Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

At Protocol: Building the Social Internet

atproto.com

Post reply on HN
Print view View original

31–40 of 43 posts

Re: At Protocol: Building the Social Internet

#31

Post by blackqueeriroh »

NiallBunting wrote: post #27
nightpool wrote: post #22

Earlier quoted context omitted.

The ability to make information private fundamentally conflicts with how ATProto is designed. All records have to be sent to all Relays and AppView nodes on the network to provide a "global view" of the network. So there's no way to keep records private without locking out some user's servers from viewing them, and since AppViews are centralized indexing services, they won't function without being able to see the ent…

Yeah, apps wouldn't be able to only listen to the firehose. There are some proposals for private files. However, I'm outside the AtProto world so not sure what exactly the suggested implementations are. I just hope they give enough control. I think the technology could potentially be used for way more than microblogging. I would love to use webapps that store the data on my devices and share it with specific people.…

What you linked is by Bluesky’s own staff - it’s him laying out the design plans for the protocol’s private data implementation.

He’s doing it to engage publicly with the community, since many of us also build on the protocol.

Re: At Protocol: Building the Social Internet

#32

Post by blackqueeriroh »

numpad0 wrote: post #23
bladeee wrote: post #16

How common is it to host your own PDS? It's not really clear to me what the advantages/disadvantage are.

The advantage of Bluesky over Twitter and Mastodon is that it's centralized with no actual plan to decentralize.

This is false. Every single people of the atproto stack can and has now been hosted and deployed by others and is operating at less get than the scale of a small community.

Re: At Protocol: Building the Social Internet

#34

Post by tardedmeme »

embedding-shape wrote: post #5
thesuitonym wrote: post #4

I often wonder if people have forgotten that you can send information through the internet without HTTP(S)

Since others starting intercepting and eavesdrop on non-authenticated traffic, it got a lot less comfy to do so though. It's not like people are adding encryption to stuff for the fun of it.

I often wonder if people have forgotten that you can use TLS through the internet without HTTP

Re: At Protocol: Building the Social Internet

#35

Post by tardedmeme »

outside1234 wrote: post #24
_kst_ wrote: post #19

The summary of what this is about is: "Atproto is a big-world open social protocol. Users publish JSON records into repositories. The changestreams of those records then sync across the network to drive applications." It's too bad that information isn't on the front page. You have click "GET STARTED" and scroll down

Is it really, at the end of the day, a server that serves your data over JSON (aka a social media post) and offers feeds (ala an RSS model) of the changes of that data and provides services for discovering the schema of that JSON (aka its lexicon)?

Yes

Re: At Protocol: Building the Social Internet

#36

Post by tardedmeme »

nightpool wrote: post #22
NiallBunting wrote: post #6

I think it would be interesting if the file servers had read/write/list/delete permissions on files. For both groups and users. It would mean the public stuff could see your files but private projects could exist. Eg. Maybe I don't want my At Protocol version of Figma making all my drawings public. If they could be shared in a group (anyone in a list in that folder or whatever). Maybe this is coming, but would intere…

The ability to make information private fundamentally conflicts with how ATProto is designed. All records have to be sent to all Relays and AppView nodes on the network to provide a "global view" of the network. So there's no way to keep records private without locking out some user's servers from viewing them, and since AppViews are centralized indexing services, they won't function without being able to see the ent…

What would happen if you returned different content depending on who was asking?

Re: At Protocol: Building the Social Internet

#37

Post by tptacek »

thesuitonym wrote: post #4

I often wonder if people have forgotten that you can send information through the internet without HTTP(S)

In addition to the middlebox problem, most (not all, but most) of the things that send information over the Internet that aren't HTTP-shaped (incl. HTTP/2 and HTTP/3) are worse than the best HTTP-shaped things. This makes sense: HTTP-shaped things are where all the energy is directed.

Re: At Protocol: Building the Social Internet

#38

Post by idiotsecant »

blackqueeriroh wrote: post #30
yawnxyz wrote: post #28

Earlier quoted context omitted.

though technically your data lives within Bluesky's servers (kind of like your money lives in a bank) so in some ways it's still the bank's money though it's allotted to you

Actually your data does not live in Bluesky’s servers if you don’t want it to. Mine doesn’t. At least another 20-30,000 people’s doesn’t either.

What keeps bluesky from simply catching the data?

Re: At Protocol: Building the Social Internet

#39

Post by esbranson »

The W3C equivalent PDS efforts are in the LWS WG.[1][2] And of course other W3C standards like DIDs, VCs, and ActivityStreams/ActivityPub. W3C is still missing standards around data replication and a global indexing / feed layer.

[1] https://www.w3.org/groups/wg/lws/

[2] https://www.w3.org/TR/lws-ucs/

Re: At Protocol: Building the Social Internet

#40

Post by skrtskrt »

g4t0r wrote: post #29

Someone posted this article[0] earlier today and I thought it was a really good primer on AT-stuff I've been playing with my own PDS and AppView for a side project, and it's really fun and interesting. [0] - https://overreacted.io/a-social-filesystem/

That article should be the official ATProto docs at this point. The current docs are hilariously self-referential and uses terminology that is either way too broad and overloaded for the component they are supposed to describe, or so far misaligned with the standardized computer science concepts that it seems intentionally opaque.

It reads like the devs think they invented a whole new world of computing and not just a little protocol.

Post reply on HN