4 ms·

I have also thought about this idea that users of a social network could keep all their data and just let different clients (Facebook, Twitter) access it. I thi

by ninetax 14y ago

I have also thought about this idea that users of a social network could keep all their data and just let different clients (Facebook, Twitter) access it. I think it would be a great idea! If I want to change my email address? Change it in one place. If I want to let my friends know something? Update it in one place. It's very DRY you know? But then I think, why even let Facebook use my data? If we have these data stores let's build a peer to peer network to let trusted friends access our information. In real life, if I want to tell my friend something I don't pass it through a 3rd party first, I tell them directly! This can let us have really fine grained control over who we share with by authenticating the users that request the information. And with the authentication can come encryption. This move to the cloud is, frankly, annoying. Why should we have to trust our data to all these people we don't know? They pretend that they give us "free" service, while actually using that data for profit. Good UI shouldn't have to come at such a cost. I have been thinking about this for a while, if there are other people interested in building such a web I would love to know.

nkoren 14y ago
I've been thinking about this for a while too, and I agree that the peer-to-peer element is essential. So far, it appears that all the would-be open social media apps/protocols/whatever rely on you having your own server. That, frankly, is a deadly failing. Per Metcalfe's Law, social networks are only valuable when they are maximally inclusive, and the requirement to have your own server is something that will always exclude the majority of (non-technical) users. The need for easy non-technical access means that there are only two viable paths for any social media application. Either: 1.) Go to web page and use application, or 2.) Download app and use application This is how Facebook / Twitter / etc. do it., and they're clearly doing it right. But there's no law which says that the application needs to keep its data on a third-party server. Why not have it run a Peer-to-peer service? In fact, I think that Peer-to-peer is especially suitable to social networking, given that social networks are typically highly clustered (eg., people who are my friends are much more likely to be each others' friends as well). This means that if the every client application is able to redistribute copies of friends' status updates as well as its own status updates, then a sync between any given pair of clients is likely to allow them to bring a large subset of their friends' statuses up to date. Of course if you're rebroadcasting your friends' status updates as a P2P host, then the question of spoofing/authentication becomes paramount. Obviously that would have to be solved by having all status updates transmitted in encrypted formats which can only be decrypted / authenticated with a users' public key. Oh, and once you've got dual-key authentication embedded in widely-used social trust networks, you could probably solve a lot of other authentication problems while you're at it. All of this would need to be completely transparent to the user, of course. If it's any more difficult to set up and use than standard Facebook / Twitter, it will fail. But I don't see any fundamental obstacles to making it that easy. Unfortunately I don't have time to do anything about this!
- ninetax 14y ago
  And even if it is just as easy, or easier, to set up as facebook or twitter, it still might fail because of the user and the egg problem. As much as I would love to use and have a network like this, until all the people I keep in touch with through facebook use it, I'll still have a facebook. I wonder if it's plausible, even possible, to have all those p2p benefits (encryption, authentication, privacy) and some how also have interoperability with facebook?
  - cyarvin 14y ago
    Can you make a Tent (or more ambitious social p2p) client which is also an FB client? Of course. FB and Twitter have APIs. Of course, they can revoke your key. But they have web clients, so they can be scraped. Faced with a genuinely distributed opponent, there's no way the existing behemoths can keep your data in their silos. For instance, FB can block tent.is at the API level or even at the IP level. But if Tent hosts pop up all around the internets, and if they are general-purpose enough that users can install their own scraping gateways, which can't be attacked centrally using technical or legal means... it's game over. To me this is one of the less recognized advantages of a distributed service.
    - maximveksler 14y ago
      I've been thinking about this idea a lot recently. I'm wondering if you would like to have a discussion? If so, please email me at maxim.hn.4418904@vekslers.org Maxim.

by ninetax 14y ago

nkoren 14y ago

I've been thinking about this for a while too, and I agree that the peer-to-peer element is essential. So far, it appears that all the would-be open social media apps/protocols/whatever rely on you having your own server. That, frankly, is a deadly failing. Per Metcalfe's Law, social networks are only valuable when they are maximally inclusive, and the requirement to have your own server is something that will always exclude the majority of (non-technical) users. The need for easy non-technical access means that there are only two viable paths for any social media application. Either: 1.) Go to web page and use application, or 2.) Download app and use application This is how Facebook / Twitter / etc. do it., and they're clearly doing it right. But there's no law which says that the application needs to keep its data on a third-party server. Why not have it run a Peer-to-peer service? In fact, I think that Peer-to-peer is especially suitable to social networking, given that social networks are typically highly clustered (eg., people who are my friends are much more likely to be each others' friends as well). This means that if the every client application is able to redistribute copies of friends' status updates as well as its own status updates, then a sync between any given pair of clients is likely to allow them to bring a large subset of their friends' statuses up to date. Of course if you're rebroadcasting your friends' status updates as a P2P host, then the question of spoofing/authentication becomes paramount. Obviously that would have to be solved by having all status updates transmitted in encrypted formats which can only be decrypted / authenticated with a users' public key. Oh, and once you've got dual-key authentication embedded in widely-used social trust networks, you could probably solve a lot of other authentication problems while you're at it. All of this would need to be completely transparent to the user, of course. If it's any more difficult to set up and use than standard Facebook / Twitter, it will fail. But I don't see any fundamental obstacles to making it that easy. Unfortunately I don't have time to do anything about this!

ninetax 14y ago
And even if it is just as easy, or easier, to set up as facebook or twitter, it still might fail because of the user and the egg problem. As much as I would love to use and have a network like this, until all the people I keep in touch with through facebook use it, I'll still have a facebook. I wonder if it's plausible, even possible, to have all those p2p benefits (encryption, authentication, privacy) and some how also have interoperability with facebook?
- cyarvin 14y ago
  Can you make a Tent (or more ambitious social p2p) client which is also an FB client? Of course. FB and Twitter have APIs. Of course, they can revoke your key. But they have web clients, so they can be scraped. Faced with a genuinely distributed opponent, there's no way the existing behemoths can keep your data in their silos. For instance, FB can block tent.is at the API level or even at the IP level. But if Tent hosts pop up all around the internets, and if they are general-purpose enough that users can install their own scraping gateways, which can't be attacked centrally using technical or legal means... it's game over. To me this is one of the less recognized advantages of a distributed service.
  - maximveksler 14y ago
    I've been thinking about this idea a lot recently. I'm wondering if you would like to have a discussion? If so, please email me at maxim.hn.4418904@vekslers.org Maxim.