These Researchers Say They Can Predict Which Tweets Will Go Viral

You can even put your own tweets through their viral-predicting algorithm.

These Researchers Say They Can Predict Which Tweets Will Go Viral
[Photo: Flickr user Ralph Hockens]

With 288 million active users sending around 500 million tweets each day, Twitter offers plenty of opportunities to grab attention.

Sometimes that attention is positive, as it was when semi-employed comedian Justin Halpern started the account Shit My Dad Says and found himself on the receiving end of book deals, mentions on The Daily Show, and a TV show starring William Shatner. Sometimes it’s negative, as it was for director of corporate communications Justine Sacco, who made an inappropriate joke about HIV to her 170 Twitter followers, and wound up briefly becoming one of the most hated women in the world.

What both of these examples have in common is that a specific tweet or series of tweets passed from person to person like a contagion. But what makes a tweet go viral?

That was the question asked by researchers at Cornell University, who think they’ve solved it by identifying the specific features necessary for a tweet to become popular. In short, a viral tweet typically has nine specific features. It will ask people to share it, using words like “please” or “retweet.” It will also be informative, use language familiar to the community it’s addressing, be phrased in the style of a newspaper headline, use words that appear in other retweeted messages, express a strong positive or negative reaction, use third-person pronouns, use generalizations, and be easy to read.

While that may sound like broad advice, the researchers combined all of these insights into an algorithm that you can check out on their website–letting you compose two different variations on the same message to see which is more likely to go viral. It may not exactly be a writing guide on the level of the classic The Elements of Style, but as a guide for getting noticed online here in 2015, it’s certainly not bad.

[via Cornell University]