IETF 91 Hawai'i SAAG, Thursday November 13, 2014 Thanks to Paul Hoffman for taking minutes. WG summaries sent to the list, a few presented. Related WG summaries presented. Time sync protocol security intro and requirements Karen O'Donoghue Two WGs: NTP and TICTOC RFC 7384 covers many issue IEEE 1588 also covering time RFC 5906 -- Autokey Time server authentication Published with know problems Network Time Security Kristof Teichel Not considering how to use security with NTP pools Phill Hallam-Baker: Identical to the requirements for private DNS Can reuse the key establishment Jim Gettys: If we want to see DNSSEC to the edge, we need secure time to the edge Threat Model Analysis of Router Backdoor Ning Zong draft-song-router-backdoor-00.txt Stefane Bortzmeyer: The draft is important, but doesn't give many ideas for how to detect Ning: Will add more later Richard Barnes: Don't get optimistic about ability of probing for backdoors (noname given): Not finding guilt is not the same as proving innocence Ning: yes. Proposal for research on human rights protocol considerations Joana Varon and Niels ten Oever draft-doria-hrpc-proposal Extend from privacy to freedom of expression and association Justin Richer: Interesting work going forward? Classification of RFC? Human rights considerations? Niels: Not sure because this is research, and thus should be in IRTF. First we should see what's there in the RFC series. John Levine: Reasonable idea. In this community, some rights are more popular than others, need to think about balancing issues Richard: Useful. Risk reading technical documents as rights statements. Instead, think of them as a window to what the impacts on rights are. Niels: What is behind window? Should we interview authors Richard: Research, propose what they say, ask authors. Maybe derive your view of the human rights positions behind RFCs and then ask authors of RFCs if they agree with (have rough consensus on?) those statements Subir Das: What was your motivation? Joana: Wondered what was more than just privacy in the documents Niels: Future-proofing the technologies that are coming out of the SDOs. We should think about more than privacy. Basil Dolmatov: How does this deal with net neutrality? Niels: Yes, we consider that. Alissa Cooper: Choose one or two rights to focus on. IAB draft on filtering considerations draft-iab-filtering-considerations draft-morris-policy-cons. Maybe limit your scope to start. David Somers-Harris: Will you research human responsibilities and rights that authors assume humans don't have? Where does this policy end? Joanna: maybe first focus on a freedom of association and expression. Niels: human rights can be in conflict with each other Survey of Worldwide Censorship Techniques Joseph Hall draft-hall-censorship-tech Walks all the ways that censors have used protocols to block traffic Aggregation - what censors want to block Identification - what point in the network is the blocking happen; also layer Prevention - how the mechanisms are implemented "Degredation is the new blocking" How to design protocols that are resistant to traffic analysis Stephane: Censorship techniques have been studied List or references are quite short You cannot help the good guys without helping the bad guys What is the point of the draft Not clear on the focus of this draft Joseph: Wants to help protocol designers prevent censorship POODLE Richard Barnes About 1/3 who turned off SSL v3 went to "error" Some major sites had v3 before POODLE Trying to get out of the fallback business Looking at getting rid of RC4 Added fallback dance in Firefox Result: RC4 pretty much disappeared They could have used better ciphersuites all along Configuration matters Ryan Sleevi: Site operators weren't reconfiguring servers, this was just inside the negotiation There could be legitimate reasons for RC4 until a few months ago Ben Kaduk: No error bars, so should not be sure David Somers-Harris: Does not mean that SSLv3 has died Some Japanese sites still use it Richard: still have a big legacy client issue Sean: des-diediedie is the original Martin Thomson: Firefox download site will still offer SSL v3 Tradeoff is to allow people to get Firefox securely Open mic Dan Harkins: Doing the human rights study will likely politicize protocols Not want the technology to have political context Niels: Agree. The question is how do we want to frame the questions Research is needed to figure that out Mark Nottingham: This is dangerous. If we favor one view, people will come to espouse the other. We should think more about the stakeholders more W3C HTML5 prioritise things Useful tool to move discussions forward draft-nottingham-stakeholder-rights Justin: We have to stop pretending that technology is a non-political decision (applause) Tom Yu: Protocols being apolitic is a fallicy Mark N: Maybe doing the stakeholder priority list in HTTPbis Mark N: (new topic) Split browsers: computation is done on boxes in the cloud Can be used to MITM TLS Started a survey to look at how many are out there One split browser is very popular in India (26%) Kathleen: Some of this is documented in the TLS attack draft from UTA, was on a recent telechat, nearing publication. Lee Howard: Politicalization is very appropriate topic for a plenary or ISOC lunch Alissa: Stakeholder rights Which stakeholder is benefitting is implicit, and that's good You can suffer a much greater loss if you make these things explicit Of course people come with their own priorities, goals, customers Max Pritikin: Identify, don't necessarily prioritize Sean Leonard: Few new drafts PKIX textural format Certificate fragment identifiers LDAP PKCS9 registrations Janardhan Iyengar: Misalignment between consensus and control We can get consensus but we can control their deployment Russ Housley: Multiple groups picking different cipher suites and will lead to interop issues. DICE chose a ciphersuite with a CCM mode; HTTPbis chose GCM Pick one, please Mark N: Stack ranking may help, if we get it wrong the market will decide and that's ok