Tattle Updates: Second Quarter of 2025

This quarter's flavour was stacks. We wish it were stacks of money, but it was just plain old software stacks. We participated in two global events quite serendipitously at the same time which made this connection stronger than we would have realized otherwise.

At the UN Open Source Week, Tarunima co-hosted an event along with Roost.tools and the UN Peacekeeping Force on the building blocks of Trust and Safety software.

At the Internet Governance Forum 2025, Denny presented Uli and Feluda at the DPGA booth. There was a lot of discussion about digital public goods that work well together and "stack" up to form solutions organizations and countries can deploy during crises. He also participated in a session on building public-spirited social media where participants from around the world discussed the building blocks of such platforms.

A digression into stacks. Technologists like to break complex things into blocks that stack in layers on top of each other. The idea is that each component is precisely defined and the way it coordinates with the layers above and below is also well defined (often called a protocol).

Here's an example of a stack diagram of the various technologies that power the internet. Each block has organizations responsible for developing and governing it.

Over the course of our work, we have encountered groups around the world working on very similar problems. There is now a growing recognition that in the Information Integrity landscape, many of us are reinventing similar solutions. Given how resource-constrained the space is, this redundancy is costly.

Most project features rely on similar building blocks such as data scrapers, access control, annotation, and search. The real work lies in adapting these building blocks and their interactions to suit each project's specific needs.

A stack perspective can help break silos between different groups working in this area. At the same time, we recognize that some amount of reinventing the wheel is inevitable. There are many reasons why groups may build a component from scratch even when an open-source version already exists.

Something we highlighted at the UN Open Source event is that standards enabling interoperability across projects can allow innovation while reducing unnecessary duplication.

Project Updates

Sorting, Search, and Dashboards: Uli's New Features at a Glance

Based on feedback from our Uli community, we rolled up our sleeves and introduced several new features for Uli.

You can now sort and paginate crowdsourced slurs, making navigation easier. Filters are also available based on date, severity, source, categories, and more.

We also added advanced multilingual search so users can find relevant results across different languages.

To make insights easier to explore, we created a dashboard that visualizes slur distribution by severity, submission trends over time, and the top 10 most contributed slurs.

New Features: https://uli-community.tattle.co.in/crowdsource-contributions
Dashboard: https://uli-community.tattle.co.in/dashboard

Sessions with Uli Community

We opened up the space for contributions to the Uli Slur List through a series of community sessions.

Over five weeks, we hosted open-call meetings with experts from organizations including Rati Foundation, Citizen Digital Foundation, Digital Empowerment Foundation, and several independent researchers. Their insights played a major role in shaping Uli's latest features.

We also connected Uli to broader conversations around Online Gender-Based Violence (OGBV). This led us to Kalindi College, Delhi University, where we conducted a full-house session on "Navigating Social Media and an Introduction to Uli."

In June, we also connected with Sajhe Sapne's alumni for a discussion on "Understanding and Navigating Online Harm on Social Media Platforms."

Image credits: Sajhe Sapne

Feluda as a Python Library, Yes That Happened!

Feluda had quite a journey this quarter. We gave it a proper revamp and made it available as a Python library, making it much easier for developers and researchers—especially those who aren't well-versed in Machine Learning—to use it for data analysis.

We're also working on making Feluda even better by improving its internal structure, polishing the documentation, and optimizing performance. All of this is setting the stage for a solid v1.0 release, so stay tuned!

On top of that, we joined the Dedicated Mentoring Program and onboarded a contributor who's been doing great work to standardize and further enhance Feluda.

Viral Spiral 1.0 is Here!

After a year of hosting both in-person playtests and online sessions, we've been through quite the adventure with Viral Spiral. Along the way, we gathered bug reports, feature requests, and encountered scaling challenges—but all of that helped us improve the game.

So we rewrote the software from the ground up with two major goals:

Make Viral Spiral easy for our design team to test and iterate.
Build a backend capable of supporting 1,000 concurrent games.

And guess what? We achieved both.

If you haven't tried Viral Spiral yet, give it a try here.

If you'd like to join one of our scheduled YouTube livestreams, learn the game, and play with us in real time, sign up here.

Say Hello to anno.milli – A New Archival Annotation Prototype!

We teamed up with the Archives at NCBS and the Milli Archives Foundation to build an MVP for an archival annotation platform, and we're excited about how it turned out.

The anno.milli prototype allows users to search and explore digital archival files, add their own annotations, and discover approved annotations contributed by others.

The application is integrated with the Archives at NCBS test catalog. While browsing the archive, users can jump directly to the anno.milli annotation app to contribute annotations.

One of the most exciting aspects of the project is that it's built around open archival standards like EAD, while also exploring the possibilities offered by web annotations.

Interestingly, the W3C Trust and Safety Working Group is currently discussing the use of web annotations for content labels added by third-party organizations such as fact-checkers and content moderators.

We hope our work with Milli encourages greater collaboration across communities working on information integrity and helps break down long-standing silos in the space.

That's all the exciting work we had stacked up for April, May, and June.

We can't wait to share what comes next quarter—there's plenty more in the pipeline!

Stay tuned, and wishing you all the best until then.

Warm Regards,

Team Tattle