On Thu, Jan 7, 2016 at 2:41 PM, Andy Isaacson <adi@hexapodia.org> wrote:
On Wed, Jan 06, 2016 at 11:51:29PM -0500, grarpamp wrote:
> Nine server council... a hoseablitly focus point similar to Tor dirauths.
> In any case... interesting.

The privaTegrity (PT) backdoor is significantly more malignant than the
Tor dirauth issue.

If you pwn the Tor dirauths, you can sign and publish a false
"consensns" to clients that will cause them to use only your relays for
new connections, thus breaking anonymity for new connections.  Doing so
leaves a trail of bits showing that this was done (mostly just on the
target system).  Tor is actively seeking solutions to make their system
more privacy-preserving and if a better option shows up in research,
they will likely adopt it.

If you pwn the PT overlords, you can retrospectively deanonymize
connections that you recorded in the past.  If PT were deployed at scale
with a, say, 12-month deanonymization window [1] then every connection
during that interval would be silently deanonymized by APT0 who has
stealthily exfiltrated the overlord private material.

[1] the whole point of the PT backdoor and its claim to "break the
    crypto war stalemate" is that a lawful investigation could go back
    and ask "who sent this bomb threat".

If PT were deployed at scale and a vulnerability were found that used
the backdoor, the developers are left with an uncomfortable choice --
fix the vuln and thereby break the backdoor, or leave users vulnerable
and preserve the so-called "lawful" access?  This is not a conflict that
I want my privacy technologists to have to navigate.

Now, cMix seems like an interesting technology (much like the tech bits
of eCash were interesting back in the 90s, a previous #chaumism[3]).  I
chatted with one of the coauthors yesterday and there's clearly an
interesting performance improvement to existing mix networks; read the
paper[2] for more details.  But the PT system built on it is predicated
on an unrealistic model of datacenter security, international
geopolitics, network economics, cyberwar, and network reliability.

[2] https://eprint.iacr.org/2016/008.pdf
[3] https://twitter.com/hashtag/chaumisms

-andy


To add;
It was surprising (to me) that Chaum should be the one to produce the first of the modern 'solving the key escrow problem' algorithms. Academia has been ignoring this particular problem for quite a while - I expect that more proposed solutions will follow, solutions that will be more difficult to prove insecure...

PT sounds like its rooted in threat models & crypto systems from 10-15 years ago - I'm not sure why anyone would migrate to this system - over anything presently available (unless everything else is made illegal, of course).

Check out cMix's 'beta' paper & make attacks as necessary.
http://www.scribd.com/doc/294737065/cMix-Anonymization-by-High-Performance-Scalable-Mixing

Some of Chaum's earlier work involved designing systems with arbiters where 3rd parties could prove evidence of misdeeds - this system, using threshold secret sharing (right?), doesn't expose the same properties.

Which means that, of course, operators of routers are free to collude and subject to TAO from various IA.

-Travis
--