This promise Elon Musk he kept it: he published one part of the Twitter source codea platform he has owned for a few months, on GitHub, portal dedicated to the development of open-source applications. This move could have a huge impact for Twitter and the social media world at large. Indeed, it becomes possible for anyone – from influencers to hackers who create bots – to know how to make a tweet successful. Among the first pieces of information, in fact, there are the values that receiving a like or a retweet have to bring your post to the top of the timeline.
Elon Musk has released the Twitter source code: here’s what emerges
Twitter’s source code is the heart of the platform. It is the mechanism that keeps the social network functioning. In particular, it is the software that select, sort and filter tweets to be shown to users in the “For you“. Relying on various factors like the preferences, interactions, trends and relevance of content. This algorithm has a big impact on the information we receive, the opinions we form and the conversations we initiate.
So far, Twitter’s source code was a business secret, like that of other social networks like Facebook, Instagram or TikTok. This means that we didn’t know exactly how it worked, what criteria it used and what effects it had. To understand the success of a tweet, it was necessary to make attempts and collect the results: this could generate mistrust on the functioning of the platform’s suggestions.
Elon Musk has decided to change that. Since he bought Twitter at the end of October 2022, in a daring $44 billion deal, promised to make the algorithm open source. Already in March 2022 he had polled his followers on the idea, writing: “The Twitter algorithm should be open source“. When he wasn’t the owner of the platform yet. And on March 31, 2023, he kept his word and posted two repositories on GitHub, chiamati “the-algorithm” e “the-algorithm-ml”, containing large portions of the Twitter source code.
What Twitter Source Code Analysis Reveals
Analyzing the source code of Twitter is a task for experts. Those who know the computer language used in its parts can understand what the individual commands do, but understanding their effect is an operation for those who understand it. Like Aakash Gupta, ex-executive of tech companies like Epic Games and Google. And that he posted what the Twitter source code reveals about his profile on social media and in more detail on Substack.
An analysis of Twitter’s code revealed how theThe social network algorithm evaluates several factors to determine the visibility of a tweet. Among these, the “I like it” seem to be the most influential factor. Indeed, I am able to multiply the reach of a post up to 30 times. Sharing follows, which has an effect 20 times greaterand the answers, which have a minimal impact.
This implies that Twitter favors likes over replies. In other words, a content that gets a lot of likes is promoted much more than a content that elicits a lot of responses. A much-discussed post doesn’t weigh as much as a highly-appreciated tweet.
The impact of images and external links
Also adding pictures is helpful, acting as a multiplier for the recommendations on the For You page. But much less than likes and shares.
Instead, insert a link to an external site, such as an article or video on YouTube, is often penalized as “spam”. Unless the profile does not already have a good popularity: In this case, the penalty is mitigated. This should allow blocking advertising spam, but ensure that newspaper articles are shared by the newspapers that publish them – including the New York Times, even if it has now lost its golden official account check.
The impact of reporting – and of Twitter Blue
Other aspects that emerged concern the reports of abuse or spam or being “mutated” from another user are negative factors, which therefore reduce the reach of a tweet. Something we expected to work like this.
Less obvious (and bound to arouse discussion) the pay for Twitter Blue subscription. In fact, doing so increases the reach of who you will reach with your tweets: an incentive to pay for the service.
Insteadmisinformation charges carry a penalty. But also the posts that dealing with Ukraine they are disadvantaged by the Twitter algorithm, regardless of how they treat the topic.
Twitter source code and profile clusters
Another interesting element emerged from the analysis of the source code of Twitter groups each user into a cluster, that is, a larger set of similar profiles. A sort of bubble, based on the profile. This means that content consistent with the group in which Twitter has placed a profile favors its reach. In reverse, less appropriate content greatly decreases the reach of a tweet.
In other words, if Twitter recognizes that you often post about technology, for example, a post about your holidays away from PCs and smartphones may not have the success you’re used to.
But also the ratio between the number of people followed and the number of followers it has its importance.
What does the publication of social algorithms mean
The publication of the Twitter source code could have several impacts on the social network managed by Elon Musk. The first and most immediate concerns the discussion about the software. Musk can get his social network talked about again in a positive way, after the many problems of the last perioddemonstrating that it has kept its word on the source code – ignoring those that are still outstanding.
However, the impact on public opinion remains to be assessed not so much of the gesture, how much of the content of the algorithm. This first analysis brings out the way in which the social network manages what it presents before our eyes. And although the first findings are more interesting in cave marketing, the fact that, for example, it penalizes the discussion on Ukraine will certainly cause debate among various experts.
And in the next few hours the analyzes will multiply: it remains to be seen whether they will be positive or negative. We will keep you informed about it.