Recent drop in Tor users from Turkmenistan; testers wanted

The number of measured Tor users in Turkmenistan decreased by about 75% between 2021-07-22 and 2021-07-31. This is only users connecting using the non-obfuscated Tor TLS protocol. Users of bridges or pluggable transports do not show a change at the same time.

We discussed the situation in Turkmenistan at the 2021-08-12 Tor anti-censorship team meeting. @gus, the lead of the Tor community team, found a volunteer tester in Turkmenistan, and together they made some observations:

  • Using Tor Browser for Android, the tester could not connect using the default bridges of obfs4, meek-azure, or snowflake.
  • However, a private obfs4 bridge with a non-public address worked.
  • http://emma.mhgb.net/, a web-based tool for evaluating Tor blocking, could not be accessed by the tester, for some reason. @gus set up another instance at http://emma.gus.computer/.
  • The country-wide number of users of default bridges (the hardcoded bridge addresses you get if you simply select “obfs4” in Tor Browser) is close to zero; this means that users who use bridges are mainly using non-default bridges.

@gus is looking for additional testers who can help uncover what is happening in Turkmenistan. You can post here, or on the Tor ticket, or send email to gus@torproject.org (PGP). A good first test to run is:

userstats-relay-country-tm-2021-04-01-2021-08-12-off

userstats-bridge-country-tm-2021-04-01-2021-08-12

userstats-bridge-combined-tm-2021-04-01-2021-08-12

If you zoom out on the graph, you can see that the decrease in the number of relay users follows a period of elevated usage that started around the beginning of 2021.

userstats-relay-country-tm-2020-01-01-2021-08-12-off