Elon Musk vows to make Twitter recommendations transparent

On Friday, the billionaire announced that Twitter would make public the algorithms used to recommend tweets to every user. A transparent approach, which he has long advocated without putting it into practice.

“On March 31, Twitter will release the source code for all computer code used to recommend tweets,” said the unpredictable owner and head of the social network Elon Musk. “People learn a lot of stupid things, but we will fix problems as soon as we find them,” he added.

The opening of the service’s black box was part of his stated motive for buying Twitter last year, as was the company’s “potential” and defense of a more absolute vision of freedom of expression.

But his first four months at the head of the network were primarily marked by waves of mass and sudden layoffs, the flight of numerous advertisers, a chaotic launch of paid subscriptions and technical failures.

“Rapid improvements,” Musk hopes.

The multibillionaire explained that Twitter’s recommendation algorithms are complex and poorly understood, even within the California group. “The transparency of the code will be very inconvenient at first, but this should lead to a rapid improvement in the quality of the recommendations,” he said. “We are developing a simplified approach to highlighting more interesting tweets,” he continued.

A measure of transparency could reassure authorities and lawmakers who want to know how the platform works. Many elected officials, especially Republicans, feel it is biased against them, while human rights NGOs criticize the service, especially under Elon Musk, for not fighting misinformation and harassment enough.

Making the code open source also means it can be used by other developers or potential competitors.

So far, no serious alternative to Twitter has emerged. Most elected politicians, organizations and individuals still use it to communicate. Jack Dorsey, former head of the network, was a supporter of open source software.

Back to top button

Adblock Detected

Please consider supporting us by disabling your ad blocker.