From: nobody@soda.berkeley.edu
Date: Sun, 30 Jan 94 11:48:58 PST
To: cypherpunks@toad.com
Subject: Re: "bomb me" traffic
Message-ID: <199401301944.LAA00407@soda.berkeley.edu>
MIME-Version: 1.0
Content-Type: text/plain


Greg Broiles offered,
"I'm sure that the good intentions are appreciated, but I don't see how
traffic which is predictable, both in terms of size (~ 45K) and timing
(every morning at X:00 AM) is going to make the task of traffic analysis
more difficult."

What I wrote was a bit misleading. I meant to say the 6-12 rounds in the
morning (these days down to ~6) were the MAXIMUM traffic load I expected to
send out at once. Given my sleep cycle, this is NOT predictably X:00 AM.
I have a convenient internet link on a PowerBook on my desk for which I am
logged in most of the day, at least in the background of the Mac Finder. So
throughout the day I will get more requests and send off the five mails.
The sizes vary from 18K to 46K and are being sent chained between 2 remailers
selected at random, including the possibility of those two being the same
remailer. I am also sending out a short (but I could make it longer or even
of random length if I learn some more Unix/perl) "Ping" through all the mailers
on my list a few times a day. I even made a "SuperPing" commands script which
sends the short message through all the possible sets of two chained remailers
to check the links BETWEEN remailers. I do not do this in both directions;
only Me->A->B->Me, not Me->B->A->Me also. Last night the pmantis->hfinney link
was not pinging but pmantis and hfinney themselves pinged fine. I need to play
more, and recheck my "script" before I comment on the reliability of those
remailer-remailer links though.

So throughout the day, as especially when I (and thus others) have free time,
I will be sending quite a few messages with sizes 18-46K through two random
remailers each. It seems to WORK, which is what matters to me. It seems this,
in whatever small amount, WILL make traffic analysis more difficult.

"Further, the characteristics which have led some remailers to be considered
unsuitable for the current project make them stronger, in terms of
resistance to traffic analysis - e.g., erratic (or slow) throughput, and
loss of "Subject:" lines."

I understand this and agree with you completeley. I am considering using those
remailers too, as I have already included a header telling people that the pieces
are EXPECTED to arrive erraticallly over a day's time.

However, I think added delays should eventually involve a few SECONDS, not a few
hours. This isn't the fucking postal service I'm trying to use. I like the speed
of internet e-mail. Stripping subject headers should be left up to the user of
the remailer and is trivial to leave out. Such delays make it impossible to say,
cooperate in a timeley manner on this or other mailing lists, or on Usenet. Such
is anon.penet.fi's delays, and those of the slow remailers. When ONE e-mail takes
upwards of a DAY to arrive, or even a few hours, I just can't carry out my plans
for world domination with my co-conspirators ;-). Adding multi-hour delays to my
e-mail is just too primative. E-mail is replacing the telephone for many uses so
if the remailers ever expect to gain a large base of users, they need to be FUN,
and delays are not. Gaining a large number of daily users is the long term solution
to traffic analysis worries, but adding hours of delays COMPARED with other
fast remailers will make people avoid those remailers.

-Xenon