Strong feelings travel faster
- Calm and dull1 share per 100 viewsFunny4 shares per 100 viewsAmazing6 shares per 100 viewsMakes people angry8 shares per 100 views
Made-up numbers, real order. Remember the weights: a share is worth more than a like, a comment more still. Now ask what makes people share and comment. Being amazed. Being angry. Being scared. A calm, true, dull post gets a like at most. A post that makes you angry gets a comment and a share.
- A post people react toMore shares and commentsShown to more people
Put that through the score. Shares are heavy. Angry posts get shares. So angry posts score high, and the feed shows them to more people, who share them again. Nobody at the app set out to spread anger. The weights reward reactions, and anger is a reaction.
- Makes people laughsame viewsMakes people furioussame views
Two posts get the same views. One makes people laugh, one makes them furious. Which does the score lift?
The funny one. Apps want people to be happy.
Not yet. Wanting is not in the score. The score counts shares, comments and time. Whichever post produces more of those wins, and happiness is not measured.Neither. Same views, same score.
Not yet. Views are one signal. Shares and comments weigh more, and the two posts will not get the same number of those.The furious one, if it gets more shares and comments.
Right. The score does not read the post. It counts what people did with it, and fury produces more comments and shares than most feelings. - Time for a true story to reach 1,500 people6 units of timeTime for a false story to reach the same1 unit of time
The false story took a sixth of the time. Real numbers. In 2018 researchers followed 126,000 stories on Twitter over ten years. False stories reached 1,500 people about six times faster than true ones. They were shared more, by people, not by robots. The likely reason: false stories were more surprising, and surprise gets a share.
Move the control to set how strongly a post makes people feel. The rule is made up: stronger feeling means more shares per hundred views, and each share reaches ten more people.
The first 100 viewers100 peoplePeople reached after three rounds111 peopleBarely1 share for every 100 views. Each share reaches ten more people. After three rounds of sharing, about 111 people have seen it.
The first 100 viewers100 peoplePeople reached after three rounds125 peopleA little2 shares for every 100 views. Each share reaches ten more people. After three rounds of sharing, about 125 people have seen it.
The first 100 viewers100 peoplePeople reached after three rounds162 peopleQuite4 shares for every 100 views. Each share reaches ten more people. After three rounds of sharing, about 162 people have seen it.
The first 100 viewers100 peoplePeople reached after three rounds253 peopleStrongly7 shares for every 100 views. Each share reaches ten more people. After three rounds of sharing, about 253 people have seen it.
The first 100 viewers100 peoplePeople reached after three rounds400 peopleVery strongly10 shares for every 100 views. Each share reaches ten more people. After three rounds of sharing, about 400 people have seen it.
Going from 'barely' to 'very strongly' multiplies the shares by ten. What does it do to the reach after three rounds?
About the same. Most people never share anything.
Not yet. Most do not, and a few shares are enough. One share in ten views, three rounds over, is what turns a hundred viewers into four hundred.Ten times. Ten times the shares is ten times the reach.
Not yet. The first hundred viewers are the same for both. Only the sharing multiplies, and after three rounds that is under four times. Give it ten rounds and it would be far more.Nearly four times, and the gap grows with every round.
Right. 111 becomes 400. Each round of sharing multiplies again, so a post shared ten times as much does not reach ten times as many at first, but it pulls away with every round.Move the control.
- One word like 'shameful'about 20 in 100 more sharesTwo such wordsmore again
Another study counted words. Posts about politics that used words of feeling and right-and-wrong, like 'disgusting' or 'shameful', were shared about 20 in 100 more for each such word. The app did not know what the words meant. It knew those posts got shares.
In that ten-year study, how much more likely was a false story to be shared than a true one?
Pick one. Nothing is scored.
A true story100 chance of being shared, true = 100A false story170 chance of being shared, true = 100About 70 in 100 more.Guesses run low because we assume people share what is true. False stories were about 70 in 100 more likely to be shared. The researchers checked whether robots did it and found that people did.Three posts go up on the same afternoon. Tap the one the score lifts highest.
Tap the part of the picture that answers it.
Ask which one makes people argue in the comments.
The council posts the new bin-day timetableA goal from last night's matchSomeone claims the mayor stole the bin moneyThe claim about the mayor
Right. The claim makes people angry, and angry people comment and share. True or not, the score only sees the shares.The goal
Not yet. A goal gets likes and some shares from fans. It is a good post, but it does not make anyone argue, and arguing is comments.The bin-day timetable
Not yet. A timetable is useful and gets saved, not shared. Useful is one of the quietest signals there is.- ShockA reactionA shareA heavy weightThe top of the feed
A video app notices that its most-shared clips are fights and near-misses. Nobody at the company chose that. Why did it happen?
Fights are what people upload most.
Not yet. Uploads are the pile. What rises out of the pile is the score's doing, and the score follows shares. A rare shocking clip can beat a thousand calm ones.Shock is a reaction, reactions are shares, and shares are weighted heavily. The score lifts what people react to.
Right. The company set weights that reward sharing. People share what shocks them. The clips that rise are the ones that fit both, and nobody had to choose them.Someone at the company likes fights.
Not yet. No person picks the top clips. The score does, from what viewers did. The taste on show is the crowd's reaction, not an employee's.
Lesson complete
Posts that make people react get shared more, so the score lifts them, whatever they are about.