Annual Report

(March 2019 - March 2020)

Annual Report Cover

Tattle is building open tools and datasets for increasing the reach of accurate information to mobile first audiences, and for enabling greater understanding of information trends on chat apps and social media.


Stock Taking

While we started working on Tattle in 2018, Tattle was legally incorporated in 2019. The organization completed its first year as a legal entity on March 25th 2020.

Tattle purposefully chose to be an open source project. We believe transparency is critical to build trust in the work we do. And trust is vital for a project aiming to address challenges of excess and misleading information. In this annual report we list all that we have done through the last one year. Through writing this we realized- we are only just getting started. We hired the first full time employee in September 2019. Now, Tattle is supported by three full time staff members and a network of volunteers and interns.

This March, we shared the Stakeholder Map that we came up with in early 2019. This map helps ground our brainstorming sessions on citizen driven action against misinformation. Amongst all these different stakeholders, Tattle focuses on three:

  • Everyday misinformation warriors
  • Fact-checking groups
  • Researchers

In this first year, Tattle’s work has been most useful for researchers. For example, the fact-checking sites dataset that Tattle released has been used by researchers to understand Covid specific trends. In 2020, we’ll focus on adding more datasets and building tools that fact-checking groups can use out-of-the-box. All the while, we’ll be adding more functionalities for motivated chat app users to search through the archive in Indian languages and through different modalities (video, audio, mixed-media).

What We’ve Built

  • Tattle Khoj: a service that tells you if an image has appeared in a fact-checking article.
  • TattleJod bot: A telegram bot that lets individuals tag and push content to the Tattle archive.
  • Tattle Archive: The main archive for annotating and organizing the data.
  • Data Collection:
    • Released datasets containing text, images and videos checked and referenced by six fact checking sites.
    • Built tech for scraping WhatsApp groups.

Presentations/ Talks

Media Mentions

  • Vice Story on Deep Fakes in India
  • NPR interview on use of Deep Fakes in India
  • The Quint republish of a blog analyzing Covid-19 related misinformation that used Tattle data.

Financials

Revenue:

AI Ethics Grant: 100K USD

Expenditures:

The primary cost drivers for the project this year were:

  • Full-time Staff for the Project
  • Technical Infrastructure
  • Travel Associated with Conferences and Research

Annual Report Cover

Contributors/Collaborators:

Through the year, several people have worked with and helped the project grow:

  • Mohit Malik: First intern who worked on the archive interface we use currently.
  • Saakshita Prabhakar: Sounding board and advisor for all things design.
  • Nilomee Jesrani: Led several design sessions that led to the wireframes for Tattle Khoj.
  • Swair Shah: AI engineer leading Tattle search, and advisor for all things ML.
  • Sunandan Chakraborty: Sounding board for data science work.
  • Venkatesh Thapan: Built the TattleJod telegram bot used for crowdsourcing content for the archive.

Finally, we'd like to remember Keshav Joshi. He worked with Tattle full time for four months in 2019. He is responsible for much of the fact-checking sites data that we've shared with researchers and is driving Tattle Khoj. He passed away in 2020. We remember him of course for his contributions to the project but also for his quirks and jokes. He believed and reminded us, that everything is learnable- an important reminder for Tattle where we must keep learning, often across disciplines, to be effective in our work.