Twitter Takes Its Recommendation Algorithm ‘Open-Source’. What’s Covered In The Code?

twitter algorithm open source

Need content for your business? Find top writers on WriterAccess!

If you’re a digital marketer who’s always wished you could take an inside peek into Twitter’s recommendation algorithm, you’re in luck. Two different Twitter source code repositories just landed on GitHub, and marketers have wasted no time combing through them in search of answers.

But how much were eagle-eyed marketing pros able to learn from the repositories, and what do the findings mean for those in the digital marketing sector moving forward? Let’s take a closer look at what you need to know,

What’s Covered in the Code?

Along with the GitHub drop, Twitter also released a blog post further explaining a few things, including some of the details revealed within the source code.

Although many people would like to know about Twitter’s algorithm and how it functions, the release focuses primarily on the “For You” feed and how the recommended tweets you see are chosen. According to Twitter, the recommendation process consists of three different stages:

  • Candidate Sourcing: In which large batches of high-value tweets are selected via various recommendation channels
  • Ranking: A machine learning process that sorts tweets according to additional perceived value
  • Filtering: The portion of the process where non-relevant tweets are removed from the feed (e.g., tweets you’ve seen previously or content that may be NSFW)

CEO Elon Musk first informed users of his intentions to release this information a while ago – March 2022, before he purchased Twitter. Now that he’s made good on his promise, he has since tweeted that “most” of the recommendation algorithm is included. He further expressed his desire to release the rest at a future date.

What Other Factors Affect a Tweet’s Visibility?

Although users have long suspected certain factors contribute to a tweet’s eventual reach, there was little hard proof to point to. However, Twitter itself has now confirmed that the following examples (along with others) are part of the mix.

  • Whether the tweet fits your personal preferences, as determined by the algorithm
  • Whether the tweet comes from inside or outside your network (as Twitter shoots for a balanced mix between the two)
  • Whether people you follow or engage with regularly have also engaged with the tweet
  • Whether the tweet was created by someone only a few degrees removed from someone you know
  • Whether you’ve recently viewed other tweets by the author
  • Whether the tweet has received negative feedback from other users
  • Whether the tweet is associated with a trending topic or community

These are in addition to well-known factors that boost visibility, like the number of likes, retweets, or comments a particular tweet receives.

It’s also worth noting that as paid verification options like Twitter Blue become more accessible and relevant across social media, whether a user pays into the pool will also affect visibility moving forward. On Twitter, this is especially the case after April 15th, as the “For You” feed will consist only of paying users past that date.

However, whether this upcoming move will actually drive more users to subscribe to Twitter Blue remains to be seen. At present, fewer than 1 percent of active Twitter users are also subscribers.

How Much Weight Do Different Factors Carry?

Although Twitter decided against officially releasing model weights as they relate to the algorithm, independent coders have managed to uncover the data regardless. Take a look at this tweet from Peter Yang for more information.

Naturally, the most critical boosts come from likes and retweets, at 30x and 20x, respectively. But being a Twitter Blue subscriber still gives a user 2-4x the visibility juice, which is nothing to sneeze at. Whether the tweet is associated with a trusted circle, contains visuals like video and photos, or has replies come in just behind Twitter Blue.

Factors that negatively affect a tweet’s reach were also uncovered. They include not having any text and only containing a URL. Of course, blocks, associated unfollows, and mutes are also negatives.

What Does This Mean for Digital Marketers?

The world of digital marketing is already in the process of adjusting to all of Twitter’s ongoing changes. However, this latest insight into how the algorithm works should be helpful moving forward.

Marketers already know that brand voice, use of visuals, buzz, and timeliness are essential on Twitter. Now they’re even more aware of how important it is to format tweets correctly and leverage visuals to expand reach. Brands may also want to consider authenticating their presence on the platform with a Twitter Blue subscription to help maximize reach.

However, the most effective way to succeed on Twitter remains to produce a consistent stream of high-quality content that engages audiences and is in step with current trends. The smarter the algorithm becomes, the more important quality will be.

Do you want to continue to be updated with Marketing best practices? I strongly suggest that you subscribe to The Beat, Rock Content’s interactive newsletter. There, you’ll find all the trends that matter in the Digital Marketing landscape. See you there!


Human Crafted Content

Find top content freelancers on WriterAccess.

Human Crafted Content

Find top content freelancers on WriterAccess.

Vanessa Dias Rock author vector
Content Specialist

Subscribe to our blog

Sign up to receive Rock Content blog posts

Rock Content WriterAccess - Start a Free Trial

Order badass content with WriterAccess. Just as we do.

Find +15,000 skilled freelance writers, editors, content strategists, translators, designers and more for hire.

Want to receive more brilliant content like this for free?

Sign up to receive our content by email and be a member of the Rock Content Community!

Talk to an expert and enhance your company’s marketing results.

Rock Content offers solutions for producing high-quality content, increasing organic traffic, building interactive experiences, and improving conversions that will transform the outcomes of your company or agency. Let’s talk.