Does it will have the features of Manyverse?

[hardBSDk]

Manyverse is the most advanced and active client of the SSB (Secure Scuttlebutt) protocol but is abandoned, do you plan to be an alternative for it?

They were moving further away from the SSB protocol because the replication system created too big databases.

[dpc]

I don't have some big plans for Rostra. Generally networks like this live and die by their popularity, and it's hard to compete with network effects of more established projects.

Under the hood is works quite similarly to SSB.

[cryptoquick]

There's no such thing as a too big database, only incompetent developers

[dpc]

@hardBSDk I was actually curious myself about this aspect. Do you have any more details w.r.t what was the issue?

[dpc]

@cryptoquick Let's not assume I was not incompetent just yet. :D

There is no escaping the fact that in a p2p network nodes replicate the storage, and especially for media it adds up. 0.5M photo here and there, 50MB video, across 300 people you follow, and we're talking gigabytes very fast. Text is lightweight, but media is not.

A centralized social network can largely deduplicate this. Store every media once (well, with redundancy) and serve on demand to all 1k followers. A p2p social network ... will just create 1k copies on every node.

I do have some ideas how this could be improved:

• fetching things on demand p2p,
• probabilistic sharding storage (so e.g. a follower of identity with 1k followers would store one in N things)

but none of this is actually implemented.

[hardBSDk]

@hardBSDk I was actually curious myself about this aspect. Do you have any more details w.r.t what was the issue?

The Manyverse replication system don't allow removal to free space ("persistence" concept) which would make people store a ever growing database, the "persistence" concept is to increase local social network trust by not removing previous statements, it works as long the local social network is small.

Sorry for the very late reply.

[hardBSDk]

@cryptoquick Let's not assume I was not incompetent just yet. :D

There is no escaping the fact that in a p2p network nodes replicate the storage, and especially for media it adds up. 0.5M photo here and there, 50MB video, across 300 people you follow, and we're talking gigabytes very fast. Text is lightweight, but media is not.

A centralized social network can largely deduplicate this. Store every media once (well, with redundancy) and serve on demand to all 1k followers. A p2p social network ... will just create 1k copies on every node.

I do have some ideas how this could be improved:

* fetching things on demand p2p,

* probabilistic sharding storage (so e.g. a follower of identity with 1k followers would store one in N things)

but none of this is actually implemented.

And compression, probably with Zstd.