Re: NSA Insecure Remailers

[tcmay@netcom.com (Timothy C. May)]

Benjamin McLemore writes:

>  All I was trying to point out with my post is what I saw as a flaw in
> my previous understanding of the depth of the security provided by
> anonymous remailers. There is a difference between relying on the
> mathematics of strong crypto to protect you from government spooks and
> prying bureacrats-- and relying on one's belief that- although one's
> security has been compromised, it will be too expensive or otherwise
> difficult for the government to use this information against you.

Cypherpunks remailers are far from "ideal digital mixes," as described
in David Chaum's February 1981 "Communications of the ACM" paper. This
is well known, and the issues of traffic analysis Benjamin raises are
also valid and known issues. A while back we had many debates about
what to do about message size padding (e.g., quantizing all outgoing
packet sizes to a standard size, or perhaps to one of several (small,
medium, large. etc.) packet sizes. And we debated adding latency, so
that a message waits until N total messages have been received before
remailing. And so on.

In any case, Chaum's ideal digital mix is hard to implement now for
several reasons, largely economic. Ideal mixes also need physical
security against tampering, against interception of internal
operations (perhaps via RF monitoring), etc. Perhaps most critical,
and least studied to date, remailers are only as good as the human
policies at the sites are. (My conception of ideal remailers involves
remailer hardware, perhaps on boards containing enough RAM and/or disk
drive space to hold the batch of messages, that is "untouched" by
human hands. Tamper-resistant modules, sealed hardware, etc. Lots of
issues here. I think "Mom and Pop remailers" could be sold on boards
similar to SoundBlaster boards.)

Chaum's original hardware-based mix also has some weaknesses, as noted
in a EuroCrypt '89 paper by Pfitzmann and others. The software-based
"DC-Net," which comes up so often on this list, is generally better.
Several Cypherpunks are interested in implementing DC-nets. So far, no
progress to report.

> It seems to me that anonymous remailers, despite my initial
> assumptions that they were cryptographically strong, are probably
> compromised by the ability of the NSA to monitor Internet backbone
> traffic, a hypothesis I would love to see disproved. Additionally, my

No, they are very far from being even cryptographically strong
(although parts of the process, involving sending encrypted messages
to the next node, are of course as secure as, say, PGP is).

> I want unbreakable security, untraceable communication and unforgeable
> digital cash--ALL of it mathematically guaranteed and none of it
> compromisable by some underpaid bureacrat who might decide to make a
> little money off of ME in his spare time.

Well, wanting something is not the same thing as getting it. Read the
1981 paper, the 1988 DC-Net paper (available at the soda.berkeley.edu
ftp site), and follow the Cypherpunks activities on DC-Nets. Then look
at the various Cypherpunks remailers...some _require_ encryption (most
don't, and most of us don't even use encryption, which means anyone
reading the packets can see what's going on! A fatal flaw or just
laziness?), some add hours of latency (though not N latency, as too
few messages are flowing), and so on. The market will push
development in possibly more secure directions.

Right now, you can see that Cyperpunks remailers, and also Julf's
penet site, have significant flaws. You get what you pay for (this is
a serious point: the lack of real commerce, the volunteer nature of
all of this, and the generally "hobby-like" nature of these systems
explains why these weaknesses are not getting fixed.

These are largely "toy" systems to provide some experience. They'll
get better with time.


--Tim May


-- 
..........................................................................
Timothy C. May         | Crypto Anarchy: encryption, digital money,  
tcmay@netcom.com       | anonymous networks, digital pseudonyms, zero
408-688-5409           | knowledge, reputations, information markets, 
W.A.S.T.E.: Aptos, CA  | black markets, collapse of governments.
Higher Power: 2^756839 | Public Key: PGP and MailSafe available.
Note: I put time and money into writing this posting. I hope you enjoy it.