[tor-talk] Tor as ecommerce platform

Maxim Kammerer mk at dee.su
Sat Aug 11 07:16:20 UTC 2012


On Sat, Aug 11, 2012 at 4:49 AM, Ted Smith <tedks at riseup.net> wrote:
> Well, you're assuming that an exit node (with some given exit policy)
> sees a fair sampling of all Tor exit traffic. This is not the case. Look
> into how Tor assigns exit nodes.

This is indeed the case — clients choose exit nodes uniformly among
Fast+Stable routers, weighted by consensus bandwidth, excluding known
bad exits and specifically excluded exit nodes (e.g., by country —
only done by a minority of users). An exit node with some minimal
bandwidth and a decent uptime will see a fair sampling of all exit
traffic; comparison of per-port statistics will require an easy
normalization.


On Sat, Aug 11, 2012 at 5:11 AM, Ted Smith <tedks at riseup.net> wrote:
> The obvious problem with this (((this, right here, is the productive
> contribution to discussion this email has: it points out the problem
> with your proposed methodologies))) is that it presumes that these top
> 50 .onion domains comprise the majority of .onion traffic through your
> node. I suspect this is not the case.

It is not a productive contribution, since it's an obvious assumption.
I even doubt that there are 50 HTTP .onion sites that see over 1000
daily visits.

> Further, this depends on a long list of assumptions (that all .onion
> sites are serving HTTP over port 80, that the HTTP sites use an
> authentication system you can detect as such, etc.).

You are just inventing corner cases. Port number and protocol can be
determined automatically, and anything that can't and is still in
top-50 will be an interesting find nevertheless.

> You don't seem to be using any actual, quantified method to arrive at
> these conclusions. As such, you're probably using your own biased human
> intuition as to all of the many (very, very debatable) facts above. The
> anecdotes you've shared are a fraction of the anecdotes you've actually
> seen, filtered through the very many cognitive biases we suffer from as
> humans.

Sure, I rely on my intuition. It is based on interacting with a
diverse userbase of Tor users. Yours is based on disliking my
conclusions and nothing else — you never brought up actual references.

> The hypothesis you assign the highest probability towards causing these
> anecdotes is "Tor is mostly used for criminal activity."

No, the hypothesis is that Tor hidden services are mostly used for
criminal activities, and that Tor exit nodes are often used for such,
when Tor is used for the purpose it was designed for, and not as a
glorified proxy.

> There are a wide variety of other hypotheses that seem to me to be equally likely to
> produce these anecdotes, such as "the link network of .onion sites that
> you are privy to contain mostly criminal sites," "criminals are more
> public about their intentions than other users of Tor," and "you have
> seen more criminals post about their activities because of your
> person-specific interests."

Could be, but you have nothing of substance behind this suggestion. It
is criticism for the purpose of criticism — good for high-school
debates and for politicians, but of little value otherwise.

-- 
Maxim Kammerer
Liberté Linux: http://dee.su/liberte


More information about the tor-talk mailing list