Skip to content
Back to the collection

Bluesky and the AT Protocol: Usable Decentralized Social Media

Martin Kleppmann, Paul Frazee, Jake Gold

Show all 9 authors

Jay Graber, Daniel Holmgren, Devin Ivy, Jeromy Johnson, Bryan Newbold, Jaz Volpert

2024 · arXiv

Read the paper
HN score
0
Mentions
6
Readers
6
Months
4
What’s in a score?

From the abstract

Bluesky is a new social network built upon the AT Protocol, a decentralized foundation for public social media. It was launched in private beta in February 2023, and has grown to over 10 million registered users by October 2024. In this paper we introduce the architecture of Bluesky and the AT Protocol, and explain how the technical design of Bluesky is informed by our goals: to enable decentralization by having multiple interoperable providers for every part of the system; to make it easy for users to switch providers; to give users agency over the content they see; and to provide a simple user experience that does not burden users with complexity arising from the system's decentralized nature. The system's openness allows anybody to contribute to content moderation and community management, and we invite the research community to use Bluesky as a dataset and testing ground for new approaches in social media moderation.

Metadata: arXiv. Abstracts retain their original authors’ rights.

In the conversation

Original HN comments, with labels from simple text rules. A label describes the whole comment and may miss context or irony. Read the discussion to make up your own mind. About the labels

6 selected excerpts · 6 total mentions

steveklabnik
Neutral mention · rule
I believe that https://gitlab.com/bluesky-community1/decentralized-ecosyste... is the canonical link, it's not the one I read originally, but is linked from the "related work" section of https://arxiv.org/pdf/2402.03239
anon7000
Neutral mention · rule
No. This isn’t true. Bluesky has always been very transparent about exactly what flavor of decentralization they offer. The whole thing is swappable microservices you can host yourself, including the relay if you really care. The relay is an optimization to make their app offering work in a performant, scalable way. The relay is described here and here: https://arxiv.org/pdf/2402.03239, https://atproto.com/guides/glossary#relay The protocol Bluesky is built on is certainly decentralized. It’s really fucking annoying that people act like it’s not “real” decentralization, when in fact it’s just a different flavor of decentralization on the technical level compared to, say, the fediverse.
GaryNumanVevo
Neutral mention · rule
Yes and no. The relay is run by Bluesky, but only as a matter of practicality because it requires a large footprint to subscribe to all the PDS events. Others have written custom AppViews and clients already. I run a "one man relay" that only scrapes my PDS, puts it into an appview (which doesn't do much) and I can see that on a basic client that I wrote. The whitepaper clarifies a lot of this: https://arxiv.org/abs/2402.03239
danabramov
Neutral mention · rule
> sound more like 'custom feeds' which users can define based on global content, using predefined criteria determined by a client (web app) which don't really a way to control the behavior of. Which makes this feature only truly operate on the global layer, and 'custom' rather than 'composable'. To be clear, custom feeds aren’t defined in the client app. Anyone can run a custom feed on their own server and with arbitrary logic. A custom feed subscribes to a global firehose (which could also be run by a third party) and uses the stream of the events in the network to produce its results. A user can then publish such a feed under their account, and other users can consume it. This is essentially event sourcing. Here’s a technical paper with details: https://arxiv.org/pdf/2402.03239.pdf We’ll have more to share on composable moderation in near future.
unshavedyak
Neutral mention · rule
I don't know enough about it to give feedback. However Martin Kleppmann's[1] involvement is a huge selling point to me. It's also, seemingly, good enough that they're building very useful things on top of it like PDS, migrating users, etc. ActivityPub (AP) felt a bit like "good enough to get something done", which is amazing on one hand - people can do a lot with it. But it also means some harder problems are totally ignored[2], so the landscape between instances felt rocky to me. I also heavily disliked how federation worked in AP, ie how the protocol felt like it favored centralized instances because small instances were less likely to be pinged, were lower priority, in…
Read the full comment
general was very spammy, etc. Not that the AP proto did anything to cause that.. it just didn't do anything to address it in my view, it was just data. Does AT fix my concerns over AP? No idea. But i appreciate a proto that had a bit more time in the oven and being used to (maybe) solve the harder problems that i experienced with AP. [1]: https://arxiv.org/abs/2402.03239 [2]: These are only my opinions from a very brief stint in trying to write an AP client to federate with Mastodon, Lemmy and Kbin. I did not dive deep into AP, so please don't judge it from my experience.
jakebsky
Neutral mention · rule
This is a big milestone for Bluesky! We've had the federation sandbox running for over six months but we're now able to commit to open federation on the production network this month as well. There's also stackable moderation coming shortly, which enables other individuals/orgs to operate moderation labeling services that users can choose to use. The technical challenges of setting up an (efficiently) scalable decentralized social network were quite interesting. The infrastructure itself is quite decentralized, with standalone PDS instances and two small shared-nothing datacenter PoPs. We're using SQLite with millions of individual databases for each user's repository and…
Read the full comment
ScyllaDB for the global indexing service (AppView). https://bsky.social/about/blog/5-5-2023-federation-architect... If anyone has questions, technical or otherwise, some of the team should be around today to answer them. Edit: HN'ers might also appreciate this paper written primarily by Martin Kleppman about Bluesky and AT Protocol https://arxiv.org/pdf/2402.03239.pdf

Keep following the thread.

More in Human–computer interaction

More from the same topics.

2000arXiv

Dancing links

Donald E. Knuth

The author presents two tricks to accelerate depth-first search algorithms for a class of combinatorial puzzle problems, such as tiling a tray by a fixed set of polyominoes.

10 mentions10 months
2HN score