[{"author": "Robert Moskowitz", "text": "

20f right now.

", "time": "2026-03-18T06:00:19.000Z"}, {"author": "Raphael Robert", "text": "

yes

", "time": "2026-03-18T06:00:45.000Z"}, {"author": "James Howe", "text": "

yes we hear you

", "time": "2026-03-18T06:00:47.000Z"}, {"author": "Robert Moskowitz", "text": "

Yes1

", "time": "2026-03-18T06:00:47.000Z"}, {"author": "Steffen Fries", "text": "

yes

", "time": "2026-03-18T06:00:47.000Z"}, {"author": "Richard Barnes", "text": "

https://www.seriouseats.com/soft-and-chewy-sugar-cookie-recipe for those wondering

", "time": "2026-03-18T06:06:47.000Z"}, {"author": "Ted Hardie", "text": "

@Rohan can you say more about the use cases that require a single message going to the list rather that individual messages to each?

", "time": "2026-03-18T06:18:34.000Z"}, {"author": "Ted Hardie", "text": "

I have to say that doesn't sound compelling compared to individual messages to each moderator, but I may be missing a scaling problem here.

", "time": "2026-03-18T06:20:09.000Z"}, {"author": "Marta Mularczyk", "text": "

what was the email address for the slides?

", "time": "2026-03-18T06:24:40.000Z"}, {"author": "Paul Wouters", "text": "

mls-chairs@ietf.org

", "time": "2026-03-18T06:24:52.000Z"}, {"author": "David Schinazi", "text": "

mls-chairs@ietf.org

", "time": "2026-03-18T06:24:55.000Z"}, {"author": "Martin Thomson", "text": "

Can I assume that this recurses, so you can have a user with multiple devices in a virtual group that is part of a support team virtual group that is part of a larger group?

", "time": "2026-03-18T06:26:27.000Z"}, {"author": "Rohan Mahy", "text": "

Ted Hardie said:

\n
\n

I have to say that doesn't sound compelling compared to individual messages to each moderator, but I may be missing a scaling problem here.

\n
\n

There are two properties which are important:

\n", "time": "2026-03-18T06:28:06.000Z"}, {"author": "Alex Lee", "text": "

Martin Thomson said:

\n
\n

Can I assume that this recurses, so you can have a user with multiple devices in a virtual group that is part of a support team virtual group that is part of a larger group?

\n
\n

That would be very insecure / hiding membership identities and ghost users. Are there no bounds on what may. or may not be a virtual client?

", "time": "2026-03-18T06:28:18.000Z"}, {"author": "Martin Thomson", "text": "

@Alex Lee How would that be different than me giving one of my devices to a spy though?

", "time": "2026-03-18T06:29:10.000Z"}, {"author": "Ted Hardie", "text": "

@Rohan would another way to accomplish that be to put the moderators into a separate group and allow the members to be external senders to that group?

", "time": "2026-03-18T06:29:28.000Z"}, {"author": "Martin Thomson", "text": "

...and how would that be different even with one layer of virtual client?

", "time": "2026-03-18T06:29:59.000Z"}, {"author": "Rohan Mahy", "text": "

Martin Thomson said:

\n
\n

Can I assume that this recurses, so you can have a user with multiple devices in a virtual group that is part of a support team virtual group that is part of a larger group?

\n
\n

If you had a gigantic retail company like Walmart, you could have a virtual group for Alice's clients. Then Alice could be in a virtual group for her store. The store could then be in a \"real\" group for all the stores.

", "time": "2026-03-18T06:30:01.000Z"}, {"author": "Martin Thomson", "text": "

@Rohan Mahy you could have just said \"yes\" :)

", "time": "2026-03-18T06:30:25.000Z"}, {"author": "Nick Sullivan", "text": "

https://datatracker.ietf.org/meeting/125/session/cfrg <- you can also upload here

", "time": "2026-03-18T06:30:56.000Z"}, {"author": "Ted Hardie", "text": "

@Rohan Sorry, I meant \"members of the group being moderated to send to the (separate) group of moderators)\"

", "time": "2026-03-18T06:31:01.000Z"}, {"author": "Richard Barnes", "text": "

is this a scalability dimension that anyone cares about? i guess maybe with ML-DSA?

", "time": "2026-03-18T06:31:42.000Z"}, {"author": "David Schinazi", "text": "

@Nick correct URL is
\nhttps://datatracker.ietf.org/meeting/125/session/mls

", "time": "2026-03-18T06:31:50.000Z"}, {"author": "Rohan Mahy", "text": "

Ted Hardie said:

\n
\n

@Rohan would another way to accomplish that be to put the moderators into a separate group and allow the members to be external senders to that group?

\n
\n

MLS currently has no way to send application messages from external senders. Only handshake messages can be external (external commits and proposals), and they are never encrypted.

", "time": "2026-03-18T06:31:55.000Z"}, {"author": "Deirdre Connolly", "text": "

Richard Barnes said:

\n
\n

is this a scalability dimension that anyone cares about? i guess maybe with ML-DSA?

\n
\n

definitely for ML-DSA

", "time": "2026-03-18T06:34:10.000Z"}, {"author": "Ted Hardie", "text": "

@Rohan if the answer to question 1 in https://datatracker.ietf.org/meeting/125/materials/slides-125-mls-targeted-messages-00 is \"yes\", does that change in a useful way for this?

", "time": "2026-03-18T06:34:14.000Z"}, {"author": "Alex Lee", "text": "

Martin Thomson said:

\n
\n

...and how would that be different even with one layer of virtual client?

\n
\n

There is probably a difference between ghost-users-by-IETF-design and misbehaving clients. Obviously, there is also the case that an insider threat copying data off to the side means group members can repudiate messages, while having hidden group members means that there is non-repudiation to unknown members.

", "time": "2026-03-18T06:34:28.000Z"}, {"author": "Deirdre Connolly", "text": "

Deirdre Connolly said:

\n
\n

Richard Barnes said:

\n
\n

is this a scalability dimension that anyone cares about? i guess maybe with ML-DSA?

\n
\n

definitely for ML-DSA

\n
\n

especially since even the lowest/smallest param for ML-DSA (-44) isn' t being registered for MLS

", "time": "2026-03-18T06:34:37.000Z"}, {"author": "Ted Hardie", "text": "

@Rohan (will follow up further offline, as we are well past this in the meeting).

", "time": "2026-03-18T06:34:54.000Z"}, {"author": "Mallory Knodel", "text": "

@alex An individual could use a client to automate language translation. Or perform actions based on watch keywords. Or spam all of your contacts a birthday invitation. I think it\u2019s a cool hook for many end usery things and the architecture Konrad presented seems like the best one.

", "time": "2026-03-18T06:34:55.000Z"}, {"author": "Alex Lee", "text": "

It seems that virtual sub-clients to a single identity is fundamentally different than grouped clients

", "time": "2026-03-18T06:34:57.000Z"}, {"author": "Mallory Knodel", "text": "

But also yes to all of the biz use cases others shared.

", "time": "2026-03-18T06:35:56.000Z"}, {"author": "Sean Turner", "text": "

@Marta can you provide privileges so I can snag the file?

", "time": "2026-03-18T06:36:07.000Z"}, {"author": "Martin Thomson", "text": "

@Alex Lee In terms of visibility in the outer group, the virtual client does hide membership. So that is a risk you need to tolerate. But the core thing here is that MLS gives you the ability to have a group of people agree on a key. If that is the secret keys necessary to participate as a single user in a group, I don't see how that is something you can stop.

", "time": "2026-03-18T06:36:25.000Z"}, {"author": "Sean Turner", "text": "

@Marta - Nick has the slides he's uploading.

", "time": "2026-03-18T06:36:57.000Z"}, {"author": "Mallory Knodel", "text": "

And importantly not on a per group basis

", "time": "2026-03-18T06:37:00.000Z"}, {"author": "Martin Thomson", "text": "

If your goal is to ensure that the membership of these \"subgroups\" is visible to the larger group, I don't think that you can enforce anything. You might add extensions to allow the subgroups to make their structure visible, but it would have to be voluntary.

", "time": "2026-03-18T06:37:36.000Z"}, {"author": "Mallory Knodel", "text": "

Yall need a moderator group chat ;)

", "time": "2026-03-18T06:37:37.000Z"}, {"author": "Martin Thomson", "text": "

Mallory Knodel said:

\n
\n

Yall need a moderator group chat ;)

\n
\n

Zulip has topics.

", "time": "2026-03-18T06:37:58.000Z"}, {"author": "Martin Thomson", "text": "

Is the reason that you want to sign the leaf (plus hashes) rather than just the outer message because you have reasons to redistribute the leaf node without the outer message?

", "time": "2026-03-18T06:39:26.000Z"}, {"author": "Martin Thomson", "text": "

...but doesn't that mean you need to carry the hashes around when you use the leaf? Are there any cases where the outer message might be leaked to a recipient of the leaf node in a bad way?

", "time": "2026-03-18T06:40:37.000Z"}, {"author": "Bas Westerbaan", "text": "

Martin Thomson said:

\n
\n

Is the reason that you want to sign the leaf (plus hashes) rather than just the outer message because you have reasons to redistribute the leaf node without the outer message?

\n
\n

If that is it, then that can be easily attained with a two leaf Merkle tree and only one signature.

", "time": "2026-03-18T06:40:43.000Z"}, {"author": "Rohan Mahy", "text": "

Ted Hardie said:

\n
\n

@Rohan if the answer to question 1 in https://datatracker.ietf.org/meeting/125/materials/slides-125-mls-targeted-messages-00 is \"yes\", does that change in a useful way for this?

\n
\n

My answer to question 1 is no. I'm not convinced that we can do this safely enough, and we would need more changes that would be non-trivial to incorporate. As Raphael said, you would substantially lose PCS with this choice.

", "time": "2026-03-18T06:40:50.000Z"}, {"author": "Martin Thomson", "text": "

@Bas Westerbaan why complicate things? I'm not following what you would gain from that.

", "time": "2026-03-18T06:41:48.000Z"}, {"author": "Richard Barnes", "text": "

IMG_1987 copy.jpg

\n
", "time": "2026-03-18T06:43:03.000Z"}, {"author": "Sean Turner", "text": "

nice

", "time": "2026-03-18T06:43:13.000Z"}, {"author": "Richard Barnes", "text": "

i don't think i made quite enough for the group so some folks might have to share

", "time": "2026-03-18T06:43:16.000Z"}, {"author": "Martin Thomson", "text": "

Send some over @Richard Barnes ?

", "time": "2026-03-18T06:43:21.000Z"}, {"author": "Richard Barnes", "text": "

Come on over!

", "time": "2026-03-18T06:43:27.000Z"}, {"author": "Rohan Mahy", "text": "

Trade you @Richard Barnes
\n96a6ee6a-0712-45a3-a3c0-ff5f3e34b326~1.jpg

\n
", "time": "2026-03-18T06:47:33.000Z"}, {"author": "Robert Moskowitz", "text": "

Green Kitkats?

", "time": "2026-03-18T06:49:58.000Z"}, {"author": "Robert Moskowitz", "text": "

Mint flavored?

", "time": "2026-03-18T06:50:28.000Z"}, {"author": "Sean Turner", "text": "

definitely not mint more like green tea

", "time": "2026-03-18T06:50:40.000Z"}, {"author": "Christopher Inacio", "text": "

macha

", "time": "2026-03-18T06:50:41.000Z"}, {"author": "Rohan Mahy", "text": "

matcha

", "time": "2026-03-18T06:50:46.000Z"}, {"author": "Martin Thomson", "text": "

The little leaf made me think basil rather than matcha, but of course tea looks like that too.

", "time": "2026-03-18T06:51:11.000Z"}, {"author": "Richard Barnes", "text": "

i had a matcha IPA from Kyoto with my ramen dinner, in honor of St. Patrick's Day
\nIMG_1984 copy.jpg

\n
", "time": "2026-03-18T06:51:28.000Z"}, {"author": "Robert Moskowitz", "text": "

I am a heavy tea drinker. I just don't find green tea interesting.

", "time": "2026-03-18T06:51:40.000Z"}, {"author": "Deirdre Connolly", "text": "

matcha is one way to achieve green beer

", "time": "2026-03-18T06:52:04.000Z"}, {"author": "Robert Moskowitz", "text": "

I have spent whole afternoons at Twinings when we have been in London.

", "time": "2026-03-18T06:52:11.000Z"}, {"author": "Richard Barnes", "text": "

the matcha flavor was very subtle, mainly a little hint on the back end

", "time": "2026-03-18T06:52:46.000Z"}, {"author": "Russ Housley", "text": "

Seems like the two-party profile could live in MLS, even if the other work goes or doesn't.

", "time": "2026-03-18T06:56:46.000Z"}, {"author": "Richard Barnes", "text": "

i'm just concerned that there be some consumer for the work

", "time": "2026-03-18T06:57:06.000Z"}, {"author": "Richard Barnes", "text": "

cookies are good, maybe a little too much nutmeg

", "time": "2026-03-18T06:58:16.000Z"}, {"author": "Richard Barnes", "text": "

+1 to \"amortized\"

", "time": "2026-03-18T06:59:33.000Z"}, {"author": "Bas Westerbaan", "text": "

Amortised as name is an improvement, yes.

", "time": "2026-03-18T06:59:44.000Z"}, {"author": "Richard Barnes", "text": "

i'm impressed that this is Safe, but that seems like nice validation of the API

", "time": "2026-03-18T07:01:58.000Z"}, {"author": "Bas Westerbaan", "text": "

So what do we do after Q-day? Keep the slower rhythm or speed up?

", "time": "2026-03-18T07:03:43.000Z"}, {"author": "Carlos Melchor", "text": "

@bas IIUC the application controls the frequency of the exchanges.

", "time": "2026-03-18T07:04:51.000Z"}, {"author": "Rohan Mahy", "text": "

FYI: I went through and proposed tentative mls-extension numbers and component IDs for all the active mls and mimi drafts. basically assigning them WG drafts in order of -00 publication, then the individual drafts in order of -00 publication. That's where the 0x0006 came from in the amortized draft

", "time": "2026-03-18T07:05:09.000Z"}, {"author": "Richard Barnes", "text": "

nope, do it

", "time": "2026-03-18T07:05:48.000Z"}, {"author": "Richard Barnes", "text": "

@Marta Mularczyk the chirping birds in the background were a lovely addition!

", "time": "2026-03-18T07:07:02.000Z"}, {"author": "Martin Thomson", "text": "

Just say no to proof of work

", "time": "2026-03-18T07:13:05.000Z"}, {"author": "David Schinazi", "text": "

Friends don't let friends proof-of-work

", "time": "2026-03-18T07:13:31.000Z"}, {"author": "Martin Thomson", "text": "

the PoW-TPM combination from Monday was the best combination I've seen so far

", "time": "2026-03-18T07:14:07.000Z"}, {"author": "Mallory Knodel", "text": "

Please kill the poll so we can see the slides

", "time": "2026-03-18T07:14:38.000Z"}, {"author": "Richard Barnes", "text": "

did he say certs are \"broke\" or \"baroque\"?

", "time": "2026-03-18T07:14:45.000Z"}, {"author": "David Schinazi", "text": "

baroque

", "time": "2026-03-18T07:14:53.000Z"}, {"author": "Mallory Knodel", "text": "

Ty

", "time": "2026-03-18T07:14:55.000Z"}, {"author": "Richard Barnes", "text": "

not that i would object to either option

", "time": "2026-03-18T07:15:02.000Z"}, {"author": "Martin Thomson", "text": "

He says X.509 is baroque, then sings the praises of SD-CWT...

", "time": "2026-03-18T07:15:17.000Z"}, {"author": "Richard Barnes", "text": "

as if SD-[JC]WT aren't baroque

", "time": "2026-03-18T07:15:17.000Z"}, {"author": "Martin Thomson", "text": "

jinx

", "time": "2026-03-18T07:15:22.000Z"}, {"author": "Martin Thomson", "text": "

It's only 33% larger than necessary Rohan

", "time": "2026-03-18T07:15:55.000Z"}, {"author": "David Schinazi", "text": "

We could come up with our own bespoke baroque encoding!

", "time": "2026-03-18T07:16:18.000Z"}, {"author": "Richard Barnes", "text": "

of course, SD-CWT is not a straightforward translation of SD-JWT to CBOR, bc apparently Rohan has lost his mind

", "time": "2026-03-18T07:16:36.000Z"}, {"author": "Martin Thomson", "text": "

I think that ASN.1 + XER is well-suited to inclusion in MLS

", "time": "2026-03-18T07:16:45.000Z"}, {"author": "David Schinazi", "text": "

Needs more QUIC varints

", "time": "2026-03-18T07:17:21.000Z"}, {"author": "Martin Thomson", "text": "

If we want to play format games: https://github.com/martinthomson/aweson

", "time": "2026-03-18T07:18:00.000Z"}, {"author": "Deirdre Connolly", "text": "

am i weird for pronouncing it 'jay double-you tee'

", "time": "2026-03-18T07:18:23.000Z"}, {"author": "Richard Barnes", "text": "

i tried to make the Welsh pronunciation \"joot\" the standard, but Mike Jones refused to yield

", "time": "2026-03-18T07:18:54.000Z"}, {"author": "Martin Thomson", "text": "

Deirdre Connolly said:

\n
\n

am i weird for pronouncing it 'jay double-you tee'

\n
\n

Absolutely not. I think that \"jot\" is weird.

", "time": "2026-03-18T07:18:56.000Z"}, {"author": "David Schinazi", "text": "

The RFC says to pronounce them jawts (though I find that very odd myself)

", "time": "2026-03-18T07:18:57.000Z"}, {"author": "Sean Turner", "text": "

I like Deirdre's pronunciation!

", "time": "2026-03-18T07:19:16.000Z"}, {"author": "Martin Thomson", "text": "

@David Schinazi depends on where you come from, but in some accents, \"jawt\" and \"jot\" are indistinguishable

", "time": "2026-03-18T07:19:39.000Z"}, {"author": "Martin Thomson", "text": "

Jay dubya tea

", "time": "2026-03-18T07:20:06.000Z"}, {"author": "David Schinazi", "text": "

They're only jots if they're from the JSON region of France. Otherwise they're just sparkling tokens

", "time": "2026-03-18T07:20:25.000Z"}, {"author": "Richard Barnes", "text": "

Martin Thomson said:

\n
\n

Jay dubya tea

\n
\n

i think this pronunciation has to be made with sweet tea

", "time": "2026-03-18T07:20:34.000Z"}, {"author": "Martin Thomson", "text": "

Boba

", "time": "2026-03-18T07:20:47.000Z"}, {"author": "Richard Barnes", "text": "

for some reason, i find \"jawt\" less offensive than \"cawt\" for CWT

", "time": "2026-03-18T07:21:32.000Z"}, {"author": "Martin Thomson", "text": "

\"caught\"?

", "time": "2026-03-18T07:21:44.000Z"}, {"author": "Richard Barnes", "text": "

well yes

", "time": "2026-03-18T07:21:54.000Z"}, {"author": "Martin Thomson", "text": "

\"jort\" is a type of hard-wearing fabric, right?

", "time": "2026-03-18T07:22:20.000Z"}, {"author": "David Schinazi", "text": "

c\u00f4te?

", "time": "2026-03-18T07:22:24.000Z"}, {"author": "Nick Sullivan", "text": "

cot/caught merger?

", "time": "2026-03-18T07:22:28.000Z"}, {"author": "Martin Thomson", "text": "

cord ?

", "time": "2026-03-18T07:22:35.000Z"}, {"author": "Richard Barnes", "text": "

David Schinazi said:

\n
\n

c\u00f4te?

\n
\n

definitely not

", "time": "2026-03-18T07:23:03.000Z"}, {"author": "Deirdre Connolly", "text": "

i like this idea (hashing the group context and signing that, allowing private rejection at the DS)

", "time": "2026-03-18T07:23:42.000Z"}, {"author": "Micke Nordin", "text": "

Regarding federation: I know the social web WG over at W3C is doing mls over activitypub

", "time": "2026-03-18T07:26:46.000Z"}, {"author": "Micke Nordin", "text": "

Apparently they have two implementations they are working on

", "time": "2026-03-18T07:27:33.000Z"}, {"author": "Sean Turner", "text": "

@Micke: we need to revive the federation draft.

", "time": "2026-03-18T07:28:32.000Z"}, {"author": "Carlos Melchor", "text": "

thank you!

", "time": "2026-03-18T07:29:02.000Z"}, {"author": "Raphael Robert", "text": "

more time for vibe coding

", "time": "2026-03-18T07:29:13.000Z"}, {"author": "Mojtaba Chenani", "text": "

Thank you!

", "time": "2026-03-18T07:29:17.000Z"}, {"author": "Robert Moskowitz", "text": "

Too cold here to dive into the pool.

", "time": "2026-03-18T07:29:22.000Z"}, {"author": "Robert Moskowitz", "text": "

bye

", "time": "2026-03-18T07:29:26.000Z"}, {"author": "Micke Nordin", "text": "

@Sean: Probably good idea, I wrote a toy implementation too of MLS over AP after their docs: https://swicg.github.io/activitypub-e2ee/mls

", "time": "2026-03-18T07:30:27.000Z"}]