| Christian Danner 2006-08-22, 7:14 am |
| "Non scrivetemi" <nonscrivetemi@pboxmix.winstonsmith.info> - Mon, 21
Aug 2006 20:03:24 +0200 (CEST):
>And why are you attempting to twist "high latency" into
>"extraordinary" in your attempt to make your point?
A latency of 5-14 hours isn't extraordinary?
>The system is designed to chose remailers randomly so the latency of
>any individual remailer shouldn't matter as long as there's SOME
>latency.
>
>Please don't respond with some sort of silly "user can chose this and
>that" nonsense, or I'll be forced to point out that's a USER issue not
>a remailer or software issue.
Concerning latency the user can't select anything but what's offered
by the remailer admins. So if there are more remailers of that kind,
what would you recommend a user who can't wait a whole day for the
delivery of his message? Either he selects the chain manually, or,
with a random selection, he has to send more copies of his mail to
make certain, that at least one doesn't walk into one of the
high-latency traps. OTOH with only low latency remailers present he's
able to tune the overall latency by varying the length of the chain
and has an additional benefit as it's harder for an adversary to
compromise that greater amount of computers.
>
>You'd have to find another way. <shrug>
>
>
>You just contradicted yourself. If "overall latency" were tantamount a
>single remailer with your "extraordinary" latency would be more secure
>than a chain of 10 with almost none. Obviously that's not the case.
>
>Actually, latency isn't the crux of the remailer network's security to
>begin with it's pooling and reordering. Latency is a side effect of
>those things, but it doesn't have to be a goal.
ACK. That's what I tried to explain. It's about the optimum
combination of chain length and message pool size. The latency of the
single hop is only a dependent parameter without any inherent
(positive) value, which however can be a problem when the speed of
delivery is relevant. However, if sender and recipient are already
under suspicion and there's not enough dummy traffic coming from the
sender, a certain amount of over-all latency is necessary to prevent
from a direct time-based correlation of outgoing and incoming mails.
> With proper reordering
>and padding even effectively real time channels can be just as secure.
Nearly, especially if there's enough traffic.
>
>You assume way too much. An analysis of the quantity or "size" of the
>traffic emanating from a location combined with careful observation of
>the latency between entry and exit traffic could certainly give clues
>to identity. Always using exceptionally long chains might partition you
>just like always using the same chain, or exit. It's a pattern, and
>patterns can be discovered.
Interesting. What method would you prefer to figure out the length of
my remailer chain?
Regards
Christian
--
OmniMix .. protect your privacy
http://www.danner-net.de/om.htm
|