“In this age of machine learning, it's not algorithms, it's data,” said David Carger, a MIT professor and computer scientist. Karger said Mask could improve Twitter by making the platform more open, so that others could build on it in new ways. “What makes Twitter important is not the algorithm,” he said. “It's people who are tweeting.”
A deeper picture of how Twitter works means more openness than handwritten algorithms. “The code is OK; The information is good; Code and data combined with a model can be great, “said Alex Engler, a fellow at the Brookings Institution's Governance Study who studies the effects of AI on society.
The machine learning models that Twitter uses are still part of the picture, as the whole system reacts in a complex way to real-time user behavior. If users are particularly interested in a particular news item, related tweets will naturally spread. “Twitter is a socio-technological system,” says a second Twitter source “It's responsive to human behavior.”
This phenomenon was illustrated by research published on Twitter in December 2021 which showed that right-leaning posts received more expansion than left-leaning posts, although the dynamics behind this phenomenon were unclear.
“That's why we audit,” said Ethan Zuckerman, a professor at Amherst University in Massachusetts who teaches public policy, communications and information. “Even those who make these tools discover amazing flaws and errors.”
A mockery of Musk's alleged motives for acquiring Twitter, Zuckerman said, the company was extraordinarily transparent about the way its algorithms worked late. In August 2021, Twitter launched a contest that gave outside researchers access to an image-cropping algorithm that exhibited biased behavior. The company is also working on ways to give users more control over the algorithms that they have at work.
Revealing some Twitter code would bring more clarity, says Damon McCoy, an associate professor at New York University who studies the security and privacy of large, complex systems, including social networks, but who created Twitter doesn't fully understand how it works. .
One concern for Twitter's engineering team is that, in all of this complexity, some of the code may be taken out of context and highlighted as a sign of bias. Excessive revelations about how Twitter's recommendation system works can lead to security issues. Access to a recommended system will make it easier to gamble and gain prominence. It may also be possible to use machine learning algorithms in a way that is subtle and difficult to detect. “Bad actors are investigating and testing the system right now,” McCoy said. Access to Twitter models “Some of the principles used to promote some content over others may help outsiders understand.”
On April 18, while Musk was stepping up his efforts to acquire Twitter, someone had access to Twitter's Github, where the company had already released some of its code, creating a new repository called “Algorithm” সম্ভবত perhaps a developer's idea that the company It can easily reveal the details of how it works. Shortly after the announcement of Musk's acquisition, it disappeared.
Additional reporting by Tom Simonite.
More great cable story