[tor-talk] Guard selection impact on anonymity for bridge users

[Lolint]

Hi

I see that guard selection is a very important criteria for ensuring anonymity with Tor. But I've never seen it being discussed in the context of bridge users who usually have to input those bridges manually.

I have a couple of questions:

o The recommended behavior is to take the 3 or 2 bridge lines from BridgeDB and STICK WITH THEM (not rotating them) if they work for as long as possible? If so why aren't there any warning against rotating them for no reason for bridge users?

o If someone has 3 different laptops, they should use the same bridge lines across all of them instead of using different bridge lines for each one, right? (I know that Tor would pick different guards for each one, but since bridge selection is manual, it leaves this option for the user)

o This one is a bit different: Since bridge users are much less than direct users (nearly 1 bridge user for each 40 direct users) and since the density of users per bridge relays is much less than the density of users per guard relays, wouldn't that make traffic correlation harder for direct users than bridge users?

Thx

--Jeff