On 21 Mar 2016, at 04:00, Philipp Winter <phw@nymity.ch> wrote:

I wrote an exitmap module [0] that can tell us how many exit relays see
a CloudFlare CAPTCHA when connecting to a given site.

First, I ran the module for coreos.com because it uses CloudFlare, but
the owner configured it to whitelist Tor. Indeed, only one out of 864
exit relays saw a CAPTCHA:
<https://atlas.torproject.org/#details/7DD29A65C370B86B5BE706EA3B1417745714C8AF>

Next, I ran the module for cloudflare.com, which does not seem to
whitelist Tor. 638 (75%) exit relays saw a CAPTCHA and 211 (25%)
didn't.

This looks great!

Do we know if CloudFlare's blocking depend on the remote website, or the website's CloudFlare settings?

Or does CloudFlare treat each Exit Relay the same regardless of which website it's accessing?

Their introductory marketing / documentation would seem to indicate it's global:

"Once CloudFlare identifies that there is a new attack, CloudFlare starts to block the attack for both the particular website and the entire community." [0]

Can the ExitMap module also record how many sites show CloudFlare's "JavaScript Challenge" [1] ?

http://www.zdziarski.com (yes, only HTTP, ugh) uses their JavaScript challenge.

And their "Totally Block Tor" [1] option? (only available to enterprise (paying?) customers)

I don't know of a CloudFlare website that blocks Tor entirely.