[tor-talk] Creating distributed social networks with WebId behind Tor

[Henry Story]

On 2 Jun 2012, at 16:06, Jacob Appelbaum wrote:

> On 05/16/2012 07:18 AM, Henry Story wrote:
>> I think those are things one can push back to later research, as it is impossible
>> to consider all interesting ideas simultaneously. My interest is at present to
>> see if I can use WebID behind Tor with onion urls. 
>> 
> 
> I'd happily try this out with you - feel free to contact me off list - I
> think it would work well for the FreedomBox community as well.

Will do. 

Currently I am working on getting the main read-write-web application working in Play2.0,
which is a web application framework written in Scala and very good at asynchronous IO...
(scala and akka.io would probably be a very good framework to implement Tor like functionality)
I think this will take me a few more weeks - probably the whole month - to get things 
working smoothly again.

I'd need to play with Tor a little bit so that I can work out what the core issues are
to get a WebID-type protocol working. I would need to dig into the following:

 A. TLS over Tor
   - the client needs to be able to use Tor URLs encryption so that it  can know that it has 
   connected to the right server, without using Certificate Authorities of course. [1]  
      I suppose I should be able to tweak the TLS libs on the java side to be able to 
    make this trust decision, without too much difficulty.... 
   - once that is done having the server request the client certificate should be great.

 B. Read-Write-Web

    I'd need to look into what writing a bit of RDF using Tor URLs would look like. Perhaps you can
   help here: could you (or anyone else here) develop a simple foaf profile using Tor URLs, and put that 
   on a server behind Tor, which I could use some Tor enhanced curl to GET?
 
     If Tor URLs allow for relative URLs (which my httpk:// sketching in mentioned previously [1] 
  would have done), then one should be able to express the graph like this in Turtle [2]

[[
   @prefix foaf: <http://xmlns.com/foaf/0.1/> .

   <#me> a foaf:Person;
      foaf:knows <http://342sdfsds.onion/people/33#clarkKent> .
]]

   Given the location the GET was made on, I should be able to turn those relative urls into
   absolute Tor ones, that I could then add to my WebID profile [3] as a link back. 
   That would of course not be secured, but it would show that the underlying data layer was functional 
   inside of Tor.

   Just with that I could already enhance my crawlers to crawl Tor based RDF.
   
	Henry
   


[1] I had thought of something similar over a year ago with the httpk scheme.
   http://lists.w3.org/Archives/Public/public-xg-webid/2011Mar/0068.html
[2] http://www.w3.org/TR/turtle/
[3] http://bblfish.net/people/henry/card#me
> 
> All the best,
> Jacob

Social Web Architect
http://bblfish.net/