But Twitter hasn’t really explained how it’s doing that — and the bits of information it has offered are confusing.
Let’s try to clear it up.
Right now, when Twitter inserts a tweet into your timeline from someone you don’t follow, it offers a short header that tries to explain why it’s there. For instance, you might see a tweet from a stranger, along with a label saying that a friend you follow on Twitter favorited the tweet, or follows the account that generated the tweet.
So you might think — reasonably — that you’re seeing that tweet because of your friends’ actions.
Sources familiar with the update say Twitter’s algorithm searches for popular tweets in your “social graph,” marked by actions like favorites and retweets. But it’s not showing you those tweets specifically because of one of your pal’s actions — it’s showing it to you because lots of your pals have shown interest in that tweet.
This means that every tweet you favorite won’t appear in someone else’s feed — it would require engagement from a number of other people, too.
Still, Twitter’s move might end up changing the way users treat the fave. Even if 50 other people favorite a tweet that Twitter surfaces, if your name is the one included in the label it certainly looks and feels like an endorsement, regardless of your intention.
That’s one of the reasons people familiar with the algorithm say that Twitter is thinking about alternate, and hopefully less confusing, ways to indicate why you’re seeing one of the tweets it found for you.
This new experiment is similar to others Twitter has introduced. MagicRecs, a Twitter tool created in May 2013, sends alerts to users when a bunch of people in their network favorite a tweet or follow someone new. Essentially, it’s just another way for users to discover new content.
The new algorithm, however, isn’t part of MagicRecs. In fact, it’s a completely separate team. Director of Product Management Trevor O’Brien, who’s leading the new charge, was still with YouTube when MagicRecs started, these sources said. Now, it will be up to O’Brien and his team to get users on board with Twitter’s new algorithm — and help them understand exactly how it works.
This article originally appeared on Recode.net.