participation"
    arXiv preprint, Columbia University · 17 December 2025
  6. Comparing information diffusion mechanisms by matching on cascade size, Volume 118, number 46, article e2100786118. Reanalysis of the false-news data: the reported structural differences "can almost entirely be explained by false-news cascades being larger"
    Proceedings of the National Academy of Sciences · 16 November 2021
  7. Elements of a TikTok video, Page displays "Last updated: Aug 18, 2026". Asks for "an attractive element early on [...] ideally within the first 5 seconds", and declines to prescribe a video length
    TikTok Creator Academy · 18 August 2026