@@ -59,3 +59,42 @@ censorship as an opportunity to distribute fraudulent software packages containi
During this time of increased need for support, our user support specialists have used translation tools and localized templates to communicate with users, but there is a limitation to the value of the support we can provide without Farsi-speaking specialists. In this activity, we will hire Farsi-speaking staff to provide Farsi-language user support for Tor-enabled tools on the Tor Project’s and Guardian Project’s live user support channels that can be accessed through email, Telegram, Whatsapp, Signal, and the official Tor Forum.
Beyond keeping Iranians connected to the open internet, live user support creates a reliable early censorship detection mechanism. When users know to report issues directly to one of our support channels, we’re able to investigate problems quickly. Sometimes this direct connection to users allows us to detect censorship hours or days sooner than automated testing. This in turn helps us develop and deploy fixes sooner, which leads directly into the second Objective.
## Objective 2: The Tor Project can quickly verify how Tor is being blocked in Iran & unblock it
To achieve this Objective, we must be able to detect new instances of filtering and blocking against Tor, or changes to existing methods; investigate how this blocking and filtering is taking place; work with in-country users to test solutions in their real-life circumstances and networks; and quickly deploy user-validated solutions. To do so, we will:
**O2.1, Research and implement a method to collect more accurate counts of Snowflake users:**
Snowflake is the most widely-used pluggable transport by Iranians. About 90% of Iranian users who connect to the Tor network using bridges do so using the Snowflake pluggable transport. Our challenge with detecting filtering or blocking against Snowflake is related to the methodology we use to count users. Currently, the system we use to count Snowflake usage—via self-reported “consensus fetch”—is known to undercount users by as much as a factor of 10. This counting method is also only able to count the number of users who make it successfully to the Tor network via Snowflake. It cannot count how many attempted users fail at any part of the process. These factors make it difficult to detect or understand the severity of the impact of censorship against Snowflake, which disproportionately impacts Iranian users.
In this activity, we will research and implement a method to more effectively detect censorship against Snowflake by more accurately counting usage. We will shift the counting method away from the “consensus fetch” and instead pull data from the Snowflake broker and Snowflake backend, which will allow us to count a combination of the number of requests, the number of Snowflake proxies, the number of attempts the broker makes to match a user to a proxy, and the number of successful sessions. The end result will be more accurate counts of unique users, their
country of origin, and whether or not the connection is working for them.
**O2.2, Set up, run, and monitor more vantage points in-country:**
Vantage points in-country allow us to monitor how Tor tools work in Iran and test fixes or changes to these tools when new filtering techniques are implemented. We will do this with the support of Miaan Group, who will help to locate and establish servers that will effectively reflect and test the Tor user experience in Iran. At this time, we do not have any live servers. In the past, we had 5 servers which now remain inactive. However, funding through this project will help us reestablish more vantage points. When the servers were live, they were used to:
- Check and continuously monitor pluggable transport connectivity and performance
- Analyze censorship mechanisms
- Test and validate countermeasures
The information we could gather from the servers made it possible to quickly test and validate anti-censorship measures. Relying only on user feedback has proved a useful but slow process.
To protect users' identities, users also often need to redact important/personal details to protect their safety. With a vantage point, we're able to quickly gather information remotely in a safer way. Access to such a vantage point makes developing anti-censorship tools faster and more efficient for this specific user group.
**O2.3, Connect to users inside of Iran to test the Tor connection and debug Tor-powered tools:**
Similarly to O2.2, connecting with in-country users gives us an understanding of how Tor tools work for Iranians and test fixes or changes when tools stop working. Direct connection to users is one of the most vital tools we have in our approach to keeping Iranians connected because it allows for a rapid testing, validation, and release cycle. We will only connect with Iranian users who directly reach out to the Tor Project and Miaan via our official support channels and help desks. In order to test Tor connection and debug Tor-powered tools, we will not ask them to run scripts or custom programs. We may ask them to tell us about what they are experiencing in their
connection to Tor, to test different settings within these tools and report the results, to share their logs from the Tor application they are using (which can be anonymized and do not require running custom software to collect), and to run OONI connectivity tests (as available). These activities are chosen for their low risk. Please see our Security Plan for more information about our approach. We will do this with the support of Miaan Group, who has already helped the Tor Project in this way over the past several months.
**O2.4, Monitor impact of censorship and internet filtering events:**
We will monitor and evaluate the impact of new censorship or filtering events. Miaan Group will support the Tor Project by sending us reports from its networks of in-country changes in internet access.
**O2.5, Develop and deploy fixes to Tor tools when censorship disrupts access in Iran:**
The culminating activity in this Objective involves reactive software patching and tweaking that is
made possible by our ability to detect changes in censorship, the information we’ve gathered about how blocking and filtering is taking place, and our work with in-country users to test solutions in real-life circumstances and networks.
This activity does not include developing brand-new tools, implementing new protocols, or implementing new pluggable transports. Instead, the development work included here centers a rapid patching, testing, validation, and release cycle in order to get around blocking and detection as it happens. This work is in coordination with the testing and feedback cycle made possible in O2.3.
