[tor-dev] Reachability testing brainstorming. (OONI, metrics, pluggable-transports)

[Aaron]

BridgeDB was extended to filter bridges that are blocked in a
specified country from its response. BridgeDB needs a list of what is
blocked, and where, for this to work.

We want to answer the question "What is blocked, and where", with the
constraint that the list of bridges should not become public.

Here are a few questions to kick off the discussion. Any insight, more
questions (or answers) are appreciated.

1. Do we want active or passive reachability testing (or both)?
2. For active scanning, OONI comes to mind. Do we trust operators with
the list of bridges? A partial list of bridges?
3. Does an open TCP port mean that a bridge is 'reachable' or should
the test complete a full handshake?
4. What about pluggable transports (PT)? Will the Bridge Authority do
independent reachability testing of PTs?
5. What concerns are there regarding detectability of active scanning?
6. For passive censorship detection, are there other indicators other
than relay/bridge usage-by-country over time?
7. Should relays report per-country usage by address/transport? What
are the upsides? Downsides?

Thanks in advance!
--Aaron