[tor-talk] access sites

Paul Syverson syverson at itd.nrl.navy.mil
Thu Apr 5 12:41:23 UTC 2012

On Thu, Apr 05, 2012 at 05:24:28AM +0300, Maxim Kammerer wrote:
> On Thu, Apr 5, 2012 at 04:15, Ted Smith <tedks at riseup.net> wrote:
> > (Also, have you missed the mail *on this list* from .mil domains?)
> There is Paul Syverson, who works at NRL, if I am not mistaken; did I
> miss anyone else?

We created Tor to protect military communications. Much like other
things invented at NRL (e.g., the joystick controller for remote
control---patented in 1923! or GPS) it also has widespread civilian
use. For most of those, the civilian, business, or other government
use is icing on the cake of the purpose that prompted research on
them.  For some, e.g., IFF (identification friend-or-foe) their
development into civilian use (ATCRBS, the air traffic control radar
beacon system) importantly facilitates military use of the shared
space. For onion routing, we argued publicly right from the start that
the diversity of users was an essential element in effective use of
the technology---even way back when we were just calling our systems
onion routing, rather than _the_ onion routing (Tor) to distinguish
from instances of onion routing developed elsewhere.

As you can see, I'm not averse to touting these creations. But I am a
researcher who does publicly published research in this area and whose
work largely benefits from visibility. As Roger and others have
pointed out earlier in this thread, people who rely on Tor to protect
sensitive communications are rarely going to be happy to have anything
revealed about their usage. You are simply not going to hear from
(most of) those people, and you are definitely not going to get a
representative sample of such use. At best you are going to be lucky
to have anecdotal examples or even just anecdotal claims of usage
from which to extrapolate.

You seem to be asking for a statistically accurate demographic study
of all users. But I am sure there are whole classes of users who don't
want even their class of activity on Tor, much less their specific
activity, known. I have no idea what classes, but that just makes
sense. And on a more individual level, for every stalking victim who
both managed to connect to Andrew and decided to trust him to help her
protect herself online there are ???  others who did not have that
opportunity or were unsure enough about trust to not reveal. (Of
course ideally anyone regardless of technical background should know
about the benefits of Tor and how to use it for their needs without
having to talk to Andrew or someone. Let's not get into any of those

A well-designed user study will tell us something interesting about
Tor users. What it will definitely not do is give us a representative
distribution of Tor users by purpose (and as already touched on usage
demographics, e.g., by country can be significantly
dynamic). And inferring user distributions from traffic distributions in
studies that are methodologically controversial is not helpful.  If
that's all we've got for now, then that's all we've got. But we should
be very careful what we infer from it.


More information about the tor-talk mailing list