[{"author": "Britta Hale", "text": "<p><span aria-label=\"wave\" class=\"emoji emoji-1f44b\" role=\"img\" title=\"wave\">:wave:</span></p>", "time": "2024-11-07T15:34:35Z"}, {"author": "Nicola Tuveri", "text": "<p><span aria-label=\"wave\" class=\"emoji emoji-1f44b\" role=\"img\" title=\"wave\">:wave:</span></p>", "time": "2024-11-07T15:35:13Z"}, {"author": "Pascal Sch\u00e4fer", "text": "<p><span aria-label=\"wave\" class=\"emoji emoji-1f44b\" role=\"img\" title=\"wave\">:wave:</span></p>", "time": "2024-11-07T15:35:52Z"}, {"author": "Deirdre Connolly", "text": "<p><span aria-label=\"wave\" class=\"emoji emoji-1f44b\" role=\"img\" title=\"wave\">:wave:</span></p>", "time": "2024-11-07T15:37:30Z"}, {"author": "Conner Ybarra", "text": "<p><span aria-label=\"wave\" class=\"emoji emoji-1f44b\" role=\"img\" title=\"wave\">:wave:</span></p>", "time": "2024-11-07T15:38:51Z"}, {"author": "Mike Ounsworth", "text": "<p>Here is \"the really really long email\" in question:<br>\n<a href=\"https://mailarchive.ietf.org/arch/msg/pqc/4DYslEv7ZjHnkt4qN_dwgbdnM2o/\">https://mailarchive.ietf.org/arch/msg/pqc/4DYslEv7ZjHnkt4qN_dwgbdnM2o/</a><br>\n(as well as the followup emails below it)</p>", "time": "2024-11-07T15:41:12Z"}, {"author": "Lorenzo Miniero", "text": "<p>@Chairs, are you getting the slides in the confidence monitor that presenters see?</p>", "time": "2024-11-07T15:44:23Z"}, {"author": "Andrew", "text": "<p>In hybrid signature spectrums, is there an example of something that would be considered SNS? It seems like the \"fused hybrid\" example in 3.2 might satisfy SNS, but I think the authors can correct me.</p>", "time": "2024-11-07T15:46:31Z"}, {"author": "Sofia Celi", "text": "<p>@lorenzo, I think now is working in the monitor</p>", "time": "2024-11-07T15:47:03Z"}, {"author": "Tara Tarakiyee", "text": "<p>Please put yourself in the queue, it helps the notetakers</p>", "time": "2024-11-07T15:49:20Z"}, {"author": "Sofia Celi", "text": "<p>yes, thank you, Tara. The person with the question is Mike Ounsworth</p>", "time": "2024-11-07T15:49:57Z"}, {"author": "Tara Tarakiyee", "text": "<p>Mike put themselves in the queue :), what about the person before?</p>", "time": "2024-11-07T15:50:57Z"}, {"author": "Ben S", "text": "<p>Agree with Mike. We need to be careful with this draft to avoid an <a href=\"https://xkcd.com/2501/\">XKCD 2501</a> problem. If you make the document for non-cryptographers full of cryptography, then they just won't read it.</p>", "time": "2024-11-07T15:51:19Z"}, {"author": "Sofia Celi", "text": "<p>ah, ok. Perhaps Deirdre Connolly</p>", "time": "2024-11-07T15:51:22Z"}, {"author": "Tara Tarakiyee", "text": "<p>Thanks!</p>", "time": "2024-11-07T15:51:34Z"}, {"author": "Ben S", "text": "<p>+1 to Kathleen on getting reviews from the intended audience</p>", "time": "2024-11-07T15:51:53Z"}, {"author": "Kathleen Moriarty", "text": "<p>Paul is correct and there could be a blog to compliment to reach other audiences</p>", "time": "2024-11-07T15:53:25Z"}, {"author": "Britta Hale", "text": "<p>@Andrew: yes, fused signatures are an example</p>", "time": "2024-11-07T15:56:44Z"}, {"author": "Deirdre Connolly", "text": "<p>Break is loaded . \"weakened\" is better, depending</p>", "time": "2024-11-07T15:56:51Z"}, {"author": "Jonathan Hammell", "text": "<p>bike shedding is now garden shedding</p>", "time": "2024-11-07T15:57:19Z"}, {"author": "Alicja Kario", "text": "<p>\"ruined\"? <span aria-label=\"big smile\" class=\"emoji emoji-1f604\" role=\"img\" title=\"big smile\">:big_smile:</span></p>", "time": "2024-11-07T15:57:49Z"}, {"author": "Sofia Celi", "text": "<p>Break is accurate imho. When an efficient algorithm comes that compromises the security of the algorithm in polynomial time (in efficient time), the algorithm is broken</p>", "time": "2024-11-07T15:58:37Z"}, {"author": "Britta Hale", "text": "<p>\"obliterated\"? It offers a softer connotation. /s</p>", "time": "2024-11-07T15:59:03Z"}, {"author": "Sofia Celi", "text": "<p>Specifically, for a cryptographic scheme, \"broken\" usually means that an adversary has discovered a way to defeat the intended security guarantees\u2014like confidentiality, integrity, or authenticity\u2014within a practical time frame (often in polynomial time relative to input size).</p>", "time": "2024-11-07T15:59:13Z"}, {"author": "Orie Steele", "text": "<p>(as an individual) Regarding my comment on BCP14 language, if there is no need to use that language, then you can consider removing the instruction on interpreting it, in section 2.... I'm seeing lots of \"should\" and \"recommended\" and very few (0?) \"SHOULD\" or \"RECOMMENDED\"</p>", "time": "2024-11-07T15:59:25Z"}, {"author": "Andrew", "text": "<p>Thanks @britta.</p>", "time": "2024-11-07T15:59:33Z"}, {"author": "Deirdre Connolly", "text": "<p>Sofia is correct; although depending on deployment settings, a non-poly attack on a primitive may be sufficiently risky to be equivalent to a 'break' on the deployment  timescales</p>", "time": "2024-11-07T16:00:09Z"}, {"author": "Deirdre Connolly", "text": "<p><span class=\"user-mention silent\" data-user-id=\"1185\">Orie Steele</span> <a href=\"#narrow/stream/360-pquip/topic/ietf-121/near/143998\">said</a>:</p>\n<blockquote>\n<p>(as an individual) Regarding my comment on BCP14 language, if there is no need to use that language, then you can consider removing the instruction on interpreting it, in section 2.... I'm seeing lots of \"should\" and \"recommended\" and very few (0?) \"SHOULD\" or \"RECOMMENDED\"</p>\n</blockquote>\n<p>I think PQUIP cannot/does not do BCP14 lnaguage</p>", "time": "2024-11-07T16:00:33Z"}, {"author": "Sofia Celi", "text": "<p>so, any cryptography can be broken given enough time and resources. But it is assumed \"practically broken\" if an algorithm is found to do it in efficient time.</p>", "time": "2024-11-07T16:01:07Z"}, {"author": "Tirumaleswar Reddy.K", "text": "<p>yes, the draft does not use BCP14 language.</p>", "time": "2024-11-07T16:01:21Z"}, {"author": "Deirdre Connolly", "text": "<p><span class=\"user-mention silent\" data-user-id=\"603\">Deirdre Connolly</span> <a href=\"#narrow/stream/360-pquip/topic/ietf-121/near/144001\">said</a>:</p>\n<blockquote>\n<p>Sofia is correct; although depending on deployment settings, a non-poly attack on a primitive may be sufficiently risky to be equivalent to a 'break' on the deployment  timescales</p>\n</blockquote>\n<p>example: collision attacks on SHA-1; we've had several before a full collision attack, and it's still ~safe for second preimage attacks; when was it broken? 'depends'</p>", "time": "2024-11-07T16:03:37Z"}, {"author": "Deirdre Connolly", "text": "<p>In terms of _broad_ recommendations, I don't mind being conservative and falling back to Sofia's pure definition</p>", "time": "2024-11-07T16:04:31Z"}, {"author": "Deirdre Connolly", "text": "<p>(This of course is modulo common implementation errors, attacks on widely used libraries, etc)</p>", "time": "2024-11-07T16:04:52Z"}, {"author": "Peter Campbell", "text": "<p>xi</p>", "time": "2024-11-07T16:11:36Z"}, {"author": "Sofia Celi", "text": "<p>so, that is correct, but still fits the definition: I think SHA-1 was widely stated as broken once the 2005 attack was published, which showed the efficient attack. In terms of implementation errors or protocol errors, I think there is a bit of the confusion: the cryptography or the security property is not broken because the implementation or protocol got it wrong. The cryptography is also not broken because of side-channels attack or a physical attack. So, I think it will good to clarify what makes a cryptographic algorithm broken (when the security property is shown to be attackable in efficient time), and when implementations or protocols implement it wrongly. A cryptographic algorithm is \"broken\" only if there is a fundamental flaw or weakness in the algorithm itself that enables an efficient attack against its intended security properties. This distinction helps separate true cryptographic weaknesses from implementation or protocol mistakes: it is different at the algoritmic level or the implementation/physical/protocol layer.</p>", "time": "2024-11-07T16:14:45Z"}, {"author": "Kathleen Moriarty", "text": "<p>@Sofie, yes, those are traditionally important distinctions in standards and good to account for</p>", "time": "2024-11-07T16:16:14Z"}, {"author": "Deirdre Connolly", "text": "<p>Yep, agreed</p>", "time": "2024-11-07T16:16:19Z"}, {"author": "Pascal Sch\u00e4fer", "text": "<p>Agreed</p>", "time": "2024-11-07T16:16:49Z"}, {"author": "Richard Barnes", "text": "<p>Ah, FIPS lawyering</p>", "time": "2024-11-07T16:17:46Z"}, {"author": "Deirdre Connolly", "text": "<p>what's the FIPS equivalent of Kremlinology</p>", "time": "2024-11-07T16:18:18Z"}, {"author": "Deirdre Connolly", "text": "<p>FIPSology, I guess</p>", "time": "2024-11-07T16:18:25Z"}, {"author": "Richard Barnes", "text": "<p>Gaithersburgology?</p>", "time": "2024-11-07T16:18:43Z"}, {"author": "Richard Barnes", "text": "<p>This stuff makes me appreciate the IETF process</p>", "time": "2024-11-07T16:19:28Z"}, {"author": "Deirdre Connolly", "text": "<p>It's especially frustrating when CRYSTALS-Kyber and CRYSTALS-Dilithium were obviously specified by the same authors, and FIPS 203 and FIPS 204 obviously weren't <span aria-label=\"melting face\" class=\"emoji emoji-1fae0\" role=\"img\" title=\"melting face\">:melting_face:</span></p>", "time": "2024-11-07T16:20:16Z"}, {"author": "Phillip Hallam-Baker", "text": "<p>I have deterministic generation of keypair from seed for RSA as well</p>", "time": "2024-11-07T16:24:31Z"}, {"author": "Sofia Celi", "text": "<p>consult your lab is probably the answer ;)</p>", "time": "2024-11-07T16:24:33Z"}, {"author": "Phillip Hallam-Baker", "text": "<p>(also X25519, x448, ML-KEM, etc.)</p>", "time": "2024-11-07T16:24:52Z"}, {"author": "Andrew", "text": "<p>Isn't X25519 a 256-bit curve? So only 128-bit secure?</p>", "time": "2024-11-07T16:25:34Z"}, {"author": "Phillip Hallam-Baker", "text": "<p>@Andrew you are correct</p>", "time": "2024-11-07T16:25:55Z"}, {"author": "Phillip Hallam-Baker", "text": "<p>But the secret is 256 bits</p>", "time": "2024-11-07T16:26:04Z"}, {"author": "Sofia Celi", "text": "<p>yes, 128 of sec</p>", "time": "2024-11-07T16:26:17Z"}, {"author": "Andrew", "text": "<p>Thanks, I was assuming Mike just misspoke.</p>", "time": "2024-11-07T16:26:39Z"}, {"author": "Richard Barnes", "text": "<p><span aria-label=\"+1\" class=\"emoji emoji-1f44d\" role=\"img\" title=\"+1\">:+1:</span></p>", "time": "2024-11-07T16:26:47Z"}, {"author": "Richard Barnes", "text": "<p>what Mike is saying about MLS seems accurate, but also totally unenforceable</p>", "time": "2024-11-07T16:27:03Z"}, {"author": "Sofia Celi", "text": "<p>I have not heard of any company so far of trying to achieve fedramp auth for MLS, so it will depend, I think</p>", "time": "2024-11-07T16:27:39Z"}, {"author": "Richard Barnes", "text": "<p>MLS (MLSpp) is in Webex FedRAMP</p>", "time": "2024-11-07T16:27:57Z"}, {"author": "Richard Barnes", "text": "<p>but AFAIK, only the crypto module needs to be verified</p>", "time": "2024-11-07T16:28:07Z"}, {"author": "Sofia Celi", "text": "<p>oh! do you have public documentation?</p>", "time": "2024-11-07T16:28:10Z"}, {"author": "Sofia Celi", "text": "<p>yes, same as TLS I will assume</p>", "time": "2024-11-07T16:28:18Z"}, {"author": "Richard Barnes", "text": "<p>and if the crypto module is verified, and allows seeds to be input, then you can do MLS</p>", "time": "2024-11-07T16:28:32Z"}, {"author": "Sofia Celi", "text": "<p>so falling back to the crypto implementation having the certification, I would assume</p>", "time": "2024-11-07T16:28:34Z"}, {"author": "Richard Barnes", "text": "<p>like, i'm not aware of NIST having a Protocol Library Validation Program, just a CMVP</p>", "time": "2024-11-07T16:29:28Z"}, {"author": "Deirdre Connolly", "text": "<p>interesting; how does TLS 1.3 fall into being FIPS, then</p>", "time": "2024-11-07T16:29:30Z"}, {"author": "Alicja Kario", "text": "<p>Robert Relya (OASIS co-chair) is my co-worker</p>", "time": "2024-11-07T16:29:31Z"}, {"author": "Alicja Kario", "text": "<p>I will talk with him about it</p>", "time": "2024-11-07T16:29:39Z"}, {"author": "Sofia Celi", "text": "<p>yeah, when I used to do fedramp auth, it was the same</p>", "time": "2024-11-07T16:29:49Z"}, {"author": "Sofia Celi", "text": "<p>TLS 1.3 is not like fips, because it is a protocol: the algorithms have to be though and some of the combiners</p>", "time": "2024-11-07T16:30:17Z"}, {"author": "Aron Wussler", "text": "<p>SP800-108 too</p>", "time": "2024-11-07T16:30:38Z"}, {"author": "Richard Barnes", "text": "<p>@Sofia <a href=\"https://blog.webex.com/collaboration/hybrid-work/webex-government-portfolio/\">https://blog.webex.com/collaboration/hybrid-work/webex-government-portfolio/</a> -- note the \"Zero Trust\" part, which is based on MLS</p>", "time": "2024-11-07T16:30:44Z"}, {"author": "Britta Hale", "text": "<p>There are efforts on NIAP approval now, and it is in requirements on others for FEDRAMP.</p>", "time": "2024-11-07T16:31:19Z"}, {"author": "Deirdre Connolly", "text": "<p>kk; I think the usage /derivation of ML-KEM keys in MLS via this derivation KDF may still get in the way</p>", "time": "2024-11-07T16:31:38Z"}, {"author": "Sofia Celi", "text": "<p>thank you @richard!</p>", "time": "2024-11-07T16:31:50Z"}, {"author": "Sean Turner", "text": "<p>Check out appendix a of PKcS#11: <a href=\"http://docs.oasis-open.org/pkcs11/pkcs11-base/v2.40/os/pkcs11-base-v2.40-os.html\">http://docs.oasis-open.org/pkcs11/pkcs11-base/v2.40/os/pkcs11-base-v2.40-os.html</a></p>", "time": "2024-11-07T16:33:05Z"}, {"author": "Sean Turner", "text": "<p>Some of those names are familiar.  Thomas Hardjono is here in Ireland.</p>", "time": "2024-11-07T16:34:12Z"}, {"author": "Sofia Celi", "text": "<p>re fips and TLS 1.3, there was an interesting discussion on it at some point at the TLS WG ML. Rich also has a lot of input on that.</p>", "time": "2024-11-07T16:35:12Z"}, {"author": "Kathleen Moriarty", "text": "<p>Sean, what version do you see Thomas on? Robert Griffin is lovely and very likely retired.</p>", "time": "2024-11-07T16:36:29Z"}, {"author": "Alicja Kario", "text": "<p>there was also a new info that KMAC() as a KDF doesn't allow concatenation of the shared secrets from a non-approved algorithm (I think on the OpenPGP meeting)</p>", "time": "2024-11-07T16:37:33Z"}, {"author": "Alicja Kario", "text": "<p>Quick note on that: there will be support for that in PKCS#11 v3.2</p>", "time": "2024-11-07T16:45:34Z"}, {"author": "John Gray", "text": "<p>PKCS11 3.2 will support the context.   They use a parameter.  I talked to the author about it at ICMC in September</p>", "time": "2024-11-07T16:46:11Z"}, {"author": "Phillip Hallam-Baker", "text": "<p>I was reading through SP-whatever, there is TONS of stuff they need to fix there.</p>", "time": "2024-11-07T16:46:59Z"}, {"author": "Phillip Hallam-Baker", "text": "<p>SP800 is going to need a revision pretty soon</p>", "time": "2024-11-07T16:47:25Z"}, {"author": "John Gray", "text": "<p>I asked the author on the Mike at ICMC and he definitely said they were supporting the context in PKCS11 3.2 via their parameter object.   That makes total sense.</p>", "time": "2024-11-07T16:47:45Z"}, {"author": "John Gray", "text": "<p>I meant on the microphone, not Mike... Lol..</p>", "time": "2024-11-07T16:48:17Z"}, {"author": "Phillip Hallam-Baker", "text": "<p>Hybrids are going to be NECESSARY util there is threshold PQC</p>", "time": "2024-11-07T16:48:31Z"}, {"author": "Yoav Nir", "text": "<p>\"simple\" and \"lightweight\" bring you bad luck</p>", "time": "2024-11-07T16:49:51Z"}, {"author": "Daniel Huigens", "text": "<p>\"limited additions\"</p>", "time": "2024-11-07T16:50:10Z"}, {"author": "Conner Ybarra", "text": "<p>\"PQ DLC\"</p>", "time": "2024-11-07T16:51:31Z"}, {"author": "Phillip Hallam-Baker", "text": "<p>@Daniel the limit was to the SIZE of the addition, not the number</p>", "time": "2024-11-07T16:52:09Z"}, {"author": "Phillip Hallam-Baker", "text": "<p>Hey, is SSH being updated to ML-KEM anywhere?</p>", "time": "2024-11-07T16:53:11Z"}, {"author": "Alicja Kario", "text": "<p>yes</p>", "time": "2024-11-07T16:53:20Z"}, {"author": "Alicja Kario", "text": "<p>draft-kampanakis-curdle-ssh-pq-ke</p>", "time": "2024-11-07T16:53:41Z"}, {"author": "Phillip Hallam-Baker", "text": "<p>Ah, they recstarted...</p>", "time": "2024-11-07T16:53:45Z"}, {"author": "Yaroslav Rosomakho", "text": "<p>draft-kampanakis-curdle-ssh-pq-ke</p>", "time": "2024-11-07T16:53:53Z"}, {"author": "John Gray", "text": "<p>The Oasis person I talked to at ICMC was Robert Relyea who is specifying PKCS#11 3.2 as part of Oasis, he went over the ML-DSA API in 3.2, it takes a PARAMS mechanism that supports the context.   He had mention 3.2 was coming out by the end of this year.</p>", "time": "2024-11-07T16:54:39Z"}, {"author": "Mike Ounsworth", "text": "<p><span class=\"user-mention silent\" data-user-id=\"5147\">Andrew</span> <a href=\"#narrow/stream/360-pquip/topic/ietf-121/near/144059\">said</a>:</p>\n<blockquote>\n<p>Thanks, I was assuming Mike just misspoke.</p>\n</blockquote>\n<p>I believe said \"ML-KEM-* and X25519 each have up to 256 bits, so you can do both halfs off the same 32 byte seed without entropy issues\". Or something like that.</p>", "time": "2024-11-07T16:54:43Z"}, {"author": "Deirdre Connolly", "text": "<p><span class=\"user-mention silent\" data-user-id=\"677\">Phillip Hallam-Baker</span> <a href=\"#narrow/stream/360-pquip/topic/ietf-121/near/144123\">said</a>:</p>\n<blockquote>\n<p>Hey, is SSH being updated to ML-KEM anywhere?</p>\n</blockquote>\n<p>Already implemented as hybrid kex in OpenSSH</p>", "time": "2024-11-07T16:55:02Z"}, {"author": "Phillip Hallam-Baker", "text": "<p>@Deirdre, thanks. Guess should have come to Vancouver etc.</p>", "time": "2024-11-07T16:55:59Z"}, {"author": "Jonathan Hammell", "text": "<p>OpenSSH has implemented draft-ietf-sshm-ntruprime-ssh, not draft-kampanakis-curdle-ssh-pq-ke</p>", "time": "2024-11-07T16:57:42Z"}, {"author": "Yaroslav Rosomakho", "text": "<blockquote>\n<p>ssh -V<br>\nOpenSSH_9.9p1, OpenSSL 3.3.2 3 Sep 2024<br>\nssh -Q kex | grep -E 'mlkem|sntrup'<br>\nsntrup761x25519-sha512<br>\n<a href=\"mailto:sntrup761x25519-sha512@openssh.com\">sntrup761x25519-sha512@openssh.com</a><br>\nmlkem768x25519-sha256</p>\n</blockquote>", "time": "2024-11-07T16:57:58Z"}, {"author": "Deirdre Connolly", "text": "<p>They have their own ML-KEM support as well</p>", "time": "2024-11-07T16:58:02Z"}, {"author": "Mike Ounsworth", "text": "<p><span class=\"user-mention\" data-user-id=\"4773\">@Sophie Schmieg</span> pointed out that the API for \"RealHash-ML-DSA\" on my slide is in fact wrong, which really underscores that this really <em>really</em> needs to be written down clearly as a standalone algorithm.</p>", "time": "2024-11-07T16:58:07Z"}, {"author": "John Preu\u00df Mattsson", "text": "<p><a href=\"https://undeadly.org/cgi?action=article;sid=20240921181110\">https://undeadly.org/cgi?action=article;sid=20240921181110</a></p>", "time": "2024-11-07T16:58:25Z"}, {"author": "Jonathan Hammell", "text": "<p>Oh, thanks for correcting me <span class=\"user-mention\" data-user-id=\"4721\">@Yaroslav Rosomakho</span>  and <span class=\"user-mention\" data-user-id=\"603\">@Deirdre Connolly</span>.  I didn't know ML-KEM was included already.  That's great.</p>", "time": "2024-11-07T16:58:48Z"}]