– DRAFT –
Improving Web Advertising BG

Wendy: We'll take a couple more minutes to allow folks to join us
… We especially welcome those today who are not usually able to join us; hope this is a more Asia-friendly time
… We have two meetings scheduled
… Start by reviewing the agenda and see where we have items to discuss today and which to discuss tomorrow
… Both of meetings will be minuted by Karen
… and thanks to everyone else who helps out in irc to offer additions and corrections
… People are welcome to attend both meetings, but not required to
… I think we are at a good time to start with our agenda curation

<wseltzer> https://github.com/w3c/web-advertising/issues/121#issuecomment-951731864

Wendy: I have listed in GitHub some possible agenda items
… Let's start with introductions

[Reads agenda]
… We have some people who have offered to present Thursday
… reporting on content taxonomy and privacy signals
… which leaves us plenty of materials to discuss this evening US/morning or other in your time zone
… We often start these calls with the introduction of new participants
… If you have not attended a Web Advertising call before and would like to introduce yourself, please feel free to speak up
… I see a hand in Zoom

Introductions and Agenda Curation

jyasskin: Jeffrey Yasskin, Google, introduces

Wendy: anyone else want to introduce themselves and your role?
… and also invite you to "present+" in irc
… those of us who have been on many of these Web Adv Business Group calls
… know we have been meeting weekly to discuss a number of use cases
… starting from a document that Ben Savage helped us to put together
… and many people have contributed use cases and descriptions
… and a table of the proposals that fit with those use cases
… to help us think about these work

<wseltzer> https://github.com/w3c/web-advertising/blob/main/support_for_advertising_use_cases.md

Wendy: linked here
… and we have a long list of proposals we have been discussing, many of them avian0themed
… bird proposals to take up some feature of the web advertising landscape
… and propose both privacy preserving and help to support monetization on the open web

<wseltzer> https://w3c.github.io/web-advertising/dashboard/

Wendy: We add to growing list, and send off to be worked on in incubation
… often in a community group or another repository
… added the link in irc
… includes a growing list of issues collected on these proposals
… and periodically we come back to see what is going on or updates from the incubation that anyone wants to share with the group
… That is one possible subject
… we also have the question of preparing for the standards track and what is ready to move forward
… I'm waiting for someone to join the queue to see if there is something you would like to share
… many of the things we queued up specifically are for tomorrow
… which leaves opportunity to ask questions, make proposals
… and I see Nick Doty

<Zakim> npdoty, you wanted to comment on dashboard

Nick: I'm looking at the dashboard
… seeing a bunch of variables and no content

Wendy: you're right
… it looks like the dashboard is not working right now; I'll ping Philippe

<wseltzer> https://github.com/w3c/web-advertising

Wendy: you can see things included on dashboard from Github repo

Wendy: It's TPAC; how's it going

Tess: Long and never ending...

Wendy: yes, it's long; it ends the end of this week

Brad: But there is a one-week break before IETF

Sam: so many meetings at the "golden hours"
… almost as bad as meeting in person

Wendy: I would especially like to hear from folks who are not able to join our usual calls
… Goal is to help expand the group and reach people around the world
… see if you have questions answered, or issues to put on the table

Wendell: Thanks, WEndy

Wendell: Last year at this time
… we had a fairly dense presentation
… we had some statistical presentations on the value of audience-based
… advocates for budgeting and what goes on
… I spoke on Verizon Media perspective
… Essentially, one of the punch lines at end of talk
… was A) get on with it
… we have been talking about this for a while, harkening back to Google-hosted meetings in Boston and San Diego
… and second aspect, if we don't solve this in this forum, we will go to other forums and folks will move there
… not just media folks but also young audiences
… We have gotten on with it
… there has been some movement, but we have hit a lull
… some of the proposals just don't work
… they are not workable politically, legally and technically
… we are as users of this web tech at Yahoo
… looking for the next tranche, next wave
… not just essays on Github but actual working code
… Michael Kleber and PARAKEET crew have put forward question
… is it necessary for a trusted server
… and what affordances should it be vested with
… proposal from Robin Berjon, NYT
… and IAB has an elaborate proposal for choice access and internal back plane for trading
… being cognizant of legal regime and the economics to make this be stable
… we have an opportunity to have a free-wheeling discussion
… there were also proposals about embellishments or refinements to web
… don't know if James Rosewell is here now
… might be good to have some reflection on those proposals as we also consider safe harbor API
… and see if folks want to respond to that or something else

<Zakim> kleber, you wanted to reply to Wendell

Kleber: I think Jeff was ahead of me
… either order is fine

Wendy: Jeff deferred to you

Kleber: thank you, Wendell for the summary
… examining my life, I don't think there has been a lull; tremendous amount of productive activity
… in the proposals discussed
… in particular I feel we are moving towards working code and live experimentation in Chrome
… for a bunch of the APIs we have talked about
… FLoC and conversion measurement are in origin trial
… have been experimented with by a variety of players
… and published findings
… and we are iterating and addressing the short comings
… by people who took up our offer
… we acknowledge it's a lot of work to try out an API in this young form
… and understand some of work will need inevitably to be thrown away and redone
… so thank you to those who have helped us
… Specifically to APIs that is our FLEDGE proposal
… the original TURTLEDOVE, with modifications from many proposals discussed in this forum
… and MS's variant called PARAKEET
… introduce some kind of trusted server into how Advertising works

<wseltzer> https://darobin.github.io/garuda/

Kleber: Robin Berjon's Garuda is an interesting idea, and like to see more discussion on it when Robin is around
… trying to think what else I have not commented on that you said
… Surprised to hear you said that a bunch of proposals technically did not work
… because we have seen some success in those we are pursuing
… those that are not politically viable
… or cannot be adopted more broadly
… now is time to plug the new Community Group
… that was just chartered and meets this Friday, a day and a half from now

<wseltzer> https://www.w3.org/community/patcg/

Kleber: It already has 76 participants signed up
… So that is a lot of evidence that there is substantial interest, including cross-browser interest
… and taking the castle-in-the-air proposals and grounding them with goal to achieve multi-browser support

Wendy: in mean time, the dashboard is back up with help from our systeam

<npdoty> thanks, and I'm enjoying catching up on the very long list of issues there

Jeff: I had originally queued up just to rescue Wendy
… which is why I was happy to defer to Michael
… but I was going to ask a counter point to point Wendell was making

<blassey> Friday's PATCG meeting: https://web-eur.cvent.com/hub/events/2b77fe3d-2536-467d-b71b-969b2e6419b5/sessions/05b7fd0e-1cba-4e38-b908-90609b8c3802

Jeff: We have been in business now for a few years
… there were some very broad proposals like TURTLEDOVE
… in the process of being tweaked
… and more specific things like measurements
… curious, at end of the day
… what really counts at W3C is what comes out of the rec track
… Do people have a perspective on what will be ready for the rec track six or twelve months from now
… may be good to discuss that with this medium-sized group
… question to anyone who has an opinion

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

Blassey: the PAT CG
… hopefully has things to incubate to point when they are ready
… see on horizon that some proposals could move into rec track WG
… not quite there, but have those conversations in PAT CG

Wendy: I think this is a good time to talk about the landscape of groups that we now have
… because with that CG and Privacy CG
… and Web Platform Incubator CG
… we have forms with contributor license to develop drafts to a greater extent
… Business group is good place for presentation of proposals
… with pointers to CG where work is happening
… where we have a specific charter like those in PAT CG
… focused on particular slices of technology
… and I appreciate that the PAT CG has chartered to try to develop a proposal for a WG charter
… then in the W3C process, a proposed working group charter would go to the entire W3C membership for review
… and to horizontal review for accessibility, security, i18n, privacy
… and the TAG
… we take a deep look at it as we charter it
… and if membership supports a working group dedicated to advancing one or more specification on to recommendation
… Key questions for that group to help us address
… should we start with a narrowly focused WG on a single spec, or take a range of specs and put them in one place
… I hope that by working to develop ideas to the next degree of specification

<Zakim> npdoty, you wanted to comment on ad topics and what happens in which group

Wendy: that that CG will be able to help us in the Team and community to answer those questions

Nick: To follow up on that summary
… I think I know about those groups, and the context
… but I'm having trouble figuring out what is happening in each group and why
… curious about something that this group likes to discuss
… a case where ad topic
… of user controls

<wseltzer> https://github.com/privacycg/proposals/issues/26

Nick: to activate more interest or less in ads
… as opposed to users implicitly consent to something
… that seems like a promising approach
… not sure why we would not call it improving web advertising

BrianMay: I'm becoming increasingly interested in auditing what's going on in our systems
… specifically auditing around advertising
… how do we know what our systems are doing
… how do we keep an eye on what participants are keeping an eye on it
… I think it's worth exploring
… given all the changes and information available to people generally

Wendy: while people think about that question
… to Nick's question about topic hints
… I think Ben Savage raised that for a discussion in this group

Heinz: hello everyone
… question about audits
… there is a group in IAB Tech Lab looking at audits
… and their members are reviewing
… and how to sell audits
… in the advertising ecosystem
… they have specs out tomorrow morning
… Alex from Tech Lab will be attending that meeting and can give more details

Wendy: Thanks

<npdoty> and just for agenda planning, is the contextual advertising update scheduled for later in the week?

<Joshua_Koran> Prebid is also working on a similar proposal on ensuring people, publishers and marketers have transparent audits of advertising

Wendy: a good case for our working liaisons with IAB Tech Lab

Brad: On topic of other audit-related things
… the @ proposal has IP blindness
… that discusses IP address use under audit
… one update
… this week around broad principals that would be a good conversation on how those audits work, how to turn those broad principles into a policy

Wendy: Thanks

<blassey> https://github.com/bslassey/ip-blindness/

Wendy: Question mentioned on what are we doing in technology
… and in trusted third parties
… in either of those cases, how do we know they are behaving as anticipated
… and meeting needs of users in those systems
… whether it's audits, or expert review
… or reliance on a third party to validate
… what's happening
… all part of the 'how do we govern these systems' and get assurance that they are doing what we expect

Brian: also add to list of things mentioned
… error identification and resolution
… and services data to incorporate into things that would allow for that kind of effort

Wendy: I'm hoping that if these aren't yet captured in our use cases that we will work to add them
… to make sure we are thinking both of the first level of how proposal meets the direct goal
… and audit, verification seems like the second level
… to make sure it is operating as expected
… while I wait for others to queue up, I'll remind people that we have the GitHub repository
… the opportunity to add new issues

<wseltzer> https://github.com/w3c/web-advertising/issues/104

Wendy: and some of those issues are hoping for info-sharing
… link to collection of related meetings
… of other groups that are discussing some of these problems
… and find those collected here
… and it's a great place to put these questions or make additions to the documents we've been gathering
… I won't take the quiet time as assurance that things are crystal clear...

Joshua: Thanks
… We've talked a lot about various proposals
… and it's often difficult to compare them to one another
… as they optimize for different benefits
… trusted server models are real-time
… others optimize for less data to be sent
… wondering if there is some sort of metrics as we progress with tests and trials
… to compare the objective outputs of this advertising system we are building
… there are users, using software to engage with publishers, being confronted by advertisers
… if we don't have metrics for at least two of those groups
… is it worthwhile if not today

<npdoty> blassey, from a quick glance, this seems like a pretty broad list of uses for IP addresses under IP blindness

Joshua: at an upcoming meeting to talk about metrics
… what do we want to measure?

BrianMay: When I'm thinking about questions Josh has asked, I look at use cases and look at what those proposals suggest
… and what changes over time as they are implemented and developed

Wendy: Is that the kind of metric you were thinking about?

Joshua: It does, but thinking about....contextual
… a true statement but doesn't help with understanding if we are doing a good job from users, publishers, marketers' perspective and their ROI
… if we look at Google ad team's review of FLoC, they said there was a metric for number of orders per dollars spent
… buy side would be yield or @
… important to have a list of these metrics
… so as various companies test they can share metrics and we can improve them

<Zakim> jeff, you wanted to comment on publishers, users, and advertisers

Jeff: I think Joshua has a very interesting idea

<blassey> npdoty: you may also be interested in https://github.com/ShivanKaul/draft-ip-address-privacy

Jeff: I wouldn't want to have a tech team guess at what those metrics would be
… but would be good to have outreach to who we think the stakeholders are [names]
… and poll them on what they would like to see
… and I'm sure we'll get some metrics
… that you may double one person's satisfaction with one algorithm and cut anothers
… would give us a quantitative reasoning about it
… so if people ask, 'why you cut people's privacy by ten percent' then publishers say it improves by 400 percent
… I would like to see everyone's input on that question

Wendy: one of the great aspects of our forum is we can form task forces to try to collect that information

Angelina: Sorry to join late
… I just caught the tail end of that conversation
… if there is anything you would like to request from the OS task force, I could ask a list of questions
… we have over 100 people

Wendy: Thank you, Angelina
… we are thinking about crowd sourcing some metrics of interest
… before we go to you and your members with a list of questions
… we can also call upon the group here to generate some questions and places where we would like to see questions asked

Brian: Suggest to Angelina, that group testing new tech
… is also a good place to test metrics they can use
… and gauge what metrics to use for those tests

Angelina: idea is to figure out some KPIs they are looking and connect to those platforms with solutions
… just email me at angelina@iabtechlab.com
… and I'll work with you guys on it

ErikT: I was going to suggest a bit of a larger point
… how do we try and welcome more people into this group and discussion
… there is a lot of browser folks
… and some of the larger publishers

<npdoty> I'm a little concerned we might, in talking about metrics, be assuming that the status quo is an acceptable baseline or likely to continue

ErikT: but smaller publishers and smaller advertisers are less represented

<npdoty> or if not, what alternative baseline would we consider?

ErikT: bringing them into this conversation and bridging that gap would be helpful

Wendy: thank you; some of things we are thinking about is sharing the documents, inviting input on that
… thinking about webinars or presentations to help get the message of what we are working on out to a larger group
… and to meet people where they are
… certainly if you or anyone here has suggestions for people we should talk to; channels we should reach out to
… and good ways to get the message out further, we would very much like to hear that

Heinz: I think there is definitely
… one thing for this group, which is tech oriented
… some things are not feasible for an end user to do
… brilliant ideas
… but NAI and IAB could help drive the use cases that goes across all the players
… good to get more input from these other industry groups to get more input
… and how to solve the problems for the parties in the advertising industry ecosystem

<wseltzer> https://github.com/w3c/web-advertising/blob/main/support_for_advertising_use_cases.md

Wendy: That comes back to one of the proposed topics for our discussion
… is a documentation sprint

Joshua: respond to Erik's point for more feedback
… I know PRAM did spend a lot of time
… talking directly to CMOs
… to gather use cases, and also some publishers and some adtech cos
… and I think Angelina put link into the document they produced
… I do think they mentioned some of the metrics there that we could capture and bring into our docs
… we don't have to reinvent

<npdoty> blassey, this is a useful doc on different privacy impacts. I just wasn't sure if IP blindness meant throwing away the IP address, or keeping the IP addresses, but only using them for a list of purposes

Wendy: thanks for the reminder that we did hear from PRAM
… and heard them present their use cases
… hope to get those captured in links
… and not duplicate material
… we are at the end of our hour
… I hope those joining for first time, have had an interesting taste and will follow more regularly
… I thank everyone for joining
… we will share these minutes and we will meet again tomorrow
… also through the TPAC event calendar
… and the PAT CG will have its first meeting on Friday
… a breather, then IETF...
… thanks all

[adjourned]

<blassey> npdoty: at a very high level, the gnatcatcher proposal is to proxy traffic to servers that are not Willfully IP Blind (i.e. that don't commit to only using IP under a policy that restricts its use to these purposes)