This shouldn’t come as a huge surprise. Meta is moving forward with their plans for Theads and the Fediverse, and their adjusted terms reflect a new impending reality for Fediverse users.
This shouldn’t come as a huge surprise. Meta is moving forward with their plans for Theads and the Fediverse, and their adjusted terms reflect a new impending reality for Fediverse users.
For all the fucks’ sakes, people.
Yes, Meta sucks. But at least get your shit together before you all start falling over each other to say how these ToS changes mean that Zuck has now given birth to Time Travelling Baby Hitler or some shit.
Meta says, for Threads to federate, they access the same data any instance does when it federates.
And as far as LEMMY.world defederating from Threads… LEMMY. That’s like saying Twitter (or W, or whatever the hell it is now), shouldn’t put Facebook posts in its timeline. Threads is a Mastodon concern. Not Lemmy.
🤦♂️ Ya fuckin’ tinfoil hat nerds. I love you all. But God damn.
I agree that this is nothing to panic over, but I want to clarify that Lemmy is not safe from this. Lemmy and Mastodon both use the same protocol (ActivityPub) and that’s also the protocol that Threads will use to federate. Just as Mastodon users can like, boost, and reply to Lemmy threads / comments, Threads users will be able to do the same. That’s why it’s important to defederate Threads on all ActivityPub-enabled instances.
Technically. Yes.
But doing so is onerous enough that I can’t see it as any sort of “threat”.
And again… Defederating does absolutely zero to restrict Meta from being able to access your info. Defederating means you don’t see Meta. It doesn’t block Meta from seeing you.
You don’t even need to dip your toes into ActivityPub to scrape most of the data. It’s public – aside (I think) from just user IP addresses on Mastodon. And in the case of Lemmy, I don’t think there’s anything you can’t access from outside of ActivityPub.
Defederating actually does stop Meta from accessing data (at least through ActivityPub) if you enable AUTHORIZED_FETCH / similar. That setting requires remote instances to authenticate themselves, which prevents blocked instances from querying anything. IIRC, Lemmy either already supports or plans to support that same feature.
Meta could, of course, just use web scraping, but that can be prevented with DISALLOW_UNAUTHENTICATED_API_ACCESS. Although admittedly, I don’t think Lemmy has this feature yet.
Even DISALLLOW_UNAUTHENTICATED_API_ACCESS can be easily bypassed by creating a client that logs into mastodon.social (for example), and just gobbles up the Federated feed.
It’s what the FediBuzz relays are now doing in order to keep single-user instances viable and not funnel everyone to the same 3 instances.
Unfortunately, if Meta wants to be shitty, they’ll be shitty. Even stuff like robots.txt & nofollow tags are just polite requests that can be ignored by shitheads.
kbin includes a “microblog” feature which is a mastodon-like implementation of ActivityPub.
Without jumping through flaming hoops, though… does the “Threads” tab really ever talk to the “Microblog” tab? (aside from your kbin account being able to interface with both)
(I do find it funny that kbin’s “Threads” is their Lemmy/Reddit-like, and not their Mastodon/Threads/Twitter-like)
I don’t use it, so I’m not super clear on it. It does feel like a bit of an afterthought.
I do know that I’ve interacted with Mastodon users in fediverse comment threads via kbin in the “regular, reddit-like” interface. My understanding is that APub is APub is APub, and the client implementations define the format you see content in, and implement or do not implement different APub features based on how the developer(s) want to shape their client.
Mastodon users can post on lemmy
Watch this
Threads is not Mastodon. Both are microblogging, while Lemmy is better described as a forum or link aggregator.
It’s possible to interact with Lemmy from Mastodon. I do so regularly by tagging a community in myastodon post. Following a community from Mastodon is also possible, but the UX is rough.
deleted by creator
If all you’re looking for is eloquence and footnotes, this person in this post’s comments already has you covered…
https://kbin.social/m/fediverse@lemmy.ml/t/410613/-/comment/2032665
deleted by creator
Good job on W, but I’m pretty sure it’s L.
You can follow lemmy stuff via Mastodon accounts. 🤔 No? Do I not quite understand how and why that works?
Your point about slowing the fuck down still stands though.
You can follow @user@instance or @community@instance on Mastodon. But you just get blasted by every post and comment, one by one, in unthreaded CHRONOLOGICAL order.
So while you can do it, the question is… should you?
Ok, that was my experience. I haven’t found a great context/use case for it yet.
It does seem a client could be made that uses the functionality. Or a purpose deployed instance or community could make use of it too.
But I agree, it’s hard to imagine a good use. And your point still stands about how panicking is unhelpful.
Okay, but like, I don’t want Meta consuming that data? At least if they wanted to scrape through reddit to put that together, they’d be going out of their way. This data is now just coming through the same API “for free”.
If I didn’t mind Meta scraping through all this, why wouldn’t I just use Threads?
This is exactly the kind of shit that pushed me here - I don’t want Meta sifting through all my shit. Its unlikely that some other instance host is going to start building psychological advertising profiles on me and sell it to the highest bidder. But you bet your ass Meta will try.
I’m curious what precautions you have taken to prevent web scraping of your posts.
Probably none. For all he knows Meta already owns a couple small instances.
They admitted to federating for research back around bluesky’s announcement.
If you don’t want your data scraped you’ll need to use e2ee.
deleted by creator