[Slightly OT] Theoretical classification of Tor
kitsune.or at gmail.com
Tue Jun 17 01:34:39 UTC 2008
-----BEGIN PGP SIGNED MESSAGE-----
Something I've been thinking about: On a theoretical basis, how would
Tor's architecture be classified?
Two extremes of the scale I'm talking about would be:
* Fully centralized, like plain ol' client/server;
* Fully decentralized, which Gnutella comes close to.
Now, Tor's data-carrying architecture is primarily decentralized -
anyone can run a node, and these nodes are chosen (more or less) at random.
However, the idea of directory authorities clouds this a bit* - and the
fact that there's a directory authority consensus system, clouds it even
If there was a single directory authority, I'd consider the system to be
partially decentralized (or mixed): While the data infrastructure is
fully decentralized, all clients look to a central authority for a
listing, producing a single point of failure (or - for the paranoid -
With multiple directory authorities which must produce a consensus,
though, I don't know how this would go. It's even further from
centralization than the example above, but not fully decentralized.
Does anybody get what I'm saying? Sorry if it's just technobabble.
*: Please don't take this as trying to spread FUD - which some people
have done in the past when mentioning directory authorities. It's not my
AAS, CompTIA A+/Network+/Security+
Owner of Tor node "kitsune"
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
-----END PGP SIGNATURE-----
More information about the tor-talk