Results from a Twitter project aiming to tackle misinformation and propaganda were published just hours before Elon Musk鈥檚 takeover of the company on 27 October, suggesting they may have been rushed out before the change at the top. The report shows that Twitter鈥檚 Birdwatch project has made some progress in countering misinformation, but external researchers fear the scheme may now be at risk of swingeing cuts as Musk attempts to make the social media company profitable.
Twitter in an effort to stop the spread of misinformation in the US. The company recruits a small subset of users to label potentially misleading tweets and write notes that include additional context. These notes can then be chosen by an algorithm for display underneath the original tweet for US users. For instance, a tweet claiming that a vaccine causes side effects could be placed in the context of a note that points to research showing that this is only the case for a fraction of those taking it. Such a claim doesn鈥檛 necessarily violate Twitter鈥檚 rules, which would trigger removal, so these notes aim to counter false claims.
Unlike some social media platforms that use simple up-and-down voting to raise or lower the visibility of content, Birdwatch uses a that seeks to get feedback from people with a wide range of views in order to cancel out bias 鈥� to effectively 鈥渂ridge鈥� the gap between partisan viewpoints. Notes are only shown under tweets if they are rated as helpful by enough people with different perspectives, from both ends of the political spectrum, based on their previous interactions with other content.
Advertisement
In its first few months, the scheme was that were being promoted, and for a lack of citations provided by contributors. A leaked internal document revealed that at one point the company even accepted an as a Birdwatch contributor. But Twitter continued to tweak the algorithms behind the tool and a new report on the project shows that progress has been made.
Researchers at Twitter ran a survey in April that showed tweets picked up by the bridging algorithm to a subset of users who aren鈥檛 Birdwatch contributors and, for some of those respondents, the notes attached to them. They found that 64 per cent of respondents who saw the notes alongside the tweets found them subjectively helpful. On average, notes selected by the algorithm reduced the likelihood of agreeing with the substance of a potentially misleading tweet by about 26 per cent, and there were no statistically significant differences between people reporting as Republican or Democrat voters, suggesting that political bias wasn鈥檛 an issue.
at the Poynter Institute for Media Studies in Florida says that seeing the report get rushed to a public release was 鈥渉eartbreaking鈥� because it highlighted that Birdwatch could be shut down.
鈥淚 think the silence from a lot of the people who work on Birdwatch now leads me to believe that it might be in jeopardy under the new ownership. It looks like they鈥檙e reducing the number of human moderators who can put checks on misinformation,鈥� he says, referring to that content moderators at the company were being frozen out of the tools required to do their job. 鈥淚f it goes away then there goes an opportunity to tackle misinformation in one of the least political and most palatable ways I think we have seen.鈥�
Musk 鈥� who didn鈥檛 respond to a request for comment 鈥� has repeatedly said that his purchase of Twitter was motivated, at least in part, by protecting free speech on the platform, but there has been little concrete detail of his plans to do so.
Keith Coleman at Twitter, who has overseen Birdwatch since its launch, didn鈥檛 respond to a request for comment but is understood to still be at the company. In a he claimed that Twitter would continue recruiting new users to Birdwatch at a rate of 10 per cent growth a week.
Twitter was contacted for comment, but no staff were available for interview.
Reference:
