Why Elon Musk Wants to ‘Open Source’ Twitter’s Algorithms
Why Elon Musk Wants to ‘Open Source’ Twitter’s Algorithms
(Bloomberg) -- Social media platforms have become some of our most important sources of news. What information gets seen by whom is shaped by tightly guarded algorithms that tech companies have spent billions of dollars to develop. People on all sides of the political spectrum have voiced concerns over the impact those algorithms have in terms of promoting divisions, misinformation and hate speech, among other things. Elon Musk, who has proposed a $44 billion takeover of Twitter Inc., has vowed to make its algorithms “open-source” to promote transparency and free speech. It’s not entirely clear what he has in mind, or what the consequences might be.
1. What is an algorithm?
A set of instructions for making a decision or performing a task. Arranging names in alphabetical order is a kind of algorithm; so is a recipe for making chocolate chip cookies. But those simple formulas bear only a distant relationship to the computerized code used by social media giants like Twitter, Facebook parent company Meta Platforms Inc. and Alphabet Inc.’s Google.
2. Why are they used?
The earliest iterations of social media platforms generally logged users’ posts and comments in reverse chronological order, with those published most recently showing up first on people’s feeds and timelines. That all changed in 2009, when Facebook began populating news-feeds with content that the algorithm determined individual users would be most interested in seeing. The rest of the platforms followed suit and started testing their own versions, citing user experience as their motivation. Engagement levels are higher when the “best” posts, as defined by the algorithm, are more easily found. This truth underpins the success of search engine giants such as Google, which rely entirely on algorithms to display quick, relevant results to queries. The change is credited with driving the platforms’ growth and has made them vastly more profitable, as advertisements are easier to sell to engaged readers than to bored scrollers.
3. How do they work?
Present day social media algorithms sort posts in a users’ feed based on a combination of signals that indicate relevancy, only one of which is how recently an item was published. The companies are able to use the vast caches of data they collect on individual and group behaviors to personalize recommendations. For example, the platforms might prioritize content from your closest friends and family if those are the accounts you interact with most often. Users who frequently interact with content of a specific political bent will likely be served more content judged to meet that preference. Each social media company prioritizes different criteria when determining which content is most relevant, although the details have remained secret. Facebook claims it prioritizes meaningful interactions to generate friendly conversations while Twitter says it prioritizes quality and relevancy to display the best content.
4. What does Musk want to change?
Musk has floated a number of notions in the name of upholding what he describes as free speech ideals, including a promise to take a minimalist approach to content restrictions. During a panel discussion at the Ted2022 conference on April 14, Musk said he would be “very reluctant to delete things” and “very cautious with permanent bans -- timeouts, I think, are better.” He’s talked about trying to “authenticate all real humans” as a means of differentiating between bots and legitimate accounts. And he has advocated for opening up to public scrutiny Twitter’s content algorithms, which identify spam or posts violating the site’s terms of service as well as ranking items for ordering user feeds, among other things.
5. What does that mean?
Most likely, “open sourcing” Twitter’s proprietary software would make some of its algorithms available for public inspection. Once others can read the code, they can use it for their own applications or could make suggestions to Twitter’s own developers for changes. In other industries, such as cryptography, nearly all relevant algorithms are open sourced so that entire communities of coders can collaborate on improving design and security features. What’s not clear in the social media context is how much transparency making an algorithm’s code available would add, as seeing the code doesn’t tell you how exactly it works in practice when operating in collaboration with the vast amounts of data that power it. And with these immensely complicated systems, it’s not clear how much seeing isolated parts of algorithmic code would reveal about how all the pieces work together.
6. What might it mean for the industry?
Would-be competitors might have a greater chance of getting a foothold in a market now dominated by giants because they’d be able to build upon the intellectual property of one of the biggest companies in the sector. That could eventually increase competition. New social media platforms struggle in particular with content moderation -- it’s hard to grow to a size at which you can afford moderators if potential users are turned off by unmoderated content. On the other hand, algorithms can be of limited use without the constant flood of new data that existing social media companies use to keep training and fine tuning their formulas.
7. Are there risks?
Yes, both for users and for social media platforms. Some experts warn that open-sourcing Twitter’s code may bring a number of problems. For example, it might give bad actors a greater understanding of how to game the system and could fuel the proliferation of bots on the platform. “When you put something out there in the open market and you make it available for everybody to use, it can be misused,” said Arun Kumar, chief data and marketing technology officer at ad company Interpublic Group of Cos. Recommendation algorithms are proprietary and form the core intellectual property of platforms like Twitter, Facebook and TikTok.
8. Is there a solution?
One option is a compromise that allows communities of users to have a role in the platform’s development while still safeguarding its commercially sensitive data. Jason Mars, computer science professor at the University of Michigan, said Twitter could establish an open-source version of its recommendation system and use it to inform the main algorithm that determines what users actually see.
The Reference Shelf
- Bloomberg Opinion’s Cathy O’Neil says open-sourcing Twitter’s algorithms will only create more problems
- A timeline of Musk’s journey from Twitter shareholder to its prospective owner
- QuickTakes on the controversy over Facebook’s algorithms and Twitter’s spam bots.
- An article in the MIT Technology Review about Musk’s plans for Twitter and open algorithms.
- Scientific American argues for more transparency on social media algorithms
- A 2021 study conducted by Twitter on whether its algorithms amplify political content.
©2022 Bloomberg L.P.