It has been nearly a year since the last email update from the Uli team! After the launch in August last year we had a six month period of demoing, planning for the next iteration and fundraising. Uli has been moving full steam since March of this year. But in the busyness of getting the next phase launched, we forgot the lovely ritual of these email updates! We are remedying the situation and will send you monthly(ish) updates for the next year of development.
There are so many details in different tracks within the Uli project, that it is difficult to figure out where to start. I'll start with the TLDR of the findings from the demos: the features in Uli such as redaction and archiving, while exciting, weren't fitting well with the way people were using social media. You can read a longer blog on reflections from the demo and feedback here. In 2021, we had identified the features we would build into Uli through interviews and focus group discussions with nearly fifty gender rights activists and researchers. Many of you reading this email were in that group. But, we now believe that there was significant diversity even within the social media experiences of this group- some people are creators, some more like reporters and some are readers. Some use social media on their browsers but most use it through mobile apps. We designed a little bit for many people's needs, but not perfectly for anybody's. So this is what we are changing this year. We are iterating on Uli in close partnership with four fantastic organizations:
FII and Khabar Lahariya are feminist media organizations that are targeted for the content they create. We are working closely with them to explore if Uli can make it easier for them to handle the negative engagement they receive. We're observing the continuum of private and 'official' social media usage and its effects with members of POV. POV and Nazariya will also be supporting on the creation of the lexicon of abuse- this is to build on the slur list we crowdsourced last year. The lexicon this year aims to capture more nuance around derogatory language- such as whether a certain word is appropriated, or casually used between friends.
In line with our commitment to feminist principle of development of technology, and specifically AI, we're also working to understand how any revenue from licensing of the Uli resources, such as the lexicon should be distributed between those who have contributed to it. Let us backtrack to understand the background on this task- all the work that has happened in Uli so far, be it the code for the browser plugin, the machine learning model or the slur list, is open source. With support from GitHub for a community manager, Uli has only grown stronger as an open source project. You can track the entire roadmap, and work happening in a sprint on Uli's GitHub repository. But, the weakening of Trust and Safety teams across platforms last year has changed our views on opening up all the data for free, for all use-cases across the spectrum. Ideally, these resources should have been developed by Trust and Safety teams (as they have been for English). But in absence of that, we think they should pay to use these datasets in moderation work. We're fortunate that Mozilla included us in their Data Future's Lab Cohort, to work on exactly this question- how should the data crowdsourced through Uli be governed? Who should access it, under what terms? And how should any revenue be shared between those who contributed to the data?
The tasks for the next few months are cut out. There is plenty to be done and you can read more about the team leading the work here:Uli. We're prioritizing team efforts and computer resources. So, for now we have disabled the feature of redaction of tweets. Instead, the feature of abusive words being blurred will work on ALL websites. We'll hold off on announcement of some of the other features under development, till we have concluded our current phase of user research with our partner organizations. If you would like to learn more or join, please feel free to email me. We've also been working hard on our documentation game so that it is easier for people to contribute to the project.
It is worth mentioning that many reports have come out over the last few months documenting the challenge and scale of online/facilitated gender based violence. A report by Center for International Governance Innovation reports on experiences of online harms in 18 different countries. Another report commissioned by the Global Partnership for Action on Gender-Based Online Harassment provides a landscape analysis for a coherent research agenda. Reading these reports, alongside user-research, has been quite interesting. One of the things we noted from the focus group discussions in 2021 was the everyday-ness of gendered abuse, and its manifestation as fatigue. The everyday-ness, beyond the doxxing and cyber-stalking that get noted in FIRs and media, continues to echo through our conversations this year. To make visible something that has become mundane and to de-normalize it- that remains a key goal for us. We have some ideas brewing (beyond building specific features in Uli), but if you are working on creative projects or solutions towards the same goals, we would love to exchange notes and brainstorm together. Please email us firstname.lastname@example.org.