[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: [tor-talk] ICANN and .onion

Hash: SHA512

On 05/19/2014 06:24 AM, Anders Andersson wrote:
> On Mon, May 19, 2014 at 7:06 AM, grarpamp <grarpamp@xxxxxxxxx> wrote:
>> Users leaking dns / failing to redirect dns into tor is not a tor problem.
*** That is a common technologist / U.S. liberal / libertarian issue to
give responsibility to "the user" for "their" "failures" to do anything.
The poor-is-lazy argument.

But for non-hackers, the reality is that apart from booting Tails and
enjoying a proper Tor setup, installing the Tor package on most distros
does not come with pre-installed DNS and *will* leak queries by default.

So technically, you may argue that it's not Tor's responsibility, and
leave it there. But if you're honest you must consider the objectives of
the software, and distros, and end-users as being part of the "Tor system".

> I think that's a rather arrogant point of view.
*** And irresponsible as well.

> If someone would register .onion I see two problems:
> 1) A malevolent registrar could redirect all .onion lookups to their own
> proxy, essentially routing all "hidden" traffic through their own machine.
*** That effectively suppresses the anonymization of Tor in that case,
against a global adversary that is Not So American and have capacity for
eyes around the globe to pin your digital footprint to your ICBM
location in real time.

> I think that's a usability issue, and not something that
> should simply be ignored. Maybe it's not something that can easily be
> solved, but that is why there must be a discussion about it. Maybe the only
> solution is to strongly warn users.
*** One complexity factor here is that not all systems resolve names the
same way, so you need to figure out from Tor whether the name was
resolved securely or not, which is not necessarily doable.

> 2) Useful websites could actually pop up under .onion, making a plugin that
> takes over that domain seem intrusive and less attractive. This is less of
> a problem I think.
*** If useful websites can pop up under .onion, fake copycats can also
pop up that will mimick the original target without the user being able
to notice due to the natural latency of the Tor network. Then you can't
trust Tor anymore to do its job: for an end-user .onion means the site
was obtained via the Tor network. If it was not, because a DNS leak
brought you to the site via the clear Web, you're done.

The only meaningful failure for the leak of .onion to the DNS is loud
failure, aka NXDOMAIN, which is why it's technically important that IANA
forbids registration of .onion in the first place. As far as the DNS
supporters at IETF are concerned, it should be Tor's responsibility to
"use DNS properly" and avoid "integrating top-level domains into the
browser's location bar for convenience" (both quotes condensed paraphrases).

I guess one of the tasks for the next P2P-Names draft is to properly
decouple the DNS issue from the browser-location issue so that .onion is
not anymore a DNS-abuse issue, nor a convenience issue, but a strong
usability issue.

Aside, a second task is to maintain the cohesion of Tor and non-Tor
systems as a single technical non-DNS-based Peer-to-Peer name resolution
issue to avoid "special treatment" of minority networks, and ensure a
future for a techno-diversity of name resolution systems. Societies of
control like to think their (human) identity systems as definitive, and
history tend to consistently prove them wrong. The last thing we want is
to force all name resolution into a top-down, non-autonomous technique,
administratively-controlled centralized system that prevents better
techniques from evolving.


Version: GnuPG v2.0.22 (GNU/Linux)

tor-talk mailing list - tor-talk@xxxxxxxxxxxxxxxxxxxx
To unsubscribe or change other settings go to