Building a better Birdwatch

As of November 2022, Birdwatch is now Community Notes.

Last year we launched the Birdwatch pilot — a new, collaborative way for people to add context to Tweets they believe are misleading. By empowering people to do this together, they can add helpful and informative context for people from different points of view. 

We’ve made a slew of improvements to Birdwatch based on feedback from contributors, the public, and academic researchers, and we’re seeing promising results. In surveys with randomly sampled people on Twitter in the US, we’ve found:

• The majority of people found Birdwatch notes (specifically those designated helpful by Birdwatch contributors) helpful. This includes people from across the political spectrum.
• People in the survey were 20-40% less likely to agree with the substance of a potentially misleading Tweet after reading a note about it, compared to those who saw a Tweet without a note.  
• These are encouraging signs that Birdwatch can be helpful and informative to people on Twitter. That’s why we’re now expanding the visibility of the pilot so we can make additional improvements with feedback from more people on Twitter. 

Expanding the visibility of Birdwatch

Starting today, a small (and randomized) group of people on Twitter in the US will see Birdwatch notes directly on some Tweets. They’ll also be able to rate notes, providing input that will help improve Birdwatch’s ability to add context that is helpful to people from different points of view.

To appear on a Tweet, notes first need to be rated helpful by enough Birdwatch contributors from different perspectives. Difference in perspectives is determined by how people have rated notes in the past, not based on demographics. You can learn more about how we assess this here.

How we got here

When we started a year ago, we set out to build Birdwatch in public, with input from people on Twitter, Birdwatch contributors, academic advisors, and researchers. We made all data contributed to Birdwatch publicly available, along with the code that uses that data to rank notes, and we continue to do so. 

Feedback and insights from the Twitter community has led to key improvements to Birdwatch:

• Strengthening the way Birdwatch identifies notes that a diverse set of people find helpful.
• Doing more to protect contributors and help them feel more comfortable contributing to Birdwatch by rolling out auto-generated aliases for everyone in the pilot.
• Helping participants understand the impact of their Birdwatch contributions by notifying them if their notes are rated helpful or if notes they rated are elevated.
• Encouraging participants to include sources and clear explanations in their notes with in-app prompt reminders.
• Improving the discoverability of notes that need a broader range of feedback by adding a personalized ‘Needs Your Help’ tab for pilot participants.

We recently announced a collaboration with the Associated Press and Reuters to help our teams assess the quality of information elevated by Birdwatch participants. We’ve also onboarded an advisory board of scholars and researchers who study misinformation, polarization, online manipulation, and harassment. Advisors are from institutions including MIT, the University of Washington, the University of Michigan School of Information, and more.

We’re thankful for all 10,000 pilot contributors that have played a key role in making Birdwatch what it is today. Along with the time and effort they invest in writing and rating notes, Birdwatch contributors have consistently provided priceless feedback on their Birdwatch experience and shared thoughtful and honest feedback about what's working and what could be better.

What’s next

We’ll continue to build in the open as we learn, iterate, and expand the pilot. Follow @Birdwatch for the latest updates and to provide feedback on how we’re doing.