wiki:org/meetings/2019Stockholm/Notes/AntiCensorshipRapidResponse
  • censorship can mean many things
    • blocking tor or its bridges
    • also websites, messaging apps
    • anything, really
  • complex: how do you measure? how do you respond to it?
  • ooni has "tor weather" analysis that looks at bootstrap time in particular countries for vanilla tor
  • we rely on user input to figure out that something is blocked
    • user in china informed us about snowflake blocking
    • ooni's slack channel has a fairly technical person from china
    • we almost always have the same questions for users and a tool should be able to answer them -> ooni probe would be helpful here
  • often not completely sure what we do in response to a censorship event
    • we have several tickets where we have some info and then we no longer know what to do (metrics/censorship category)
  • standard set of guidelines for people would be useful
  • lots of open helpdesk tickets that are unanswered
    • nobody taking care of helpdesk at the moment
  • would be good to move forward with "censorship table" in the short run
  • we could use ripe atlas's traceroute test to scan bridges
    • does not cover every country extensively but it's a good start
  • eventually, it would be great to dynamically create a list of bridges to test
    • in the beginning, a static list would be fine
    • we may also want some config knobs that we can adjust
    • make it easy for ooni to get a bridge from bridgedb + we can do it manually in the beginning + eventually, it should be automated + how would we accomplish this?
  • ooni's probe orchestration is opt-in
    • backend can tell probes when it's time to censor
  • currently, probes learn test lists from backend at runtime
  • from measurement to publishing, <24h are necessary
    • processing takes time
    • ooni did some work on the "fast path"
    • idea is to be able to respond in, ideally, 5-10 minutes
    • measurement axes are time, geo coverage, website coverage + pick two out of three
    • should help a censorship alert system + user subscribes to country and get alert when something happens
  • where does it stop being rapid response and turns into "working on the problem"?
    • matter of detection and confirmation
    • "what should our role be?" + ooni works on both advocacy and technical work
  • rapid response should also involve circumvention
    • if you don't help with that quickly, it won't be very useful later on
    • timeliness also important for outreach and advocacy + nobody cares about old news + if you respond *too* quickly your work may be inaccurate + question is how we can support each other effectively to strike the right

balance

+ however, even if you're late, there's still value. you feed into ongoing

activities

+ big liability risk if you get your results wrong

  • ooni has "events calendar" that helps them be prepared
    • it's outdated at this point, though
    • censorship often correlated with elections
    • "censorship measurement campaign" -> recruit ooni users before the elections
    • censorship also correlate with protests
  • ooni slack channel is used for notifications
  • groups have their own whatsapp groups from where we can learn about censorship
  • we want to be *faster* and more *confident* in our results
    • we don't want to wait 24h
    • ooni also publishes results so people can write their own reports based on ooni's data
  • once tor testing is part of probes, we can compare it to reachability of other websites
    • "can i reach tor but not wikipedia"?
  • have a torproject user at pirate bay with a bunch of tor browser seeds
    • concern is that highly-rated seeds are outdated tor browsers
    • we need some kind of trusted uploader
  • only a few countries block tor
    • some block and then unblock
    • need relations with technologies on the ground
  • are things getting blocked and we just don't hear about it?
    • we may do ok on a country level but not at an isp level
    • keepiton mailing list useful
  • censorship shouldn't only be available to highly technical people
  • bridgedb distribution should take into account where user is coming from
    • we don't have information that says what bridge is necessary in what country
  • are censors reacting to ooni probes?
    • some countries are blocking ooni.io (egypt)
    • in many countries ooni.torproject.org is blocked because of torproject.org blocking
  • you cannot avoid burning bridges but you can reason about the rate they get burnt. it's a question of experimental design.
  • if we were the censor we wouldn't surveil ooni probe traffic but rather run the client ourselves
Last modified 6 weeks ago Last modified on Jul 14, 2019, 2:18:04 PM