3 ms·

Peergos – A decentralised, secure file storage, sharing and social network

56 points by arm 11y ago

ianopolous 11y ago
Founder here, we aren't really ready for a "slashdotting" at the moment, and are in the process of creating a new website with more and better explanations, but if you have any questions we will try to answer them here.
shazow 11y ago
Looks like it's built on top of IPFS. Very cool. (Recent HN thread on IPFS in case HN readers missed out: https://news.ycombinator.com/item?id=11134766 https://news.ycombinator.com/item?id=11134766) ianpolous, can you comment on your experience of building a service on top of IPFS? How coupled is Peergos to IPFS, and how interoperable are your client implementations with potential other clients? Are there plans for third-party clients to interact with Peergos data formats over the same network?
- ianopolous 11y ago
  We actually started a year before IPFS so we built our own DHT, but then IPFS came along and did a better job (the DHT part was never our primary focus) so we adopted theirs. IPFS is a pleasure to work with, both programmatically and people-wise. Juan and Jeremy and the rest of the team are super friendly and responsive, and make great design decisions. Without sounding like I've drunk some coolaid, I really do think that IPFS is the future. Currently we use IPFS for the P2P networking and treat it as a distributed hash table, so it has a really simple interface. As soon as IPNS is mature enough we will also switch to using that over our current centralised signed hash updates. Eventually we hope to use Filecoin as well for getting others to store your fragments. We have a centralised "core node" which stores username <--> public key mappings and pending follow requests, however both of these are well suited to some kind of blockchain for full decentralization. We currently proxy all queries to IPFS through our own server (either our public server, or the one you run on your machine), but that is not a fundamental requirement. We are working on a FUSE implementation as well. Everything we do is open source so yes anyone could write their own client.
  - shazow 11y ago
    If I were going to implement my own client for a subset of Peergos, is the protocol format documented somewhere? Is it a priority to encourage this kind of developer participation, or are you going to focus on your own implementation before encouraging participation? (Bonus points for demo minimal implementations.) As another commenter said, I'm also more excited about a social network "protocol" than another social network.
    - ianopolous 11y ago
      At the moment, no, you're the first person who's asked :-). It is however a pretty simple REST api, which would be easy for us to document. The main things you'll need in the client are TweetNaCl, scrypt and an implementation of our slight simplification of cryptree [1]. We have Java and Javascript implementations (though the Java isn't quite up-to-date). [1] http://boga.googlecode.com/svn/trunk/res/Docs/wuala-cryptree.pdf http://boga.googlecode.com/svn/trunk/res/Docs/wuala-cryptree...
atrudeau 11y ago
Mmmm, I've seen dozens of decentralized social networks come and go. Hope this one makes it.
- mbreese 11y ago
  If anything will take hold, my money is on a social network protocol, with the possibility for multiple implementations rather than a distributed application. Until I see an RFC, I'm not optimistic.
  - atrudeau 11y ago
    Well, xmpp was fully specified and it went the way of the dog, and did a fraction of what a full social network does. But I like the idea. Perhaps combining xmpp with a social graph and some content blobs for photos and whatnot would be a start.
    - jjn2009 11y ago
      used to be a part of facebook chat api at one point which was probably the height of the protocol: https://developers.facebook.com/docs/chat https://developers.facebook.com/docs/chat
  - __jal 11y ago
    I fully agree. Social, like security, is a process, not a product. I'm still convinced that most of Twitter's problems would not exist if they had written an RFC instead of a business plan.
  - chc4 11y ago
    There's OStatus and ActivityStreams, which are implemented by Diaspora and GNU Social along with several other decentralized networks. Hell, OStatus is even an (extremely vague, actually horrible but whatever) W3C spec. edit: Ok, I was kinda wrong. Apparently it's just a community hub hosted on w3 and not officially affiliated. https://www.w3.org/community/ostatus/ https://www.w3.org/community/ostatus/ http://activitystrea.ms/ http://activitystrea.ms/

56 points by arm 11y ago

Founder here, we aren't really ready for a "slashdotting" at the moment, and are in the process of creating a new website with more and better explanations, but if you have any questions we will try to answer them here.

shazow 11y ago

Looks like it's built on top of IPFS. Very cool. (Recent HN thread on IPFS in case HN readers missed out: https://news.ycombinator.com/item?id=11134766 https://news.ycombinator.com/item?id=11134766) ianpolous, can you comment on your experience of building a service on top of IPFS? How coupled is Peergos to IPFS, and how interoperable are your client implementations with potential other clients? Are there plans for third-party clients to interact with Peergos data formats over the same network?

ianopolous 11y ago
We actually started a year before IPFS so we built our own DHT, but then IPFS came along and did a better job (the DHT part was never our primary focus) so we adopted theirs. IPFS is a pleasure to work with, both programmatically and people-wise. Juan and Jeremy and the rest of the team are super friendly and responsive, and make great design decisions. Without sounding like I've drunk some coolaid, I really do think that IPFS is the future. Currently we use IPFS for the P2P networking and treat it as a distributed hash table, so it has a really simple interface. As soon as IPNS is mature enough we will also switch to using that over our current centralised signed hash updates. Eventually we hope to use Filecoin as well for getting others to store your fragments. We have a centralised "core node" which stores username <--> public key mappings and pending follow requests, however both of these are well suited to some kind of blockchain for full decentralization. We currently proxy all queries to IPFS through our own server (either our public server, or the one you run on your machine), but that is not a fundamental requirement. We are working on a FUSE implementation as well. Everything we do is open source so yes anyone could write their own client.
- shazow 11y ago
  If I were going to implement my own client for a subset of Peergos, is the protocol format documented somewhere? Is it a priority to encourage this kind of developer participation, or are you going to focus on your own implementation before encouraging participation? (Bonus points for demo minimal implementations.) As another commenter said, I'm also more excited about a social network "protocol" than another social network.
  - ianopolous 11y ago
    At the moment, no, you're the first person who's asked :-). It is however a pretty simple REST api, which would be easy for us to document. The main things you'll need in the client are TweetNaCl, scrypt and an implementation of our slight simplification of cryptree [1]. We have Java and Javascript implementations (though the Java isn't quite up-to-date). [1] http://boga.googlecode.com/svn/trunk/res/Docs/wuala-cryptree.pdf http://boga.googlecode.com/svn/trunk/res/Docs/wuala-cryptree...

atrudeau 11y ago

Mmmm, I've seen dozens of decentralized social networks come and go. Hope this one makes it.

mbreese 11y ago
If anything will take hold, my money is on a social network protocol, with the possibility for multiple implementations rather than a distributed application. Until I see an RFC, I'm not optimistic.
- atrudeau 11y ago
  Well, xmpp was fully specified and it went the way of the dog, and did a fraction of what a full social network does. But I like the idea. Perhaps combining xmpp with a social graph and some content blobs for photos and whatnot would be a start.
  - jjn2009 11y ago
    used to be a part of facebook chat api at one point which was probably the height of the protocol: https://developers.facebook.com/docs/chat https://developers.facebook.com/docs/chat
- __jal 11y ago
  I fully agree. Social, like security, is a process, not a product. I'm still convinced that most of Twitter's problems would not exist if they had written an RFC instead of a business plan.
- chc4 11y ago
  There's OStatus and ActivityStreams, which are implemented by Diaspora and GNU Social along with several other decentralized networks. Hell, OStatus is even an (extremely vague, actually horrible but whatever) W3C spec. edit: Ok, I was kinda wrong. Apparently it's just a community hub hosted on w3 and not officially affiliated. https://www.w3.org/community/ostatus/ https://www.w3.org/community/ostatus/ http://activitystrea.ms/ http://activitystrea.ms/