3 ms·

Is that inevitable though? A graph of independent but connected nodes could theoretically scale indefinitely in the case where I hold my data and you hold yours

Is that inevitable though? A graph of independent but connected nodes could theoretically scale indefinitely in the case where I hold my data and you hold yours. If I want to add someone to my network, I can either get their direct address or navigate through friends-of-friends (graph traversal). The only necessary delegation would be a mailbox provider, but the contents of any messages could be encrypted. Not that any of this is easy to engineer, but in my mind it seems viable.

rakoo 7y ago
It actually does exist already, in the form of Secure Scuttlebutt (https://en.wikipedia.org/wiki/Secure_Scuttlebutt https://en.wikipedia.org/wiki/Secure_Scuttlebutt). Friends connect directly to each other. When you want to get news about someone, you download from them directly, along with news from friends of friends. If both peers aren't connected at the same time it is possible to use pubs, which are kind of hubs accepting all content and distributing to whomever wants it. Pubs have no special functions apart from being a store-and-forward, so you're not associated to a pub in particular (you can be on multiple, or none at all) and no pub is more "important" than others. Nothing is lost when a pub closes. Cryptography makes sure that only the intended reader can read (unless it's a public message), and that messages can't be forged. The nice things is that it's all just a message passing infrastructure, and applications can and have been built on top of it, like a git repository system or a music sharing application. So, yeah, it is feasible.
- kixiQu 7y ago
  The amount of data you have to store for even the network's current larval state does not to me suggest that †he current version is real practical to scale.
  - rakoo 7y ago
    I don't know, people have been on it since the beginning and have only a few Gigs in their db, which is mostly non-text (pics, videos, audio, ...). You'd need to store them in any case. It's true that initial sync is way too slow but they're working on it.
- beaconstudios 7y ago
  neat, I didn't know Scuttlebutt worked that way!
pythonaut_16 7y ago
Maybe a place for a hybrid identity solution as well? Where you could be "user@mastodon.social" but also have a more numerical/cryptographic id (like a GUID or Pub/Private key pair). One to make it easy to find and connect as users, and the second to make it easier to move between instances.
vanderZwan 7y ago
Apologies for the confusion: I mean delegation of power on the "social" front of things; "social networks" shouldn't be interpreted as Facebook or Twitter but as the literal concept of humans forming a social network through any kind of system. And yes, I believe that it is inevitable in that everyone moderating everyone doesn't scale. Your comment and replies to it makes good points about the technical side of this tricky problem. What I was getting at is the original social needs that drives the demand for these kinds of technical solutions to begin with. Those are very tricky things to figure out, especially since there are both "group needs" and "individual needs" that can be at odds.

rakoo 7y ago

It actually does exist already, in the form of Secure Scuttlebutt (https://en.wikipedia.org/wiki/Secure_Scuttlebutt https://en.wikipedia.org/wiki/Secure_Scuttlebutt). Friends connect directly to each other. When you want to get news about someone, you download from them directly, along with news from friends of friends. If both peers aren't connected at the same time it is possible to use pubs, which are kind of hubs accepting all content and distributing to whomever wants it. Pubs have no special functions apart from being a store-and-forward, so you're not associated to a pub in particular (you can be on multiple, or none at all) and no pub is more "important" than others. Nothing is lost when a pub closes. Cryptography makes sure that only the intended reader can read (unless it's a public message), and that messages can't be forged. The nice things is that it's all just a message passing infrastructure, and applications can and have been built on top of it, like a git repository system or a music sharing application. So, yeah, it is feasible.

kixiQu 7y ago
The amount of data you have to store for even the network's current larval state does not to me suggest that †he current version is real practical to scale.
- rakoo 7y ago
  I don't know, people have been on it since the beginning and have only a few Gigs in their db, which is mostly non-text (pics, videos, audio, ...). You'd need to store them in any case. It's true that initial sync is way too slow but they're working on it.
beaconstudios 7y ago
neat, I didn't know Scuttlebutt worked that way!

pythonaut_16 7y ago

Maybe a place for a hybrid identity solution as well? Where you could be "user@mastodon.social" but also have a more numerical/cryptographic id (like a GUID or Pub/Private key pair). One to make it easy to find and connect as users, and the second to make it easier to move between instances.

vanderZwan 7y ago

Apologies for the confusion: I mean delegation of power on the "social" front of things; "social networks" shouldn't be interpreted as Facebook or Twitter but as the literal concept of humans forming a social network through any kind of system. And yes, I believe that it is inevitable in that everyone moderating everyone doesn't scale. Your comment and replies to it makes good points about the technical side of this tricky problem. What I was getting at is the original social needs that drives the demand for these kinds of technical solutions to begin with. Those are very tricky things to figure out, especially since there are both "group needs" and "individual needs" that can be at odds.