Original post: https://bsky.app/profile/ssg.dev/post/3lmuz3nr62k26
Email from Bluesky in the screenshot:
Hi there,
We are writing to inform you that we have received a formal request from a legal authority in Turkey regarding the removal of your account associated with the following handle (@carekavga.bsky.social) on Bluesky.
The legal authority has claimed that this content violates local laws in Turkey. As a result, we are required to review the request in accordance with local regulations and Bluesky’s policies.
Following a thorough review, we have determined that the content in question violates local laws in Turkey, as outlined in the legal request. In compliance with these legal provisions, we have restricted access to your account for users.
pardon my ignorance, but how is a de-centralized and de-federated online community bound to such annoyances?
Assuming you are serious:
Bluesky is … arguably ‘federated’, but it is centralized, not decentralized.
https://gigazine.net/gsc_news/en/20241128-bluesky-decentralization
Their model (AT Protocol) relies on a central, authoritative … ‘Relay’, that all ‘federated’ users and posts on federated PDS (personal data servers) must go through, to actually reach the ‘AppView’, ie, what all other people/users can actually see.
So, this is not a many to many, tangled spider web of connections, the way lemmy, and other parts of the actual fediverse are.
It is a top down hierarchy, a pyramid.
And Bluesky runs the Relay, the chokepoint.
If Bluesky cuts off the PDS your account is on, everyone on it is now gone.
The actual fediverse, Mastadon, Lemmy, etc, runs on ActivityPub.
In that model… every instance is essentially self contained, and every instance that is federated communicates with every other instance that is federated.
Each instance can decide what other instances they want to federate with… and users on each instance can personally block even more other users, communities, or entire instances if they choose to, but that only effects what that particular user sees.
That is what you call decentralized, approaching, or also having elements of being ‘distributed’.
To bring up an example without getting into the drama that led to it:
The ‘Tankie Triad’ of ml, lemmygrad and hexbear have had a number of other instances defederate from them.
But, there are also a good number of instances that have not done so.
So that means if your account is on hexbear… you can’t see or post on an instamce that has blocked your instance.
But, if you (a hexbear…ian?), post on a neutral instance… users on that neutral instance will see the post.
But but, if a user from an instance that has defederated from hexbear goes to to the neutral instance… they will not see the hexbearian’s post.
This sounds complicated, and it is, but … thats the whole point of a decentralized system. It is more complex in the abstract… but the entire system ends up being more robust, more adaptable, more customizable… without a central authority in direct control of the entire system.
i was asking in good faith, and i can’t thank you enough for providing such a thorough and effective answer.
it almost sounds like bluesky is just a baby twitter in the making, and it’ll probably end up the same way. i’m really digging the actual fediverse thing, mainly because it seems to be one of the only places that money and vc bs hasn’t been able to touch.
I just wanted to clarify, as… at least for myself, even here on lemmy, discussions about this have been going on for at least 6 to 9 months, and … a good number of people have not been engaging in those discussions in good faith.
But yes, I am happy to answer, glad you found it helpful!
Apologies for the hilariously simplistic graphics… i literally just drew them on my gas station tier phone haha. But I think they get the point across.
Yep, it pretty much literally is twitter 2.0 (3.0?), was founded by Jack Dorsey, … its not even a non profit, it is a for profit ‘benefit’ corporation, which basically just means its corporate bylaws claim that it attempts to benefit the public in some way.
IE, literally the corporate / legal version of virtue signalling… it is still ultimately a for profit corporation that will put profit and growth above everything else… and hopefully by now, people understand how that literally always turns out.
So the decentralized version makes sense to me. The blue sky model you describe sounds like just farming out the server load. What am I missing?
That is literally how I read it as well, BlueSky is farming out server load to enthusiastic and dedicated users, while also just going ham on the PR / propoganda / marketing making themselves appear to be something they are not.
Unless I missed something and BlueSky is actually letting people run and custom configure their own relays at least semi independently… yeah, they’re basically being quite shady and misleading.
For relays yes, but for PDS that’s not at all true. The PDS architecture lets you own your data and migrate it away from Bluesky servers or even from the BS apps, when/if they will be available. Something that ActivityPub severely lacks. Try to migrate your account from one Lemmy instance to another.
Yes, you can host your own PDS server, that is known and stated.
The entire design of a lemmy instance is meant to be more ‘self contained’, as I already mentioned. This is what enables the federation network to organize in a ‘many to many’ connection style, as opposed to a ‘many to one’.
A lemmy instance roughly has many/most of the capabilities of a PDS, Relay, and AppView… all rolled into one.
This is a fundamental difference of a ‘true’ federation model… all the members of the federation are capable of operating independently.
If you are in a federation of unequals, with built in dependencies… your ‘federation’ is much more like a king with vassal states, not a voluntary association.
Yes, migration of a user account from one instance to another would be complicated… but … so would migrating a user from one PDS to another.
I don’t even know how you could fully ‘migrate away from BlueSky servers’… when BlueSky run the only Relays.
Also, many (most?) actual client apps for viewing lemmy, posting on it, etc… they pretty much hold a lot of your particular user customizations, at least as it comes to visual theming, independently, locally, not even related to the actual user account on an instance you are using.
They also support easy switching between different lemmy user/instance accounts…
…
Also also, as far as I am aware… if you have an account on a lemmy instance, you can delete your account and this will wipe out all of that account’s posts and comments across the whole fediverse, aside from modlogs and internet archive web snapshotting type stuff.
I … think you can also export your own data as well?
Not 100% sure on these last two parts, maybe an instance admin or powermod could chime in… but I think this is correct?
They are fundamentally different, the whole ActivityPub federation vs ATProtocol decentralization has been talked to death in technical detail.
Not true. Bluesky has PDS migration in its design. In ActivityPub it is simply not possible
I mean…
This is literally the first thing you see on the page you just linked.
And it was last updated 7 months ago.
So I think you mean to say that account migration in BlueSky is currently in development, and is problematic and essentially experimental, and maybe sometime in the future this will change but also maybe not, who knows.
You are right though that is not possible in ActivityPub.
That is why I said design and not implemented and perfectly working
That there are actually multiple relays. There’s no hard coded single relay, that would be ridiculous and idk why people keep repeating it
There is a hard coded relay in the official bluesky app, just like it has a hard coded moderation service. But both of those are changeable with third party appviews/clients
I was oversimplifying a bit such that it wouldn’t be overwhelming to a self-described uninformed person asking for an explanation.
Yes, there are multiple actual relays but they functionally constitute a single layer or class of components in a birds eye view of the whole system.
As far as I am aware, no one other than BlueSky runs the relays, or has the code to do so.
If I am wrong about that, I would appreciate a source indicating such.
Does anyone other than BlueSky actually run a relay?
Several people have self hosted relays. Afaik nothing that anyone has used in “production”, everyone just uses the default one. I expect that will change as people figure it out, and trust in bsky pbc drops with things like the current Turkish censorship incident
Example of self hosting https://bsky.app/profile/why.bsky.team/post/3lkwg2djrfk23
The code to run a relay is here https://github.com/bluesky-social/indigo
Your “example of self hosting” is not an example of self hosting the relay, just an appview which is still being fully dependent of other Bluesky services like the relay. It’s pretty unlikely that the relay would be at all practical to host on a RPi5. But even if it was the problem still remains that the network is set up in a way where self-hosting it only results in you creating your own separate bubble, not meaningfully participating in the official one.
I also doubt anyone has selfhosted relays long-term since right now there’s very little purpose to that and the resource requirements are massive as well as keep growing at a fast pace in terms of the disk space required.
Can you explain what do you think “backfill” means in the context of the linked post?
Sorry if that sounds disrespectful but we kinda need to have shared definitions for stuff
I have zero need to play games with you. Make your case if you have one.
Backfill means that the AppView has to request and download and then be able to present… the entire history of all posts from everyone on BlueSky.
If you are familiar with crypto, its like how you have to download either the entire blockchain, or nowadays, a trimmed down/compressed version of it… before you can interact with it.
If you are familiar with any kind of database like a forum or something… when migrating, you have to actually import a copy of all the preexisting users, posts, forum structure, posts, etc… if you want the new forum to actually contain what the old forum did, before you allow people to start making new posts.
When this rando is setting up his own AppView… he is asking the BlueSky Relays to give his AppView all the older posts, before the AppView is caught up, and can then begin to function in realtime with the rest of the network.
I don’t mean to be rude, but if you genuienly don’t know what ‘backfill’ means in this context, it is very likely you have essentially zero experience with or knowledge of systems that involve large databases … it is a very common and well known term to anyone with basically over a year of doing most kinds of db admin/server admin work.
… Yeah, as 73ms already pointed out… that first link is just someone setting up an AppView.
To truly run an independent BlueSky system… you would have to run your own PDS, your own Relay, and your own AppView.
Your second link does actually have code and a rough setup guide to running your own Relay, so I will give you thanks and credit for showing that at least it is possible to theoretically do this…
But you say ‘several people run their own Relays’ and then do not evidence that.
The Relay config here is just… how to host your own Relay that would act as a member of BlueSky’s Relay network.
Basically, that is just how to transfer some of BlueSky’s server hosting costs … to yourself.
If you set up a totally independent Relay… could it even interface with BlueSky’s Relays?
As far as I can tell: No.
It would be totally independent… a parellel network, not a federated one that interfaces with the rest of BlueSky, and is thus not actually able to ‘federate’.
What… you would have to do… is set up your own Relay, connect it to basically all the other preexisting PDSs you want to include, then also run your own PDS, then also run your own AppView, and connect it to your own Relay… or just trust someother person running their own AppView, or just trust the official ones.
(But… I think that to connect your own Relay to preexisting PDSs… that would require those PDSs to… disconnect from the mainline BlueSky Relay system… because they can only point to one Relay system at a time… so that’s kind of a problem.)
That would be the only way to make your own … sort of branch of the BlueSky system, that at least in theory might be resistant to centralized censorship from BlueSky.
And again… I am not aware of anyone who has yet done this, or if it would even work at a technical level.
When dealing with software and tech companies, a good rule of thumb is that a planned or possible feature… doesn’t actually exist untill its been provably demonstrated to exist and work.
What is the advantage of Bluesky’s model over Xitter? Are they just outsourcing servers while still holding censorship and manipulation power?
As I see it the only advantage is that it is not run by Elon Musk.
And by ‘advantage’ I mean the ‘advantage’ of using a corporate product that, so far, is doing its best to drive people away from an actually censorship resistant Fediverse, using inclusive rainbow capitalist language to lure in the large majority of people who are not tech savvy enough to realize they are basically lying to / misleading them.
My first time seeing this and I love it. I’m going to assume its pronounced ‘shitter’ and you can’t convince me otherwise if its not.
That is indeed the intended pronunciation.
That Relay chokepoint is a serious architecture flaw, even for the central company running it (Bluesky). They might fix it in the future, but I doubt it’s high priority for them.
For those who don’t know, Bluesky isn’t really federated. The only way to host a non-Bluesky instance required 1TB of storage in July 2024, and 5 TB of storage in Nov 2024. Could be way more than that now.
You basically have to be a company to federate into the ATProto (Bluesky) ecosystem. You can’t just “stand up an instance”.
Lots of detail: https://dustycloud.org/blog/how-decentralized-is-bluesky/
(I know you’ve already realized that you were conflating Mastodon with Bluesky, I’m putting this here for others who come along so they can get the facts).
5TB requires you to be a company??? My personal NAS already has 92TB
That’s not an outlandish amount of storage. You can get more than that for $200.
At the rate it’s growing, it’s going to get outlandish very quickly.
It’s not an outlandish amount, but for instance I have my own VPS where I host a variety of services, and it still has under 1TB storage. Most hobbyists who rent a VPS would have less storage than that.
Why rent? If you have fiber and aren’t behind CGNAT you can host from your home
I rent because of government surveillance; I want my server in a different country.
Yeah anyone who runs a node is laughing at those numbers
My Jellyfin server is 6 times that… And my gaming PC is double that… Seriously, this person thinks 5TB is a lot? Don’t we have SD Cards/Flash Drives this big now? I’d be WAY more concerned about the bandwidth requirements.
Edit: laughing my ass off at the downvotes. Yes, my server has 30TB. Yes my PC has around 12TB. It wasn’t expensive or hard. The hard drives in my Jellyfin are NAS drives… Bunch of people acting like you need quantum computers to run a node lmfao. Storage space is easy. It’s the networking and bandwidth part that’s hard. So yeah, complaining that 5TB of storage puts it out of reach of the average person when one 12tb NAS drive cost $200? Just bitching. Plain and simple.
its still not a small amount of storage. and no, there’s still not really sd cards or flash drives bigger than 1tb, but obviously even if there were and they were super cheap, that would still never suffice as server storage. plus, if you’re hosting a node you’d want at least 4 or 5 times that storage to use a raid 5 or 6 array + at least one onsite backup, and one off-site backup.
now we’re talking thousands of dollars in equipment just for storage, not the actual server itself, internet connection, etc.
You literally just described my Jellyfin, minus the raid because I don’t feel like setting it up. Think all in all I’m down about $1200 for it. Not thousands. You do realized a 12TB NAS drive is $200, right? Only reason my build cost as much is because I have a few 2TB ssds in there which were just leftovers from the PC anyways. I could’ve done it all for $500.
Off-site backup isn’t required. Nice, but not required at all. In the literal sense, you don’t need it. It’s good to have, but an extra.
So yeah, 5TB, literally the only metric I was discussing, isn’t much. Maybe in the future the person should say all the nuance and not “5TB is unreasonable for the average person”. It’s not. Plain and simple.
maybe your hobbiest server doesn’t need a off-site backup but an instance of a massive social media network expected to be used by many users absolutely will. and sorry, but your nas simply will not cut it as far as throughput goes. it’s just not designed for that much activity.
You keep missing the point so hard I think it’s intentional.
First off, necessary and recommended are two very different things. 4chan running for the last decade on outdated software with no backups is proof that you absolutely can run things without a backup. It’s not wise, but not REQUIRED.
Secondly, OP was up there acting like 5Tb is prohibitively expensive and is gonna keep instances from being made. As me and other hobbyist have pointed out, 5tb is a joke. Those of us running little bullshit servers have WAY more. So asking someone trying to set up a social media server to have 5TB is nothing. If that seems like a lot to you, then you shouldn’t even try because it’s clearly way beyond your depth.
your home computers would probably not have the reliability or the disk performance required to run it.
it keeps constantly growing by terabytes and needs to be fast too though. Means you’re going to pay more than most private individuals are able to long-term just for the privilege of running that one component.
That’s only if you want to maintain a full archive. You don’t actually have to store a full archive to run a relay
source for that?
https://bsky.app/profile/bnewbold.net/post/3lkpdjgj5pk2i
You’re right that Bluesky isn’t federated, but it most definitely is centralized.
The answer it’s, they’re neither thing right now. And the claim has been made that in order to run your own instance that forwarded all traffic generated by the primary instance, you would need equivalent hardware to what BlueSky currently has. Vs Mastdon, which is…
interesting! so i’m probably conflating my expectations for bluesky with lemmy, when all the while i should actually be on mastadon. i was starting to wonder if bluesky was just a new us dem party project :\
Wouldn’t your “home” server in an activity pub network always be subject to such requests?
The difference is that if your home server is outside of Turkey then you can tell them to kick rocks. Bluesky probably complies because they don’t want to be blocked from Turkey. In a truly decentralized system like activitypub, only the server hosting the account / content in question risks being blocked, which means almost nothing the closer you get to a single account instance. Meanwhile every other server not in Turkey would not notice a difference.
Edit: this was under the assumption that they took it down completely, but it looks like they only geofenced it. Regardless, if they are pressured enough they would be capable of completing hiding an account worldwide, which isn’t possible with activitypub without the legal alignment of every instance’s country since bluesky on the other hand has sole control of the only relay.
I’m not an expert on how activity pub works, but… You’re saying if I had an account on mastodon.social, and if mastodon.social took down a post from my @user@mastodon.social account that, regardless of takedown reason, it would still be visible from other instances?
I’m trying to understand precisely where the resiliency lies.
I’m saying that if your home server (mastodon.social in your example) is outside of Turkey, then there is less reason for them to comply in the first place because they only risk the mastodon.social server being blocked in Turkey. That one is a bad example because they’re one of the largest and they might have a bunch of users in Turkey, so if you want to be extra safe, you’d want to pick a server that isn’t so big so that they are less likely to care about complying with some other county that they might not have any users from.
If the server you use is based inside the country that has a problem with your content, then you’d be screwed - though all the other servers will still mirror and cache your content for a bit even if you get taken down.
The resiliency lies in the fact that you can choose to register in a country that is politically friendly towards your posts or if your home country is friendly but you want to avoid being taken down, you can self host a single user instance and refuse any requests from other countries.
Edit: Now that I think about it, there’s also the fact that as long as the account itself isn’t limited by their home server, the content in question would be accessible through the federated copies, so if the home server isn’t within Turkey / jurisdiction and doesn’t take down the account, the country trying to take down the content would need to send takedown requests or request to geofence the content to each individual server on the entire fediverse - since the home server would be freely federating it to every server with users who follow the content, otherwise they would need to block every fediverse server and every new one every day that more pop up.
deleted by creator