teor
2017-12-21 23:02:30 UTC
Hi all,
If bridges users configure at least two bridges in different /16s,
it makes it harder to run some kinds of statistical attacks on the
paths that they build through the Tor network. (For example,
working out that relays in that /16 are never used by the client.)
For context, see this Twitter thread:
https://twitter.com/twbtwb/status/943974051000721408
I think that:
* most of the current default Tor Browser bridges are in at least two
/16s, except for the single-bridge pluggable transports,
* almost all the 3-bridge sets given out by BridgeDB are in at least
two /16s.
Should we:
* document this as a recommendation? Where?
* add code to Tor that warns when it isn't true? (Maybe not?)
* make sure all of the Tor Browser bridges are across two or
three /16s?
* make sure BridgeDB gives out bridge sets in two or three
/16s?
T
If bridges users configure at least two bridges in different /16s,
it makes it harder to run some kinds of statistical attacks on the
paths that they build through the Tor network. (For example,
working out that relays in that /16 are never used by the client.)
For context, see this Twitter thread:
https://twitter.com/twbtwb/status/943974051000721408
I think that:
* most of the current default Tor Browser bridges are in at least two
/16s, except for the single-bridge pluggable transports,
* almost all the 3-bridge sets given out by BridgeDB are in at least
two /16s.
Should we:
* document this as a recommendation? Where?
* add code to Tor that warns when it isn't true? (Maybe not?)
* make sure all of the Tor Browser bridges are across two or
three /16s?
* make sure BridgeDB gives out bridge sets in two or three
/16s?
T