--- title: "Facebook’s new feature: your top 7 friends" description: "Facebook tests new top 7 friends feature to combat feed relevance issues. Analysis of how machine learning and user signals reshape social graph optimization." keywords: ["Facebook","top friends feature","social media relevance","machine learning algorithms","EdgeRank","friend graph optimization","user engagement","social network updates"] ai_summary: "Facebook introduces a top 7 friends feature to enhance feed relevance using machine learning and user signals." date: 2012-06-19 lastmod: 2026-07-31 canonical_url: https://www.tomtunguz.com/facebooks-new-feature-your-top-7-friends/ author: "Tomasz Tunguz" ---

Facebook is asking me to redefine my closest 7 friends.

I logged into Facebook today to see this new feature, starring your friends. Below this text is a Google circles like UI that asks me to star friends whose updates are particularly important to me.

![image](https://res.cloudinary.com/dzawgnnlr/image/upload/q_auto/f_auto/w_auto/image_260_25876274)

With hundreds of friends, the average Facebook user logs in to see wall of updates that aren’t quite relevant. Some of us have reduced our friend graphs and unfriended contacts in order to return the feed to relevance, which isn’t a desired user behavior. Facebook, after all, wants to encourage expansive graphs.

The challenge that feeds everywhere face is the local minimum problem. Machine learning algorithms like EdgeRank can easily fall into local minimums, where the same people appear on the feed because they’ve been appearing on the feed. The people in the feed become stale and either the model has to be rebuilt or a strong signal must be injected into the algorithm to change results.

This beta feature might indicate Facebook is looking for another strong signal of friend-closeness, to recenter the feed on the people a user most cares about and return relevance to the feed.

Or it might indicate a push to build new features around a top 7 friend list.