[tor-relays] Unwarranted discrimination of relays with dynamic IP

[grarpamp]

On Fri, Dec 23, 2016 at 12:06 PM, Rana <ranaventures at gmail.com> wrote:
> If the small relays are largely unused (eg if 10% of the relays carry 90% of the Tor traffic - does anyone have an exact statistics on this?) and if, in addition,  there is no increased anonymity benefit in having a lot of small relays, then why bother?

There's the obvious that if you have a lot of unused relays,
and essentially only your traffic happens to traverse three of them,
then you're much more easily subject to active observation
by the relays themselves, and passive observation by GPA's.

Therein one might expect dirauths to restrict node count
to network saturation levels only, instead of the 7000 we have
today.

But even if using three fully saturated relays, you can still be
deanoned by as little as one guard and one exit / destination.
That works essentially the same for hidden services too.

There are often threads on tor-talk about filling anonymous
overlay networks with dynamic fill traffic / traffic buckets /
packet switching, udp, etc to prevent ease of that happening,
in particular since client nodes would be participating too,
but it goes nowhere.