[p2p-hackers] A globally distributed peer-to-peer data archive

[Tony Arcieri]

On Sat, Aug 27, 2011 at 11:39 AM, David Barrett <dbarrett at quinthar.com>wrote:

> Perhaps I misunderstand Tahoe.  Is it intended for groups to share data
> (eg, I add a file, you discover and download it), or is it purely for me to
> backup my data?
>
> If the latter -- if it's purely for individual backup -- then I agree
> there's no risk from the RIAA: you're not sharing it so there's no problem.
>
> But if the former -- if there's some way I can join your Tahoe cluster and
> access data that's already there -- then would you be willing to let the
> RIAA join your cluster with the keys to access the data there?
>

I think Tahoe fits either use case of a shared file repository or a
distributed backup system. There are two levels of credentials you can
share: the introducer furls, which grant access to your grid, and the
rootcap for a particular directory on the grid, which grants access to a
specific set of files. By Tahoe's design you should be able to give the RIAA
the introducer furls to your grid. They can then create directories and
store data on your grid, and will accept encrypted chunks of files from
other peers. However, unless they actually have a rootcap for directories
containing infringing content, they have no way of knowing whether or not
the chunks they receive from other peers contain this data or not, since
it's encrypted.


> Because for something like Tahoe to work on a global scale, eventually
> you're not going to know with certainty all the other members -- and any one
> of those could in fact be the RIAA trolling for people to sue.
>

I'm envisioning a system where there wouldn't be an encrypted introducer URL
to gain access to the grid, instead additional methods to provide a web of
trust and collaborative filtering of particular peers would be used to
prevent abuse. However, I think Tahoe's rootcap system is still essential.
Cryptographic credentials would be required to access any data on the grid.

For example, is it possible to determine who (username or IP) added the
> content in the first place?


To my knowledge Tahoe doesn't log which nodes are responsible for the
creation of content on the grid. Instead there are rootcaps which control
read and write access to directories on the grid, and anyone with the
"writecap" can add content to a particular directory. I may be mistaken as I
have just started using Tahoe. In the system I'm envisioning, the anonymity
of who created a particular file would definitely be preserved.


>  Is there some way to differentiate between when a node is requesting a
> specific piece of content from you, versus when it's just rebalancing some
> tree?
>
> If so, then it's limited to groups of trusting friends -- still a great
> tool, but not a vast treasure of human knowledge as it sounds like you want
> to build.
>
> But if not, if it's truly impossible to know whether a node is downloading
> and storing something because the user actually wants it or because the
> system forced them to get it unaware, then that could be a pretty good
> defense against overzealous copyright enforcement.
>
> Though I wonder if a simpler model is to just start with something like
> BitTorrent, except every time you download anything, you also download
> something else randomly.  So half of the content you download/seed is what
> you actually picked, and the other half is random.  Then there's no way from
> the outside to tell which was for you, versus which is caching random
> content to increase the uptime and performance of the whole system.


I'd certainly agree BitTorrent could be the underlying technology underlying
this system, layering the discovery and crypto on top of a BitTorrent-based
system underneath, although I'd prefer it has more filesystem-like semantics
like Tahoe. Perhaps BitTorrent with an "earliest-first" instead of
"rarest-first" chunk selection mechanism would work.

-- 
Tony Arcieri

_______________________________________________
p2p-hackers mailing list
p2p-hackers at lists.zooko.com
http://lists.zooko.com/mailman/listinfo/p2p-hackers


----- End forwarded message -----
-- 
Eugen* Leitl <a href="http://leitl.org">leitl</a> http://leitl.org
______________________________________________________________
ICBM: 48.07100, 11.36820 http://www.ativel.com http://postbiota.org
8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE