Cornell Tech STech Lab Bluesky Account Activity Labeler

Why We're Doing This

Social media platforms play a central role in how people communicate, organize, and access information. But understanding more about who you're engaging with online is often difficult.

In this research project we explore whether making certain behavioral signals more visible to users can help address some of these problems. Specifically, we are building tools to surface transparent, objective information about account behavior on the Bluesky platform.

What We Do

We built a labeling system for Bluesky. Our system is using the AT Protocol to collect publicly available account activity, such as how often a profile's bio is updated, or patterns in following and unfollowing behavior, and synthesizes this into labels that subscribers can see when browsing the platform.

Importantly, our labels highlight specific, observable behaviors and let users draw their own conclusions. Mass frequent posting, for example, might reflect automation, but it might make sense if it's the account of a major news outlet or is otherwise a prolific poster. We present the signal, and users decide what to make of it.

How We Use Data

To understand the effectiveness of our labels in improving the engagement on Bluesky, we may observe users' activity on Bluesky through the platform's public data stream (the "firehose").

We do not collect any private messages or any information that is not publicly available on Bluesky or the AT Protocol. Data will be stored securely and used only for research purposes. Any findings we publish will not identify individual participants.

Contact

If you have questions about the research, please reach out: stech_labeler@cornell.edu

We would also welcome your suggestions or if you have encountered any bugs or issues with our labels.