Elon Musk, the owner of X, has pledged to address the issue of “shadowbanning” on the social media platform that was previously known as Twitter. Shadowbanning refers to the practice of making changes to a user’s content so that it is not widely circulated and remains hidden from public view, without outright banning the user. Responding to users on X, Musk apologized for the delay in resolving this matter and elaborated on the challenges that X is currently facing in providing transparency about shadowbanning.
The concept of shadowbanning has raised concerns since Twitter’s early days, as users were unaware that their tweets were being restricted in this manner. When shadowbanned, users can continue posting on the platform as usual, but their tweets receive limited engagement and responses compared to their regular activity. This issue has often been a subject of political debate, with figures like Trump and other Republicans claiming that Twitter was suppressing their tweets.
Upon assuming control of Twitter, Musk endeavored to demonstrate that shadowbanning was a widespread practice on the platform. He released certain information to the public, though this mainly provided a behind-the-scenes view of the intricate process of moderating social media content.
Despite this, Musk has stressed the importance of informing users if they have been shadowbanned. During the Viva Tech conference, he mentioned that X was working on a solution to inform users about any impact on their accounts within the Twitter system.
In a recent tweet, Musk delved into the complexity of the issue, expressing frustration with the numerous layers of “trust & safety” software that contribute to prolonged investigations into suspended or shadowbanned accounts. He announced a ground-up rewrite of the codebase to simplify this process.
However, Musk’s explanation did not fully clarify the challenges facing X’s current endeavor. A more detailed understanding was provided by Yoel Roth, the former Head of Trust and Safety at Twitter, in a post on the decentralized Twitter alternative, Bluesky.
Roth acknowledged the validity of Musk’s statements, despite their somewhat convoluted nature. He began by explaining that social media platforms typically maintain records of banned users and the reasons for their bans. Initially, these records might be in the form of a spreadsheet or Google Doc, particularly for smaller companies. As the platform expands, this information is often integrated into metadata directly linked to user accounts.
Roth highlighted that Twitter still employs a system of free-text notes attached to user accounts to store enforcement-related metadata. He noted that this approach was retained due to prioritizing other tasks, despite the challenges of programmatically parsing such free-text notes. These notes, while understandable by humans, are not easily structured for automated communication with users about their account status.
Roth concurred that revising the code governing enforcement attribution was a prudent step and was in progress before Twitter’s acquisition. The implementation was originally planned for this year. He playfully speculated whether the delays might have been caused by Musk’s takeover of Twitter and the subsequent organizational changes.
Roth elaborated on the complexity X faces, particularly in dealing with the infrastructure that addresses spam. He acknowledged that while this complexity is necessary to combat spam effectively, it makes pinpointing the exact cause of a user’s shadowban challenging.
Concluding his observations, Roth expressed his support for Twitter’s infrastructure overhaul and wished them success. He humorously mentioned leaving documents about the process in his Google Drive at the company.
Considering the intricate challenges described by Roth, it seems unlikely that X will promptly implement the technology to display users’ account status, as promised, unless the project had been reprioritized earlier.