When discussing Bluesky, people on here really need to stop thinking in terms of instances, i.e.
-
When discussing Bluesky, people on here really need to stop thinking in terms of instances, i.e. running the entire stack themselves.
If many people run only parts of it -- and this is happening now that running relays has become easier -- and members of the network spread out, then that's perfectly fine. If your relay does some bullshit, you can take your PDS to a different one and get to keep your entire history. Relay admins and PDS admins are less incentivized to play stupid blocking politics because neither are a useful boundary for communities. This is better than what we have here.
The real problem is getting people to spread out to different PDSes and relays because most simply don't care about decentralization. This is pretty much like the "mastodon.social vs. the rest of the fediverse" situation, just more severe. -
When discussing Bluesky, people on here really need to stop thinking in terms of instances, i.e. running the entire stack themselves.
If many people run only parts of it -- and this is happening now that running relays has become easier -- and members of the network spread out, then that's perfectly fine. If your relay does some bullshit, you can take your PDS to a different one and get to keep your entire history. Relay admins and PDS admins are less incentivized to play stupid blocking politics because neither are a useful boundary for communities. This is better than what we have here.
The real problem is getting people to spread out to different PDSes and relays because most simply don't care about decentralization. This is pretty much like the "mastodon.social vs. the rest of the fediverse" situation, just more severe.@volpeon I do wonder how the communication works between PDSes that are using different relays - I know that communication isn't 100% blocked cause you can just directly query a PDS. That being said, my current (and likely incorrect) understanding of ATProto aren't different relays a bit disjoint from each other and you can't really discover someone whose PDS is using a different relay to yours. Would be appreciated to know in which ways I am misunderstanding the protocol.
-
When discussing Bluesky, people on here really need to stop thinking in terms of instances, i.e. running the entire stack themselves.
If many people run only parts of it -- and this is happening now that running relays has become easier -- and members of the network spread out, then that's perfectly fine. If your relay does some bullshit, you can take your PDS to a different one and get to keep your entire history. Relay admins and PDS admins are less incentivized to play stupid blocking politics because neither are a useful boundary for communities. This is better than what we have here.
The real problem is getting people to spread out to different PDSes and relays because most simply don't care about decentralization. This is pretty much like the "mastodon.social vs. the rest of the fediverse" situation, just more severe.@volpeon I’ve been running a PDS for fun and it’s been interesting, but I’ve seen from the mouths of users on bsky that they don’t care about decentralization, they just want log on and post
I’ve tried to argue that not caring about where your data is and who has control over your online presence is dangerous, especially now
But to these people this is just “nerd shit” and I’m not sure they can be swayed
-
When discussing Bluesky, people on here really need to stop thinking in terms of instances, i.e. running the entire stack themselves.
If many people run only parts of it -- and this is happening now that running relays has become easier -- and members of the network spread out, then that's perfectly fine. If your relay does some bullshit, you can take your PDS to a different one and get to keep your entire history. Relay admins and PDS admins are less incentivized to play stupid blocking politics because neither are a useful boundary for communities. This is better than what we have here.
The real problem is getting people to spread out to different PDSes and relays because most simply don't care about decentralization. This is pretty much like the "mastodon.social vs. the rest of the fediverse" situation, just more severe. -
@volpeon I do wonder how the communication works between PDSes that are using different relays - I know that communication isn't 100% blocked cause you can just directly query a PDS. That being said, my current (and likely incorrect) understanding of ATProto aren't different relays a bit disjoint from each other and you can't really discover someone whose PDS is using a different relay to yours. Would be appreciated to know in which ways I am misunderstanding the protocol.
@makiki Relays can talk to each other, otherwise you wouldn't see any Wafrn posts on Bluesky (they use the Blacksky relay). I don't know how specifically this works, though. I would assume that because a relay merely bundles messages from multiple PDSes, it can also act like a PDS and tell other relays about its existence so they subscribe to it.
-
@makiki Relays can talk to each other, otherwise you wouldn't see any Wafrn posts on Bluesky (they use the Blacksky relay). I don't know how specifically this works, though. I would assume that because a relay merely bundles messages from multiple PDSes, it can also act like a PDS and tell other relays about its existence so they subscribe to it.
-
@makiki Relays can talk to each other, otherwise you wouldn't see any Wafrn posts on Bluesky (they use the Blacksky relay). I don't know how specifically this works, though. I would assume that because a relay merely bundles messages from multiple PDSes, it can also act like a PDS and tell other relays about its existence so they subscribe to it.
Relays are not like that.
A relay asks to all the pds it knows “hey tell me whenever you get a post”
We do by default the blacksky, but the bsky one also knows about us and asks about us
Im gona use this comparason again:
Fedi is a decentralised platform the same way trans are a transport platform
Bluesky is blimps: “what do you mean you dont need rails”
Now every fedi user I have talked about bluesky has committed the same mistake (the other way arround too btw!)
-
Relays are not like that.
A relay asks to all the pds it knows “hey tell me whenever you get a post”
We do by default the blacksky, but the bsky one also knows about us and asks about us
Im gona use this comparason again:
Fedi is a decentralised platform the same way trans are a transport platform
Bluesky is blimps: “what do you mean you dont need rails”
Now every fedi user I have talked about bluesky has committed the same mistake (the other way arround too btw!)
@gabboman @volpeon @makiki the TLDR here is that "you do not 'go to' a relay, a relay 'comes to' you". extremely common mistake
'moving a pds to a different relay' does not really mean anything as the PDS relay configuration simply means your pds politely tells that relay (through a public, unauthenticated endpoint anyone can point curl on your behalf to) it exists. the relay may or may not end up caring -
@gabboman @volpeon @makiki the TLDR here is that "you do not 'go to' a relay, a relay 'comes to' you". extremely common mistake
'moving a pds to a different relay' does not really mean anything as the PDS relay configuration simply means your pds politely tells that relay (through a public, unauthenticated endpoint anyone can point curl on your behalf to) it exists. the relay may or may not end up caring@gabboman @volpeon @makiki if someone from a relay that does not know about you replies to, that relay may very well decide to end up start listening to you as well. so as long as one relay knows about you then the rest will get to you automagically soon enough. some do also bootstrap themselves by going through every entry in did:plc (which is centralized still, so easy to do) -
@gabboman @volpeon @makiki if someone from a relay that does not know about you replies to, that relay may very well decide to end up start listening to you as well. so as long as one relay knows about you then the rest will get to you automagically soon enough. some do also bootstrap themselves by going through every entry in did:plc (which is centralized still, so easy to do)@gabboman @volpeon @makiki but again the important decentralization part is the appview, there are not many of yet. this is getting easier with things like constellation and slices which may end up resulting in "appviewless" apps that build on top of those general purpose indexes (like tangled.sh/@whey.party/red-dwarf)
you'd still need people to rehost them but you'd get more functionality for the resources you spend on that compared to rehosting the bluesky appview -
When discussing Bluesky, people on here really need to stop thinking in terms of instances, i.e. running the entire stack themselves.
If many people run only parts of it -- and this is happening now that running relays has become easier -- and members of the network spread out, then that's perfectly fine. If your relay does some bullshit, you can take your PDS to a different one and get to keep your entire history. Relay admins and PDS admins are less incentivized to play stupid blocking politics because neither are a useful boundary for communities. This is better than what we have here.
The real problem is getting people to spread out to different PDSes and relays because most simply don't care about decentralization. This is pretty much like the "mastodon.social vs. the rest of the fediverse" situation, just more severe.I have gotten one dismissive reply, so let me add something: Blind fanaticism isn't just unnecessary, it actively prevents positive change from happening because it refuses acknowledgement of problems that always do exist. Nothing is perfect.
I have criticized the fediverse a lot over the years. Especially the instance = community boundary is a huge problem because they serve as both a social space and a gateway to the network all in one package. However, community matters and provider matters are very different. If an instance defederates from another, people don't just lose access to that community - they lose access to communicating with people on an instance entirely until they create a new account elsewhere.
Imagine if a Discord server admin banned you and Discord also blocked your communication with the server members even in other servers or 1-on-1. That's fedi.
This means your choice of an instance is extremely important and happens at a time when you don't even know anything about the fediverse at all.
This is why I think what absolutely needs to happen here is to give instances less importance by implementing communities on a higher layer, independent of those boundaries. Is this happening? I doesn't look like that to me. Hell, we don't even improve on the microblogging formula in a meaningful way. I think the ecosystem at large has become complacent with the existing structure and has no interest in making fundamental changes.
ATProto seems closer to the ideal I have in mind, so obviously I've taken an interest in it.
(Improved wording) -
@gabboman @volpeon @makiki but again the important decentralization part is the appview, there are not many of yet. this is getting easier with things like constellation and slices which may end up resulting in "appviewless" apps that build on top of those general purpose indexes (like tangled.sh/@whey.party/red-dwarf)
you'd still need people to rehost them but you'd get more functionality for the resources you spend on that compared to rehosting the bluesky appview -
-
-
I have gotten one dismissive reply, so let me add something: Blind fanaticism isn't just unnecessary, it actively prevents positive change from happening because it refuses acknowledgement of problems that always do exist. Nothing is perfect.
I have criticized the fediverse a lot over the years. Especially the instance = community boundary is a huge problem because they serve as both a social space and a gateway to the network all in one package. However, community matters and provider matters are very different. If an instance defederates from another, people don't just lose access to that community - they lose access to communicating with people on an instance entirely until they create a new account elsewhere.
Imagine if a Discord server admin banned you and Discord also blocked your communication with the server members even in other servers or 1-on-1. That's fedi.
This means your choice of an instance is extremely important and happens at a time when you don't even know anything about the fediverse at all.
This is why I think what absolutely needs to happen here is to give instances less importance by implementing communities on a higher layer, independent of those boundaries. Is this happening? I doesn't look like that to me. Hell, we don't even improve on the microblogging formula in a meaningful way. I think the ecosystem at large has become complacent with the existing structure and has no interest in making fundamental changes.
ATProto seems closer to the ideal I have in mind, so obviously I've taken an interest in it.
(Improved wording)@volpeon I never took the "dictator admin" thing seriously because not only does that sound childish at face value, but in my experience, what I find when I go looking for myself because internet drama queens love to keep it vague - it's racism. It's white fragility. That's what gets called petty and stupid on here.
If I'm not taking you or any other posts like this seriously, it's not because I'm a blind fanatic, it's just that I don't trust you guys. I have to be careful.
As a general rule, if a story is exaggerated, out of context, sounds fictional, and doesn't show any proof, then I doubt it.
-
@gabboman @volpeon @makiki but again the important decentralization part is the appview, there are not many of yet. this is getting easier with things like constellation and slices which may end up resulting in "appviewless" apps that build on top of those general purpose indexes (like tangled.sh/@whey.party/red-dwarf)
you'd still need people to rehost them but you'd get more functionality for the resources you spend on that compared to rehosting the bluesky appview -
appviews get the data from relays and pds and do store them, and you query them instead of queriying the pds directly.
they also do some extra shit like "wait potatohater69 is blocked by mrpotato, i should deatach this reply from the thread"
-
appviews get the data from relays and pds and do store them, and you query them instead of queriying the pds directly.
they also do some extra shit like "wait potatohater69 is blocked by mrpotato, i should deatach this reply from the thread"
-
so is fedi for people who knows atproto!
seriously, both are decentralized architectures but knowing one doesnt helps to understand the other one. in fact it makes it worse
-
so is fedi for people who knows atproto!
seriously, both are decentralized architectures but knowing one doesnt helps to understand the other one. in fact it makes it worse
@gabboman @volpeon @kopper
With AP there are at least some similarities to e-mail to rely on for the high level understanding and I'd hazard a guess most of the pain with AP is in the interop between servers (ex. wafrn <-> mastodon or akkoma <-> sharkey, or even kbin <-> gts) rather than the high level idea.I wouldn't be surprised if that interop is way less painful with ATProto thanks to relays and shit due to more stuff having to be actually standarized to make this work.
