7 ms·

> 1. ... For distributed systems there are more ways to exploit the APIs and gather data on users. However there is no global state, users only push-pull feeds

> 1. ... For distributed systems there are more ways to exploit the APIs and gather data on users. However there is no global state, users only push-pull feeds of friends and friends of friends. (In part for privacy and in part for performance, you don't need to carry all the data for a social network to be useful) > 3. GDPR compliance about deleting data is almost impossible in a distributed system. However that doesn't apply to individuals that aren't providing a public service. > 4. ... Open source software and distributed software tends to be much harder to use. Yes, however there has been some exiting discussion around what UX could become possible in a decentralized context. See https://coolguy.website/writing/the-future-will-be-technical/ https://coolguy.website/writing/the-future-will-be-technical... > 5. Any future concern/issue will be much harder to resolve if there are thousands of different instance running decentralized social networks. There are no instances, only peers. (Pubs are sort of easy to connect to peers and there is work being done on peer discovery to outdate most pubs; they could still be used to connect communits around topics, hobbies, etc as a way to reduce the distance between peers) > 6. Using AI to detect abusive content or spot fake news is much harder if you only have a subset of the data. There is no retweeting/sharing, though there is work being done on Out of Order Messaging to propagate a single message along the follow graph and there is work being done on flagging/tagging feeds/ids and posts. (Along with some semantics around how to interpret the flags/tags to improve UX and user control)

rapnie 9y ago
> Pubs are sort of easy to connect to peers and there is work being done on peer discovery to outdate most pubs; this is cool! could you explain more? i thought Pubs were needed to connect with peers on internal networks, to alleviate NAT traversal issues. what i would like to know more about regarding Pubs - but could not find any info on some time ago - is how much it will cost, more or less, to host your own Pub in terms of data transfer/storage, CPU, etc. and whether rate limits can be set to manage this. nice to see your docs.. i have them bookmarked for later :)
- fabianhjr 9y ago
  Hi rapnie, I suggest you try the protocol guide, it has a section on pubs: https://ssbc.github.io/scuttlebutt-protocol-guide/#pubs https://ssbc.github.io/scuttlebutt-protocol-guide/#pubs
Kalium 9y ago
> There are no instances, only peers. (Pubs are sort of easy to connect to peers and there is work being done on peer discovery to outdate most pubs; they could still be used to connect communits around topics, hobbies, etc as a way to reduce the distance between peers) Why do you draw this distinction between "instances" and "peers"? A conventional understanding of the word "instance" would suggest it means "a running instance of the software system in question", which would suggest the word has no meaning distinct from what "peer" would mean in this context. Am I missing a critical and non-obvious distinction? > There is no retweeting/sharing My experience is that users really like these features.
- strypey 9y ago
  > Why do you draw this distinction between "instances" and "peers" I understand why you ask this, but the word "instance" has come to mean the server running something like MediaGoblin or Mastodon (the server in a server/client model), while a "peer" is an end-user device running a P2P app like BitTorrent or ScuttleButt (that is both client and server). P2P networks have their cons, but the major pro for a network not backed by large wads of capital is that they scale as they add users. Server-to-server federated networks using OStatus, Diaspora protocol, Zot, or ActivityPub have to scale by convincing people with scarce sysadmin skills to set up more servers, and figure out how to sustain them (both organizationally and financially). > My experience is that users really like [retweeting/sharing] features. Sure, but that doesn't mean every piece of network software needs to have them. They are one of the key tools for abusing The Stacks; post something you want to trend (spam, fake news), get a botnet of zombie users to boost it. Given the way Scuttlebutt transports and stores posts, they would create a lot of extra overhead, and potential for abuse, for no major benefit (people can always cut'n'paste if they really want to reshare).

rapnie 9y ago

> Pubs are sort of easy to connect to peers and there is work being done on peer discovery to outdate most pubs; this is cool! could you explain more? i thought Pubs were needed to connect with peers on internal networks, to alleviate NAT traversal issues. what i would like to know more about regarding Pubs - but could not find any info on some time ago - is how much it will cost, more or less, to host your own Pub in terms of data transfer/storage, CPU, etc. and whether rate limits can be set to manage this. nice to see your docs.. i have them bookmarked for later :)

fabianhjr 9y ago
Hi rapnie, I suggest you try the protocol guide, it has a section on pubs: https://ssbc.github.io/scuttlebutt-protocol-guide/#pubs https://ssbc.github.io/scuttlebutt-protocol-guide/#pubs

Kalium 9y ago

> There are no instances, only peers. (Pubs are sort of easy to connect to peers and there is work being done on peer discovery to outdate most pubs; they could still be used to connect communits around topics, hobbies, etc as a way to reduce the distance between peers) Why do you draw this distinction between "instances" and "peers"? A conventional understanding of the word "instance" would suggest it means "a running instance of the software system in question", which would suggest the word has no meaning distinct from what "peer" would mean in this context. Am I missing a critical and non-obvious distinction? > There is no retweeting/sharing My experience is that users really like these features.

strypey 9y ago
> Why do you draw this distinction between "instances" and "peers" I understand why you ask this, but the word "instance" has come to mean the server running something like MediaGoblin or Mastodon (the server in a server/client model), while a "peer" is an end-user device running a P2P app like BitTorrent or ScuttleButt (that is both client and server). P2P networks have their cons, but the major pro for a network not backed by large wads of capital is that they scale as they add users. Server-to-server federated networks using OStatus, Diaspora protocol, Zot, or ActivityPub have to scale by convincing people with scarce sysadmin skills to set up more servers, and figure out how to sustain them (both organizationally and financially). > My experience is that users really like [retweeting/sharing] features. Sure, but that doesn't mean every piece of network software needs to have them. They are one of the key tools for abusing The Stacks; post something you want to trend (spam, fake news), get a botnet of zombie users to boost it. Given the way Scuttlebutt transports and stores posts, they would create a lot of extra overhead, and potential for abuse, for no major benefit (people can always cut'n'paste if they really want to reshare).