[research from December 2021]

General

Each time a user opens their twitter, the algorithm studies the tweets from the accounts a user follows and gives them a relevance score.

• The tweet itself: its recency(the reverse-chronological order is still there! Twitter is never fully automated and preserves its “live-feed” system), presence of media cards (image or video), and overall engagement (including retweets, clicks, favorites, and time spent reading it)
• The tweet’s author: user’s past interactions with this author, the strength of connection to them, and the origin of the relationship
• User: tweets a user found engaging in the past, how often and how heavily they use Twitter

Twitter timeline

• people have an option to have a chronological vs algorithmic timeline

Criteria

• How recent it is
• Use of rich media (images, gif, video)
• Engagement (high number of likes, replies, retweets)
• Prominence of author
• Relationship of user to author (has the user often engaged with that author?)
• User behavior

Don’ts