Imagine a new kind of VPN service which doesn't require any kind of configuration and magically just works without a central server.

Who Am I?

Before I go on, I should point out that I have no background in network infrastructure, P2P network development, or pretty much any of the specific technologies this touches. My background lies in web development, and this idea is based on my understanding of these technologies. As such, please point out any errors, misconceptions or other issues you might find. But keep in mind, this is simply an idea of a potentially awesome technology.

The Idea

For a while now it's been bugging me how messy it can be to get a VPN up and running. Or just getting setup through any means to gain remote access to different computers. Yesterday I had an idea somewhat based on Hamachi and BitTorrent.

Imagine a VPN service which instead of connecting to a central server, uses BitTorrent's DHT implementation to find peers which are part of the same “network”, just like BitTorrent clients finds peers who have the same torrent. Once peers have been found, secure and encrypted connections are setup between the local machine and all remote peers, creating a Virtual Private Network across the Internet which in essence works a lot like Apple's Bonjour technology; Everybody talks directly to everybody.

The Details

I believe the concept is rather simple, but implementing it could be another story. From a functionality point of view, these are some of my initial ideas/notes:

• The networks you are connected to are managed by “key files”, containing the following:
  • A random unique hash string which identifies the network when the client searches for peers via DHT.
  • An encryption key (or a set of encryption keys), which are used to encrypt all traffic between peers.
• Create a network: You generate a new key file which is populated with a random hash signature and encryption keys for you automatically. Somewhat like how SSH keys are generated.
• Join a network: You simply copy the key file from one computer to another. The second computer will find the first computer via DHT and start communicating with it automatically.
• PEX (Peer Exchange) can also be used to faster discover all peers in your Network. Meaning once you find one peer via DHT, it will tell you about all peers it knows about.
• Once peers are connected to each other, Apple's Bonjour technology (or something similar) could be used to get around potential IP conflicts and the like.
• In theory, the VPN client could behave just like a BitTorrent client when searching for peers via DHT. Once a peer is found however, it uses it's own protocol to setup a secure VPN connection. This could allow us to take advantage of BitTorrent DHT nodes already online, effectively piggybacking on BitTorrent users.
• Everything should be open source. Cause really, what's the point otherwise? :)

Conclusion

To me, the coolest and most interesting point is piggybacking on existing BitTorrent DHT nodes. If this is something BitTorrent client developers will consider evil, I don't know.

That's why I'm writing this post. I'm hoping to get some feedback, suggestions, insults, and ideas from people who are smarter, and know a lot more about these kind of technologies than I do.

In terms of an actual implementation, personally I would love to build this thing. But my networking and low-level programming skills are not very great at all, it would take some time before I could have anything presentable. But if someone else might be interested in starting it, I would very much like to be part of the effort, in any way I can till I've gained the required skills.

After I updated my profile picture today, a friend of mine responded with:

Build a service that changes your profile picture in all social networks!

He also didn't like the new profile picture, but that's his problem. My first response was fuck off “Gravatar?”. Obviously he didn't mean Gravatar, I just mentioned it to annoy him.

A service which automagically just updates your Facebook, Twitter, Flickr, YouTube, Vimeo, Gravatar, ... profile pictures would be quite cool. Thinking about it a bit more, there are three problems with building such a service:

• A lot of these sites you will need to crawl programmatically using some kind of web-crawler library. It will be a pain to write, and even bigger pain whenever they change anything in the HTML layout of their pages.
• Some sites have specific and/or strange restrictions for image dimensions, file size, and even file format.
• Will people actually trust such a service with passwords for all of their online social networking accounts?

The later problem, trust, is definitely the biggest one. And I'm not sure you could overcome it unless the service is officially sponsored and/or operated by Google or somebody. I do think it could be a fun project to undertake, but I think it's pretty doomed right from the start unfortunately. Although if I updated my profile pictures more than once every 4-5 years, I might just build a prototype for myself at least.