this post was submitted on 27 Nov 2024
298 points (95.4% liked)

Technology

60346 readers
6480 users here now

This is a most excellent place for technology news and articles.

Our Rules

  1. Follow the lemmy.world rules.
  2. Only tech related content.
  3. Be excellent to each another!
  4. Mod approved content bots can post up to 10 articles per day.
  5. Threads asking for personal tech support may be deleted.
  6. Politics threads may be removed.
  7. No memes allowed as posts, OK to post as comments.
  8. Only approved bots from the list below, to ask if your bot can be added please contact us.
  9. Check for duplicates before posting, duplicates may be removed

Approved Bots

founded 2 years ago
MODERATORS
[email protected]
[email protected]
[email protected]
298
Someone Made a Dataset of One Million Bluesky Posts for 'Machine Learning Research' (www.404media.co)
submitted 1 month ago by [email protected] to c/[email protected]
87 comments fedilink hide all child comments
 

A machine learning librarian at Hugging Face just released a dataset composed of one million Bluesky posts, complete with when they were posted and who posted them, intended for machine learning research.

Daniel van Strien posted about the dataset on Bluesky on Tuesday:

“This dataset contains 1 million public posts collected from Bluesky Social's firehose API, intended for machine learning research and experimentation with social media data,” the dataset description says. “Each post contains text content, metadata, and information about media attachments and reply relationships.”

The data isn’t anonymous. In the dataset, each post is listed alongside the users’ decentralized identifier, or DID; van Strien also made a search tool for finding users based on their DID and published it on Hugging Face. A quick skim through the first few hundred of the million posts shows people doing normal types of Bluesky posting—arguing about politics, talking about concerts, saying stuff like “The cat is gay” and “When’s the last time yall had Boston baked beans?”—but the dataset has also swept up a lot of adult content, too.

you are viewing a single comment's thread
view the rest of the comments
[–] [email protected] 114 points 1 month ago (25 children)

I don't know why anyone would be surprised about this. Bluesky is a distributed system using an open protocol. The whole point of it is that there's no central control.

Same goes for the Fediverse, of course. Everybody should be prepared for the "surprise" that all our posts and comments here are also being used for AI training purposes.

[–] [email protected] 20 points 1 month ago* (last edited 1 month ago) (14 children)

It's not distributed, nor really designed at all like the fediverse. It is deeply centralized, and its architecture requires it to be centralized, or at least to have only huge players with a "gods eye view" for it to work.

Atproto was initially designed as a straight drop in replacement for twitter, so its design makes sense, but its not at all like the Fediverse.

One of the authorities of ActivityPub, the fediverse protocol, just did a very kind but still very blunt breakdown of Bluesky's design choices. she is a big fan of the people involved and some of its positives, but it is not fediverse like, not at all. In her words, it doesn't scale down, only up. You cant have a small bluesky server. To work, you need all data sent to everyone, on every instance. The data demands for just the current influx is TBs/month of data, and climbing (according to the link below, they use 16TB of nvme storage right now after the recent surge, which would be thousands /month on any cloud service. This will climb dramatically).

All data being public is a design choice by Bluesky. It is also a different design choice by the fediverse that comes to the same outcome, but that does have an answer if we want it. I know gotosocial did something interesting to make fully private votes by using a empty shell profile that votes, but tying that in a tricky way to your account. So there are fediverse answers to privacy, but there may not be bluesky answers.

EDIT: One of the blueksy/atproto devs replied to the above link today. The gist reinforces the point that the service is intended to be run by large orgs, including corporations, but also big non profits like the internet archive or Wikipedia. His take is that user experience is key, and for that you need big money and easy features. They are hoping that since the pieces of atproto can be hosted separately by separate giant orgs, that market forces will make it viable to be decentralized.

[–] [email protected] -1 points 1 month ago (2 children)

that dev is full of shit. nothing in decentralized systems limits ease of use and functionality. just makes the software harder to write. the invisible hand of the market nonsense is classic misdirection.

[–] taladar 5 points 1 month ago (1 children)

Plenty of things are more difficult in decentralized systems.

You have to store all kinds of data either in multiple copies/caches or get long delays on certain operations such as search or even just displaying aggregated data (such as a post and its comments from different instances on Lemmy).

You have the problem of different jurisdictions and moderation policies for different instances.

You will have a hard time exporting or deleting all data related to a specific user when required by law (e.g. GDPR).

[–] [email protected] 1 points 1 month ago

Difficult != Can't be done. I'm well aware of the difficulties. Distributed systems design is one of my specialties.

GDPR only applies to your servers. Data deletion is probably the easiest part to deal with.

load more comments (11 replies)
load more comments (21 replies)