The United States still hasn’t revealed the recipe for coca-cola, but at least we have the one for the hit tweets. As he had promised, Elon Musk has published an open source version of a piece of the Twitter algorithm, making it possible to finally understand what makes a tweet work, or an account gain influence.
Everyone is looking for Tweep Credibility
The published code has been analyzed by several experts. And a few surprises about how it works come to light. We learn for example that there is a “Tweepcred”, a score of 0 and 100 which makes it possible to classify users and apply certain formulas to their publications. A tweepcred derived from Google’s famous PageRank algorithm, which is used to rank web pages.
Twitter thus calculates a reputation score for a user based on factors such as account age, number of followers, and device usage. And first surprise, Twitter also calculates a ratio between the number of subscribers of an account and the number of subscriptions (the number of users it follows). Note also that above 65 of tweepcred, you can publish as many tweets as you wish, and the algorithm will consider them all. Below 65, only 3 will be taken into account.
Each interaction multiplies the reach of a tweet
If you use Twitter, you probably know that retweets and likes make the tweet visible on other people’s newsfeeds. According to Aakash Gupta, each retweet offers a x20 boost compared to a reply, and each like a x30 boost. Adding a media, image or video multiplies the visibility by 2, which seemed already understood by most users.
For his part, Steven Tey, another engineer who dissected extracts of the code, notes that a retweet multiplies by 2 the possibility that a tweet is suggested (the like would have a reach of 1), but several interactions have an influence unexpected on the possibility that a tweet gives visibility to your account.
Thus, clicking on your tweet and staying there for more than 2 minutes increases its reach by 22. If users click on your profile via your tweet and like or reply to another tweet, the reach is x24. Replying to your tweet acts as a 54x multiplier. Finally, if you answer this answer yourself, the scope is x150. As a result, a user has every incentive to reply to everyone to continue fueling the success of a tweet.
Obviously, some interactions are conversely negative for the success of a tweet. This is the case if a user asks to “post less often” on your tweets, blocks or mutes you (-74x), or worse, flags your tweet (-738x).