“In this age of machine learning, it’s not algorithms, it’s data,” David Karger, a professor and computer scientist at MIT. Karger says that Musk can improve Twitter by making the platform more open, so that others can build on it in new ways. “What makes Twitter important is not the algorithms,” he said. “Those are the people who are tweeting.”
A deeper picture of how Twitter works also means opening up to more than just handwritten algorithms. “The code is fine; better data; code and data combined into a model may be best,” Alex Engler, a fellow in governance studies at the Brookings Institution who studies the impact of AI on society. Engler added that understanding the decision-making processes that Twitter’s algorithms are trained to perform will also be important.
The machine learning models Twitter uses are still only part of the picture, because the entire system also reacts to user behavior in real time in complex ways. If a user is particularly interested in a certain news story, then related tweets will naturally be amplified. “Twitter is a social engineering system,” said a second Twitter source. “It responds to human behavior.”
This fact has been illustrated by Research that Twitter has published in December 2021 showed that right-leaning articles received more amplification than left-leaning articles, although the motivation behind this phenomenon is unclear.
“That’s why we test,” says Ethan Zuckerman, a professor at the University of Massachusetts Amherst who teaches public policy, communication, and information. “Even the creators of these tools end up discovering surprising omissions and flaws.”
Zuckerman says that one irony of Musk’s alleged motives for acquiring Twitter is that the company has been very transparent about how its algorithm works by the end of the year. In August 2021, Twitter has launched a contest gave outside researchers access to an image cropping algorithm that exhibited biased behavior. According to people with knowledge of the work, the company is also working to find ways to give users greater control over the algorithms that display content.
Releasing some Twitter code will provide greater transparency, say Damon McCoyan associate professor at New York University who studies the security and privacy of large, complex systems including social networks, but even those who built Twitter may not fully understand how its mode of operation.
One concern for Twitter’s engineering team is that amid all this complexity, some code can be taken out of context and flagged as a sign of bias. Revealing too much about how Twitter’s recommendation system works can also lead to security problems. Access to the recommendation system will make it easier to play the system and stand out. It is also possible to exploit machine learning algorithms in ways that can be subtle and difficult to detect. “The bad guys are now probing the system and testing it,” McCoy said. Approaching Twitter’s models “can also help outsiders understand some of the principles used to enhance some content over others.”
On April 18, as Musk was escalating efforts to acquire Twitter, someone with access to Twitter’s Github, where the company had released some of its code, created an archive. called “algorithms”—perhaps a developer dug into the idea that the company could easily reveal the details of how it works. As soon as Musk’s acquisition was announced, it disappeared.
Additional reporting by Tom Simonite.
Stories with WIRED are more amazing
