[tor-talk] CloudFlare

[Gregory Maxwell]

On Thu, Apr 18, 2013 at 1:01 PM, Matthew Finkel
<matthew.finkel at gmail.com> wrote:
> Wikimedia is actually willing to discuss an alternative setup if a
> usable one is found. Their current implementation is not really
> acceptable, but there also isn't really a working/implemented alternative
> solution, at this point (and it's not exactly at the top of their list
> to implement their own).

It's the same old story:  There are persistent highly annoying trouble
makers— not even many of them— who are effectively deterred by
blocking whatever proxies they use. Eventually they hit tor, and thus
tor must be blocked from editing.  This abuse isn't imaginary.

The various magical nymtoken ideas would probably be acceptable— they
just need to make it so that an unbounded supply of identities is not
any cheaper than it already is— but they need to be implemented and
not have a high deployment or operating cost.

There are some people who hold the position that instant doubling of
identities (w/ and w/o tor) that attackers would get is not acceptable
but with things like
http://en.wikipedia.org/wiki/Wikipedia:Wikipedia_Signpost/2013-04-08/News_and_notes
 and Tor's effectiveness at evading censorship I expect that most can
be convinced that it's worth it.  Harder would be the fact that
English Wikipedia (and many other larger Wikipedias) blocks most data
centers and VPS services with large rangeblocks as they get used as
account multipliers by socks and an obvious nym implementation would
partially defeat that.