Private Advertising Technology CG

29 October 2021


alextcone, AramZS, blassey, bmay_, btsavage, charlieharrison, cpn, dmarti, ErikAnderson, eriktabueneck, hober, jatindermann, jeff, johnwilander, Kazuhiro_Hoya, kleber, luk-wlodarczyk, martinthomson, mib_guprew, mjv_, npdoty, rowan_m, wbaker, weiler, wseltzer
AramZS, spt
btsavage, wseltzer

Meeting minutes

<mib_guprew> mib_guprew->seanturner

<wbaker> adbg is "pro slide" ... checks notes. ok, yup.

<martinthomson> scribe btsavage

<johnwilander> No recording, please.

AramZS: Welcome. This is the first PAT-CG

AramZS: Intros. Info on GitHub. Thanks to folks at Privacy-CG. Borrowed from setup, documentation, repos.

AramZS: Admin, meetings and proposals repo.

AramZS: I am Aram zucker-scharff from WaPo. I'm one of the chairs. I have no proposals of my own. Will be as neutral as possible.

Sean: I'm coming in from IETF. I'll be paid by a pool of people. I'll be independent and rule on consensus. Will only judge proposals on merits - no intentions to bring my own.

<wseltzer> https://github.com/patcg

AramZS: There is an email list, but try to use GitHub if possible.

<wseltzer> https://patcg.github.io/charter.html

AramZS: Let's walk through the charter. Specifically we are looking to incubate web features for advertising purposes that are in the interests of users.

AramZS: This is pretty much open to everyone. It's a community group.

AramZS: Looking to incubate new features and policies around them to provide users with privacy guarantees that have a technical basis.

AramZS: Charter is public and on GitHub. You can leave feedback on GibHub in admin repo.

AramZS: Background on why GitHub. It's open, all in one place, easier to track. If you want to propose text yourself - we can discuss that.

AramZS: Will look to minute in IRC. It's a Zoom chat. Looking to move over to WebEx which we hope will be easier for people.

bmay_: There was a lot of good links on that slide presentation can you share those links.

AramZS: It's linked as a slideshow and a PDF on GitHub

<AramZS> Link: https://github.com/patcg/meetings/tree/main/2021/10-tpac

johnwilander: We are here as observers, not members of the CG as of now. Is there any bar a proposal needs to meet to become a work item in this CG? Such as parties expressing interest?

AramZS: Looking for proposals to be implemented eventually. Must be a proposal that addresses advertising concerns while providing strong privacy protection. Not here to do pure privacy stuff.

johnwilander: Many of us are struggling to take active part in all the groups active in this space. One way to limit the time you have to spend is to only spend it on things which are on path to become standards. But incubation is incubation and I'm all for that. So I'm wondering what your bar will be.

AramZS: That gets us to our first work item; to make a working group charter. What will move on to standardization.

npdoty: Nick Doty, CDT. Thanks for getting us started. I'll try to stick with GitHub. I do sometimes get a little bit frustrated with not doing anything over email. Issues being opened in a repo is awkward. Can we use a public mailing list for announcements? I'd like to have some discussion over email and GitHub for specific issues.

<alextcone> "Watch" feature on GitHub plus setting up GitHub notifications

AramZS: There is an email list. Everyone who joins the community group is automatically added. You can sign up for email alerts for the things you want. Scheduled events and big changes can be supplemented by an email as well. W3 has a shared calendar which is also useful to have.

Sean: We are not trying to surprise anyone. We plan to follow the rules and notify the group of things.

mjv_: <<<silence>>>

<alextcone> If we can avoid email threads that would be fantastic. Email threads for W3C are particularly gnarly to follow.

<npdoty> yeah, great, I trust you all! I just found the charter surprisingly detailed and obscure about notification explicitly being through issues on particular repos

ekr: I do share John's concern about the number of venues. This should subsume most of the discussion on privacy preserving advertising. That was the understanding that the people who set this up had. I'll be objecting to substantial discussion in other venues. That's how to manage load. I don't have a particular bar in mind, but there should be substantial interest before we adopt things. We don't want 20 things that have little interest taki[CUT]

ekr: When making progress need one thing per area.

AramZS: There are a number of different proposals that will move here.

<ekr> I too would prefer Google docs

eriktabueneck: I've seen Google Docs work way better than IRC for minuting and queuing.

<kleber> +1 to shared google doc

<luk-wlodarczyk> +1 for Google Docs

<npdoty> +1 for IRC rather than all constantly typing into a Google Doc

<bmay_> +1 on google doc

+1 to shared Google Docs

<johnwilander> I very much agree with Eric that we need to have a fairly high bar. Spending hours on end on specs that have no known path to standardization is holding back work on specs that do have such a path.

<robin> let's split it both ways and use cryptpad

<kleber> worst of both worlds...

<pbannist> +1 for Google docs

<ekr> why not both?!

<josh_karlin> +1 for google docs

<ekr> (that was a joke)

<blassey> -1 on crypt pad

AramZS: Let's put a pin in it. I like IRC.

<blassey> I am also an IRC fan

<bmay_> +1 on trying gdocs next meeting.

<ekr> s/martin/marshall /

mjv: Thanks Aram and Sean. On behalf of the Chrome team working on Privacy Sandbox want to say we are excited. This group will provide a much needed forum to incubate. These proposals are important to support advertising use-cases and provide privacy.

mjv: Chrome is interested in moving over FloC and Fledge and others to this group.

mjv: We will have some new code about topics showing up soon. Just a heads up. This group is the right place for such proposals.

<npdoty> <3

<johnwilander> Regarding the bar, the "Document Adoption" section says "further work is likely to lead to independent interoperable implementations."

ekr: This question about Docs vs IRC raises a more fundamental question. Conventional minuting is this point by point recording of everything. I find it useless. We make a recording if we want that. This is not a working group - so we don't need that. If it's not formally required can we NOT do that.

Wendy: A record must be kept, but doesn't stipulate the level of detail. Arguments advanced should be there. Community Groups have some flexibility. Up to the group and the chairs.

<npdoty> I find minutes very useful in case I can't make it to a meeting. and much easier than watching a recording (or inhibiting participants by recording every call)

ekr: I move we only minute the positions - not everything.

<blassey> btsavage: as a heads up, you can use "..." to represent the previously named speaker rather than retyping their name each time

AramZS: I'd like to keep it detailed for now - let's put a pin in it. We can discuss it later.

<ekr> @ndoty: automatic transcription is now good enough that it probably could substitute for people taking minutes

<AramZS> btsavage: I also see the issue with there being too many groups. Any Facebook proposals I think we will also be migrating to this group. I think this group's charter with a focus on primarily technical means of privacy meets what I see as a primary path to specification.

<kleber> (I am disturbed that we're making mt call in at 3am)

martinthomson: Thanks Ben. This is what we had in mind in the charter. The thing that will serve us here is the charter and the narrower scope. People have spoken about the need for constaints. Marshall listed a bunch of things Chrome is working on - we need to select a relatively smaller list of things to do first and keep that focus.

<eriktabueneck> +1 to kleber's concern

martinthomson: Other groups spend a lot of time shuffling through a lot of proposals, and don't make progress.

+1 about this timezone... it's 12:32 am here in Singapore

<npdoty> @ekr, I doubt that -- people are much better at getting the key details, acronyms, etc. -- but we could test it out if everyone consents to recording

bmay_: I'm also in a lot of groups. Closer to 10. I'm wondering if the other larger companies working on various proposals are willing to make the same move so that we can limit the number of groups. Microsoft PARAKEET and Apple PCM? Will they move here too?

jatindermann: I'm Jatinder from Microsoft. Looking for a venue to get all the stakeholders together. Publishers can get revenue, open and fair competiton, people get privacy. We have a few proposals and want to do origin trial and get real data and get feedback. Would love to do that with this goup. Would love to test other proposals as well.

<robin> strong support for testing, and specifically for testing PARAKEET and reporting on results

jatindermann: Tactical question; is there a regular discussion time?

<martinthomson> from my perspective, weekly meetings at this hour is a non-starter

<ekr> I would actually advocate not having a weekly meeting. Experience with QUIC and TLS indicates that it's far better to have occasional long meetings

AramZS: Plan is to start off with a weekly meeting. I'll open an issue to select the time. We appreciate that people are calling in from a variety of timezones. Will likely rotate timezones. APAC friendly one week, EMEA next, rotate back and forth.

<ekr> The problem with rotating like this is that then decisions get made and then reversed

Sean: we understand this may cause problems - but we think it's important to get input from around the world.

<ekr> ... depending on which staff there are

AramZS: Likely some repetition that way.

<martinthomson> my preference would be a mix of async work and major events

<blassey> agree with ekr's point about decisions being made and reversed

hober: wrt moving PCM or other Privacy-CG stuff over I've got a couple relevant hats. With my Apple hat off and Privacy CG hat on, I think it makes sense to have a division of labor. Technology that's specifically about private advertising is here. That seems sensible.

<ekr> hober's proposed division makes sense to me :)

hober: With Apple hat on, Privacy CG hat off, it often takes us a while to join groups. Our internal approval process is slow. Cannot make guarantees PCM can move anytime soon. I personally do not object to the idea.

<mib_guprew> @martinthomson that is the plan. All participants can help keep that on track.

bmay_: As I'm working with these other groups, I'm trying to encourage people to join W3C. Should I continue to point them at the improving web advertising group, or point them at this group.

AramZS: I think that probably both groups make sense. It does depend on what that person is interested in. I think that group is still generating a lot of useful stuff like use-cases.

AramZS: To the extent people want to be informed, the web-adv group is probably a good place to get started. Wendy?

Wendy: I take feedback about what I hear about people's interests. Community groups are under a specific license. It's a good place to work on drafts / proposals.

Wendy: The business group will aim to not overlap that. Can route people here for discussions. Not repeat.

bmay_: I'll direct people to web-adv and let them decide. Thanks Tess for the clarification.

ekr: I think the suggestion of weekly calls is quite bad. An hour leads to people swapping in state then the meeting ends. secondly, time rotation leads to different groups making decisions. How do you come to consensus. It's bad. Also tough to schedule. In my experience what works better is less common but much longer events. 4-8 hours. Rotate those.

<weiler> [ I appreciate having minuting of EKR. ]

ekr: It creates a sense of teamwork. Feel like a team spirit. Move forward. Monthly or every 6 weeks. Most discussion in GitHub.

AramZS: I think F2F would be great and lots of discussion on GitHub. Unfortunately COVID means that we are here, not F2F. Not sure people are ready for F2F. Blocking out a whole day remote seems difficult as well. We can make a plan for a different cadence once we get everything up and running.

<npdoty> I do hope we'll get to a point where people can block out a day on the calendar for virtual f2f meetings

<martinthomson> travel restrictions are still in place here; at least for 2 more days

<martinthomson> but I'm not exposing myself to COVID until my son is vaccinated

ekr: I agree F2F is not practical right now. I'd rather a 4 hour video call every month than 1 hour every week.

<blassey> my experience matches ekr's

<Zakim> kleber, you wanted to talk about web-adv vs patcg

<robin> +1 to fewer & longer meetings

kleber: I'd like to agree with ekr. Fewer, longer, in person meetings are more productive. I'd like to tie this back to the difference between this group and the web-adv group where I do support the hourly calls.

<npdoty> there can be a problem of things waiting to happen until the next call, and so monthly calls can slow things down. but if we get active ongoing asynchronous work, then we can have meetings only for longer discussion.

kleber: web-adv is at its best when it's discussing use-cases. Churn through a lot of ideas about plausible ideas for how to approach this.

<wbaker> My perception is that it took six to nine minths after the San Diego workshop on User Permission & Consent for the ADBG group to start hitting a stride and interst vortext that the first F2F was even warranted at Google in 2019-09. So we've got a bit of a "beginnings are perilous times" problem here. We need to show that patcg can come together somehow (regularly or whatever) and provide substantial value. It'll naturally be clear that

<wbaker> it does or doesn't after a time. And being able to catch up on the written minutes is what I do for "all those meetings". Self medicate and read the minutes later.

kleber: If this group is intended to focus on concrete and specific engineering solutions to problems we already understand, the async model interspersed with longer meetings is better.

<blassey> yes, async on github and queuing up deep technical discussions for the longer calls

<ekr> aram: this is an acceptable outcome from my perspective

<martinthomson> longer discussions are going to be more useful given the technical nature of the material; you just don't get far enough into the details to make any progress with short meetings

<kleber> consensus to seek consensus!

AramZS: We can do longer less frequent meetings if everyone wants that.

bmay_: We already have a team. These are familiar faces. I feel like we are a part of the same effort. I think web-adv is a good place to float quick ideas and get responses.

<npdoty> do we already have the detailed shared understanding of the technical goals? the charter and repos seem empty at the moment

<ekr> npdoty: no, we don't.

<ekr> I think we have some understanding of the goals for some pieces of the system.

bmay_: Ben, don't leave web adv too quickly. Your voice is important.

<wseltzer> [+1 to bmay!]

AramZS: Last few things to note from the charter. We are looking to generate some guiding documents.

<mib_guprew> note that there are now about 88 CG members

AramZS: There will be an editor for that, and the construction of the working group.

bmay_: I suggest that if we want a model where web-adv is weekly and this one is longer and less frequent, let's agree on how things move here.

<ekr> I would actually prefer we don't have any formal process there.

AramZS: That group can discuss as can the authors of proposals.

<ekr> There certainly should be no requirement that your document ever appear at web-adv before coming here

AramZS: We also have these technical requirements for privacy.

<ekr> Remember, this is *already* an incubation group

AramZS: I agree there is no required pathway to get here. You can start here if you want. No restrictions.

<Zakim> npdoty, you wanted to comment on user agent only

<blassey> just for clarity, web adv hasn't adopted any technical proposals (though they're discussed there)

npdoty: I'm not sure we need the hard details about the scope in the charter. I have some questions. Guidance for people making proposals. What is this about user-agents and stuff? Are we trying to exclude servers? Do we want browser-only stuff?

<blassey> most proposals are currently either adopted by WICG or PrivacyCG

<kleber> npdoty great question

AramZS: We will focus on W3C charter which is browser. Server stuff only relevant as it relates to proposals.

<ekr> I think the way Aram is talking about this is roughly correct. I think something topologically like TURTLEDOVE (regardless of its technical merit) would be in scope.

<ErikAnderson> From a github process perspective, it would be good to think through at what point you will create a repo for a proposal. The Privacy CG has been willing to provide repos for both adopted Work Items and for proposals which aren't far enough along to be adopted but which are generating more discussion/activity than can easily be handled in a single github issue.

<robin> I plan to consider GARUDA in scope :)

<ekr> I am strongly against the creation of repos for non-adopted proposals. I think that's been hugely confusing in Privacy CG

<AramZS> go ahead wseltzer

<kleber> +1, if GARUDA isn't in scope then I'm not sure this makes sense!

btsavage: to npdoty, blassey and I worked on that part of the charter:
… digital advertising isn't limited to web browsers

<luk-wlodarczyk> lol

<bmay_> +1 meta

btsavage: FB/meta advertising is shown largely in mobile apps, interacting with web browsers
… we should be able to talk about interoperating ads between those context

<ekr> I would think mobile apps are already in scope

btsavage: I want to be able to talk about digital advertising

<robin> strong support for btsavage

<bmay_> +1 ben

AramZS: I was thinking more about excluding server-to-server communications

<charlieharrison> +1 to ben as well!

AramZS: I'm not here to limit the proposals coming in. Let's discuss them as needed.

<alextcone> PARAKEET is probably the edge of the proposals right? ("Edge") pun intended

Sean: Feel free to contact the chairs if you have questions. We can provide guidance.

<npdoty> the current charter language doesn't make me think, yes, mobile apps-related work too, but I agree that seems like a useful scope

<ekr> There is also opportunity to have stuff partly here and partly elsewhere. For instance, some of the measurement stuff partly might depend on privacy preserving measurement stuff at IETF. I know @charlieharrison can talk about htat

martinthomson: Ben captured the key point. I see the remit of this group as users. People who use the internet and the web. So I don't see that as a constraint on servers. It'll depend on the specific proposal how servers are involved.

martinthomson: I suggest we start with that topic.

<mib_guprew> we have reached the top of the hour

blassey: The intention of that language was to bring webviews in scope and their interactions. Interoperability. If anyone has a better way to word that - patch is welcome.

<Zakim> robin, you wanted to offer a suggestion for what "web proposals" should be

<npdoty> martinthomson, yes, that seems like a very valuable discussion. I wanted to make sure the charter language wasn't trying to rule out everything except client-side

robin: In support of "broader than browsers". The web is a set of values - not a list of technologies. That might be a good way of scoping things in and out. What belongs on the web even if it's not tech.

<martinthomson> npdoty: i think that - whatever the charter says - we seem to have good alignment amongst participants


<kleber> we can always change the charter as needed :-)

<npdoty> yes! let's just not have the charter actively confusing people

Minutes manually created (not a transcript), formatted by scribe.perl version 136 (Thu May 27 13:50:24 2021 UTC).


Succeeded: s/I don't see a Zoom link in the administrative repos or on the TPAC event page. am I missing something?//

Succeeded: s/Cvent should generate a zoom link shortly.//

Succeeded: s/wseltzer do you know how far in advance it pops up?//

Succeeded: s/oop spoke too soon, I see it now//

Succeeded: s/WebX/WebEx/

Failed: s/martin/marshall /

Succeeded: s/I/[ I/

Succeeded: s/The web is not limited to a set of values/The web is a set of values/

Maybe present: ekr, mjv, robin, Sean, Wendy