iiinova — where your digital self keeps evolving

Reposted by onur.ascigil

Scoiattolo @scarnecchia.net · 21/03/2026

Interesting paper on the impact of blocklists on the Bluesky social graph. Self-recommending. (will read it again after coffee and likely have more to say then)

openaccess.city.ac.uk

City Research Online - Open or Blocked Skies? Community Moderation Practices in Bluesky

3206

Reposted by onur.ascigil

Leo Balduf @leobalduf.bsky.social · 21/06/2025

I'll be at #ICWSM 2025 next week to present our paper about Bluesky Starter Packs. For the occasion, I've created a Starter Pack with all the organizers, speakers, and authors of this year I could find on Bluesky! Link: go.bsky.app/GDkQ3y7 Let me know if I missed anyone!

63013

Reposted by onur.ascigil

Michał Król @harnen.bsky.social · 23/02/2025

Are you interested in @bsky.app research? We gather and make available multiple datasets including posts, social graph snapshots, labellers, feed generators and more. More information is available on our website: bsky.leobalduf.com

bsky.leobalduf.com

Bluesky Research

055

Reposted by onur.ascigil

Leo Balduf @leobalduf.bsky.social · 27/01/2025

Oh hey, our most recent work on @bsky.app was cited by Nature! :o

084

Reposted by onur.ascigil

Ignacio Castro @ignactro.eurosky.social · 27/01/2025

Did Starter Packs turbocharged @bsky.app growth? Yes, but... We analyse it in the paper quoted in @nature.com article below:🧵 bsky.app/profile/harn...

075

Reposted by onur.ascigil

Michał Król @harnen.bsky.social · 26/01/2025

@bsky.app grew tremendously over the last year. Recently, the platform introduced “starter packs” (SP) to help users bootstrap their social graph. Anybody can create these SPs. They’re lists of accounts (e.g., “people creating anime”) that others can then follow. 🧵1/11

12010

Reposted by onur.ascigil

Ignacio Castro @ignactro.eurosky.social · 14/11/2024

Code and paper analysing Bluesky 👉 dl.acm.org/doi/10.1145/...

075

Reposted by onur.ascigil

ACM Internet Measurement Conference @acm-imc.bsky.social · 04/11/2024

We have an #IMC2024 paper on Bluesky itself: Looking AT the Blue Skies of Bluesky with Leon Balduf (Technical University of Darmstadt) & Saidu Sokoto (City, University of London) dl.acm.org/doi/10.1145/...

dl.acm.org

Looking AT the Blue Skies of Bluesky | Proceedings of the 2024 ACM on Internet Measurement Conference

0124

Reposted by onur.ascigil

Ignacio Castro @ignactro.eurosky.social · 04/11/2024

Twitter shows the risk of the centralised/monolithic/vertically integrated organisation approach with a benevolent dictator at the top: including the elusive nature of benevolence. 2 #IMC2024 papers examine Mastodon & Bluesky exploration of the design space of alternatives: [1/4]

4218

Reposted by onur.ascigil

Michał Król @harnen.bsky.social · 18/10/2024

The full paper "Guardians of the Galaxy: Content Moderation in the InterPlanetary File System" is available here: usenix.org/conference/u... Written with @ignactro.bsky.social @asonur.bsky.social @bibo7086.bsky.social @leobalduf.bsky.social

usenix.org

Guardians of the Galaxy: Content Moderation in the InterPlanetary File System | USENIXusenix_logo_notag_white

053

Reposted by onur.ascigil

Ignacio Castro @ignactro.eurosky.social · 26/08/2024

To appear soon at ACM IMC 2024! 👇

1112

Reposted by onur.ascigil

Paul Rohr @pevohr.bsky.social · 25/08/2024

Cool preprint leveraging access to the "data is public by default" architecture of the #ATprotocol to infer specific moderation practices by the bsky T+S team. They also calculated how little overlap there is between that team's work + the additive work being done by other labelers. Neat stuff!

For the official Bluesky Labeler, we observe two groups of labels: In the lower half of the plot, we find porn, nudity, corpse etc., which are applied within seconds, indicating automated systems. On the other hand, labels such as, e.g., spam, sexual-figurative, intolerant, and !takedown take longer to be applied, pointing to manual processes. ...

It seems that heavy-handed moderation decisions such as removing data are deliberated instead of automated. This is reassuring, especially for the !takedown label having significant consequences for the affected users.

In general, we find that label values generally show little overlap between the Bluesky Labeler and community Labelers, i.e., they seem to deal with mostly disjoint topics: Only 56,856 (1.8 %) objects are labeled by the Bluesky Labeler and any community Labeler. The Bluesky Labeler mostly labels NSFW content and upholds some community standards whereas the community Labelers seem to operate in specific niches. This is enabled by the flexibility (or lack of predefined) label values and user choice in how to react to labels.

1123

Reposted by onur.ascigil

bryan newbold @bnewbold.net · 25/08/2024

pretty interesting pre-print about bluesky/atproto! has an analysis of labeler response time (by label value), and digs in to feed service providers. the intro describes the atproto design goals and what is unique pretty well, IMO.

17230