The cheap low risk node majority attack, pki, geoip, etc

Steve Kinney admin at pilobilus.net
Tue Jun 7 11:59:26 PDT 2016


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 06/07/2016 12:48 AM, grarpamp wrote:
> On 6/6/16, Steve Kinney <admin at pilobilus.net> wrote:

[...]

>> The only defense I can think of is to assure that message
>> traffic passes back and forth between mutually hostile national
>> jurisdictions before delivery.
> 
> This is suggested often on tor-talk. And tor devs continually pass
> on it.
> 
>> This would be a bit of a hairball to implement
> 
> Not really. Tor already loads GeoIP. So 20 or so lines of code and
> you've got a separate country for each hop. A few more lines to
> define groups like FVEY / BRICS, hemispheres, regions, AS, etc. 
> Users could isolate on whatever they wanted.
> 
> And a bunch more lines to include attributes as to "verified to be
> a human node operator in person" pki web of trust into the
> consensus. At least that way it raises the cost and risk to
> adversaries who today just use their Govt credit card to order up 
> VPS nodes all over the world.

Making users prove they are human without disclosing any personally
identifying information is not an easy task.  As far as I know, nobody
has found a way to do it.

The closest the IT industries have come so far is the CAPTCHA.
Automated CAPTCHA breaking methods include forwarding the challenge
image, puzzle or etc. to 3rd party websites where users solve CAPTCHAS
to access content, and submitting those users' solutions back the site
your automated gadget wants to fool.

> Does it benefit? Tor devs say trust the random node selection. 
> Others say at least some subset of users know the / their 
> environment better and could use such tools to advantage.
> 
> Tor still refuses to do it. So like mixmaster, you have to do it
> yourself. That sucks. It could stand to be talked over a bit more.

Routing TOR traffic back and forth across mutually hostile borders
would indeed be easy to do.  It only requires, as Dr. Strangelove
would say, The WILL to do so!

However, this would lead to...

>> lots of slippery variables and potential counter-actions by 
>> hostiles that would have to be taken into account.

The State Department funds TOR to defeat national firewalls of
"repressive regimes", enabling communication favorable to the U.S.
National Interest. TOR is a two edged sword that can also be used
against the National Interest, so the NSA thinks TOR stinks.  But
apparently State, CIA etc. will not give TOR up, because of its role
in the development and management of dissident groups in countries
slated for destruction.

A really big bump in TOR traffic crossing the borders of "repressive
regimes" could create traffic jams that degrade TOR's performance
right where the State Department wants TOR to work best, and escalate
the arms race between TOR and the operators of national firewall
filters.  This would shift the cost/benefit ratio for TOR's sponsors
in the wrong direction.  Hence "hairball."

One possible solution would be to build privacy directly into Internet
protocols.  That could drive the costs associated with network
surveillance way up compared to today's rates, while making TOR and
etc. redundant.

https://tools.ietf.org/html/rfc6973

Of course, every State and Corporate stakeholder will unite against
implementing any such scheme, even if practicable solutions are found.
 But someday, somewhere a window of opportunity might open up, and
meanwhile "there's no such thing as wasted basic research."

:o/


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.22 (GNU/Linux)

iQEcBAEBAgAGBQJXVxmOAAoJEECU6c5XzmuqfwEIAJRNYp3byLHPUiU/hXhGbjR2
BgXN0IiXO7PnU1tRBmaeXaBSt6Bb51XlZzSiD1pa76GOMFnF8aGEaZnucPxUUnat
0LagY7w6XtX1WoOQPvaoQxij5EPkrfaU3Wk+OErcF4dW96/w7KV7RFUayQMs1zD9
O8MUstF89RaE1eDU8Iw+EqhvUdcZoDVwkTluq6xwLrNBxz4lRmWAr/5CuFrx8Z5J
Y2IQ21VZctYUO0lbVljh4TfF4mSvS68ddZVlfmZbmpjHZYQWSJFTmnrkgRLRtzoQ
HkWlyrxZQ1hqkUt0tp8iQSISj/RrRtm9U1SuR9sZWQNXs/D6jFlu6u7l+j9nhNM=
=KYGq
-----END PGP SIGNATURE-----



More information about the cypherpunks mailing list