Twitter pilots Birdwatch crowd-sourced misinformation tool

Twitter has put its 'Birdwatch', a crowdsourced misinformation tool that will allow users to annotate tweets they believe to be misleading, into a pilot test in the US.

Users in the US are being invited to sign up and participate in the first phase of the pilot, which will permit approved participants to make annotations to tweets they believe to be misleading or argue for their authenticity. Those who want to sign up must have a US-based phone carrier, verified email and phone number, and no recent Twitter rule violations.

In the pilot phase the notes will be collated on a separate Birdwatch site, although Twitter hopes to eventually make the notes visible directly on tweets when there is consensus from a broad and diverse set of contributors. The idea being that the notes would act as a sort of public 'ledger' of a Tweet's authenticity. 

The notes are intended to provide informative context, and pilot participants will be able to rate the helpfulness of notes added by other contributors in the Birdwatch site.

The feature was first spotted in development by app researcher Jane Manchun Wong in August last year. Since then, Twitter says it has conducted more than 100 interviews with its users across the political spectrum and has received broad general support for the tool.

The social-media network believes that leaning on its community to identify misleading information in tweets could be a faster and more effective way to prevent misinformation from spreading.

But it also recognises that community-driven systems present "a number of challenges", Twitter's VP of product Keith Coleman said in a blog post announcing the pilot program on Monday (January 25). This includes ensuring it doesn't get dominated by a simple majority or become biased based on its distribution of contributors, and preventing manipulation attempts.

"We’ll be focused on these things throughout the pilot," Coleman said.

Twitter said the pilot will be limited to a small test group of 1,000 users in the US but will grow over time. It will prioritise admitting users who differ in the people they follow and the kinds of tweets they engage in, in an attempt to reduce the likelihood that participants would be predominantly from one ideology, background or interest space.

"We know that the broader and more diverse the group, the better Birdwatch will be at effectively addressing misinformation," Coleman said in the blog.

Twitter wants to build Birdwatch in the open to enable experts, researchers, and the public to analyse or audit the tool and identify opportunities or flaws. All data contributed to Birdwatch will be publicly available and downloadable, and algorithms developed to power Birdwatch will be published publicly in the Birdwatch Guide.

The community-driven tool is part of Twitter's efforts to address the continuing threat of misinformation and its influence over politics and public health, especially in the midst of a pandemic and the uncertainty surrounding vaccination. Birdwatch is intended to supplement Twitter's own fact-checking labels.

"We apply labels and add context to Tweets, but we don't want to limit efforts to circumstances where something breaks our rules or receives widespread public attention. We also want to broaden the range of voices that are part of tackling this problem, and we believe a community-driven approach can help," Coleman explained.