[timthelion]

Aren't there few enough urls that are registered that it should be easy to figure out which ones are blocked? It looks like there are only about a billion http://www.internetlivestats.com/total-number-of-websites/

With 1000 connections at 1 dns lookup per second that would be just 12 days. Anyone want to spin up some AWS instances in India and find out?

[zuron7]

Not worth it. The blocking that the government uses blocks only the http version of the sites. Append an s manually and access is restored. But to the uninitiated that block poses enough of a challenge.

[wav-part]

Not true anymore. https are getting blocked too.

Last few lines of `curl --trace - https://thepiratebay.org/` on Jio ISP. Full log https://gist.github.com/anonymous/27ddfa674233d8d17a007f1b3f...

    => Send SSL data, 5 bytes (0x5)
    0000: 15 03 03 00 1a                                  .....
    == Info: TLSv1.2 (OUT), TLS alert, Server hello (2):
    => Send SSL data, 2 bytes (0x2)
    0000: 02 46                                           .F
    == Info: error:1408F10B:SSL routines:SSL3_GET_RECORD:wrong version number
    == Info: Closing connection 0

So TLS handshake is being interrupted. If you think about it, https blocking is actually more efficient.

curl log does not show the last fake rcvd packet. Anyone know how to do that ?

[timthelion]

But if we are speculating on what sort of websites are blocked, that method would reduce the uncertainty. It might even lead to the uncovering of something scandalous, like the discovery of a website being blocked for revealing evil truths about the current government.