[tor-bugs] #8240 [Tor]: Raise our guard rotation period, if appropriate

[Tor Bug Tracker & Wiki]

#8240: Raise our guard rotation period, if appropriate
-------------------------+-------------------------------------------------
     Reporter:  arma     |      Owner:
         Type:  defect   |     Status:  assigned
     Priority:  major    |  Milestone:  Tor: 0.2.7.x-final
    Component:  Tor      |    Version:  Tor: 0.2.7
   Resolution:           |   Keywords:  tor-client, needs-proposal,
Actual Points:           |  023-backport, unfrozen, 027-triaged-1-in,
       Points:  medium   |  SponsorU
                         |  Parent ID:
-------------------------+-------------------------------------------------

Comment (by arma):

 Replying to [comment:50 asn]:
 > OTOH, a 9 month rotation period gives us about 10 years before we have
 50% prob of being unsafe, and 4.5 years before we have 25% probability of
 being unsafe. This seems like a pretty big safety guarantee compared to
 the rest of the security of Tor.

 My research director hat hasn't been fitting on my head lately, so I bet
 this is actually answered in one of our tech reports and I just don't
 remember, but: it seems likely to me that natural network churn will have
 a big impact once we're looking at such large timescales. That is, the
 safety does not scale up linearly with the rotation period. I think some
 of the COGS paper and graphs were aiming to measure how it diverges.

 It would be great for all of us to have better intuition on what levels of
 safety are possible on the real network (with its natural churn). My
 current guess is that moving to a 9 month rotation period doesn't actually
 give us much more than a 5 month rotation period (at least not reliably)
 -- but I'd like to be arguing based on facts rather than guesses.

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/8240#comment:51>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online