evil nodes

[Chris Palmer]

alexyz at uol.com.br writes:

> Trustworthiness is dificult to measure but there is less and there is
> more. While I wouldn´t put my life on the line for my ISP, it is bound
> by a contract with me and is accountable by law for certain
> behaviours.

You don't have much, if any, leverage against your ISP. The worst you
can do is take your monthly payment away from them, and they've got many
more customers who are unaware of or don't care about the issues you do.

> That relation enforces some trust that I will never have with any Tor
> node. And obviously not all of them are alike. Based on your
> information I may decide to enforce my communications through specific
> exit nodes that I deem more trustworthy.

Oh yes, using an exit node you feel is more trustworthy is an excellent
idea. That's why the ExitNodes configuration directive exists, and the
.<nodename>.exit feature.

> But that will not be effective if middle and entry nodes can attack
> the same way. Will the encryption between the nodes somehow deter
> this?

Same as SSL, or any other encryption. Your data is encrypted multiple
times while it's inside the Tor network; only the exit node sees the
final plaintext after it peels off the last layer of the encryption.

If an adversary alters the ciphertext of an encrypted cell of Tor
traffic, the cell will fail the integrity check done by each Tor node in
your circuit, and the cell will have to be resent.

When using Tor, threats against data integrity come from the exit node,
from the destination server, and from your own client software. That's a
major win compared to the situation when not using Tor: your data is
under integrity threat from *every* intermediary, instead of just the
last! This, combined with the ability to choose your exit node, puts you
way ahead of the game.

> Threre is risk in everything we do, I just wish to understand those
> risks so I can manage them. SSL is a great tool to deal with this
> situation but it is not always available. What do you think is more
> likely, that an attacker uses statistical analysis on big portions of
> the internet or inject code in the streams to allow him later to have
> a trojan phone home and identify the computer?

The weakest link in the security chain is probably your own behavior,
followed by your client software. As unsecure as the Internet is, the
endpoints are far less secure. I notice you are using Windows...


-- 
http://www.eff.org/about/staff/#chris_palmer

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
URL: <http://lists.torproject.org/pipermail/tor-talk/attachments/20050602/794048f1/attachment.pgp>