<scribe> scribe: Karen Myers

Wendy: Welcome, folks
... take a moment more for people to join
... As we are assembling
... Don, for your agenda item, do you want to share anything on screen?
... Don, you may be muted...

<dmarti> https://github.com/dmarti/in-browser-auction-publisher-issues

Don: I don't need to share anything on screen
... just link to one document
... just pasted link into irc (above)

Wendy: seeing 50 people present, let's look at the agenda
... start with agenda curation and introduction
... Don Marti has asked to share publisher use cases
... I had termed it threat modeling and heard that was confusing
... so used the terms in the link Don shared
... then if we get to dashboard highlights, we can look over issues collected in other places
... and other issues we have been discussion
... any other business?
... and next week we will have the Neustar PELICAN presentation
... Thanks, Karen for link to the video recording of the W3C NY Metro Chapter event [on MAdTech]

Publisher use cases and issues

Wendy: anyone like to introduce themselves who is new to the call?
... Ok
... Publisher use cases and issues

<wseltzer> https://github.com/dmarti/in-browser-auction-publisher-issues

Wendy: Don, you shared a link
... for in-browser auction publisher issues
... I invite you to introduce us to that

Don Marti: I would like to introduce first of all

scribe: where publishers need to get certain things done
... within the ad placement process
... this is what we would like to plan, the first of two outlines
... the first covering decisions for in-browser auctions and placement in web tech
... and second doc covering reporting of relevant publisher and other stakeholders
... A brief intro where CafeMedia sits in this whole stack
... Paul Bannister has been involved in this group for a while
... Paul will explain CafeMedia's role as management service for the publisher

Paul: we run advertising for 3K independent publishers
... everyone agrees online ad publishing is complex
... we function as agent for those publishers
... we don't create the content
... for adv and adtech we function as the publisher
... we run ad stack, prebid, what any publisher would do
... from POV of running adv on site
... looking at Comscore, we may be the largest publisher
... that's non-platform open-web
... so we have deep insight into what is going on
... we want to write up what are key publisher use cases
... not currently handled and put them out there
... as we continue to refine proposals
... look at publisher perspectives as a key part of that conversation

Don: thank you, Paul
... Goal is to introduce the outline for publisher requirements
... and try to get links to relevant issues in all the many related proposals that could be required to implement this
... I recently started at CafeMedia
... one of things I noticed first is CafeMedia does have that role from within the publisher side of the operations
... CafeMedia has team on WordPress, maintains ads.txt
... for small site that cannot have its own ad ops team

<jrosewell> s/ads.txt

Don: CafeMedia is the outsourced opps team

<wseltzer> https://github.com/dmarti/in-browser-auction-publisher-issues#publisher-use-cases

Don: let's jump into publisher use cases; and file issue on here or contact me
... where we are missing something
... First one is pretty basic
... A publisher needs to decide which ads appear on their site
... there are definitely ads that are completely appropriate for a music review site
... that would not work on a recipe site
... There are ads that could be taken out of context and do well in one place, not another
... may want to say, don't want this piece of creative, or not have this advertiser, or not run this vertical
... on pages within a site
... The second one
... is besides just human viewable ads is the question of malware and malvertising
... case where advertisers generally use third party crawlers
... when discovered, publisher wants to prevent that from showing up
... relevant to in-browser ad discussions
... conversations on web bundle in advance and use in browser session if interest based session associated with ad ends up winning
... if ad has malware, it needs to be detected and not shown to user
... and publisher needs to be confident they are not showing malware
... Final case is when advertisers choose to pause all their advertising across all media with short notice
... this happens with an unfortunate news story
... poisoning of food, plane crash, whenever advertising is not appropriate
... you can pause ads, but hard for publisher to make case to have their site be in competition with sites where ads can be bought
... Second one is publishers need to fulfill directly sold ad placements
... not as simple as, if we do direct sold, we don't do RTB
... very often publisher will want to see RTB bids if they are coming in as more attractive
... if they come in as more attractive than direct sold
... but they need to tweak rate at which direct sold ads serve so they come through with required number of impressions
... More verbiage in the document
... I can pull together a session to dig into this in more detail
... there is alot more material on dynamic allocation and pacing
... happy to dive into that
... Third one
... is known as flooring, setting a price for page section or price
... also need to do flooring
... on a vertical
... in a publisher site they may say this is the floor price for automotive category
... or this is floor price for an RTB placement that might compete with one of our direct sales channels
... A lot of give and take looking for RTB revenue v. direct sales and fulfilling those deals
... Number four, is pretty general one across all kinds of ad supported media: print, TV, etc.
... you don't put competitive ads next to each other
... figure out when this occurs
... if there is contextual ad for Toyota and IG ad for Honda
... this is standard across ad supported media
... a walled garden site is in a position to guarantee
... a web publisher needs to guarantee it as well
... and publisher needs to say which brands are competitors for a vertical
... Amazon could competitor for book store
... but for cloud service, hard to apply that
... up to publishers on what they want to present
... Number five is complex one
... result of case by case adjustment being applied by publisher to the auction
... errors in serving ad requiring price adjustment
... post brand safety blocking; bid adjustments for not getting paid
... and discrepancies...
... interested to dig into more details on this one as well
... Finally, complying with regulatory requirements
... in US no alcohol
... or things related to children
... publisher understands local regs that apply to them
... future ad systems needs to allow publisher to implement the regulations
... and show to potential advertisers that they do comply with the regulations
... I am happy to take issues on this Github repository

<Zakim> wseltzer, you wanted to refer to https://github.com/w3c/web-advertising/blob/master/support_for_advertising_use_cases.md#publisher-needs

Don: or set up a session to discuss further

Wendy: Thanks, Don
... that is a nice overview with substantial detail
... wanted to point back to the use cases table we were developing earlier
... publishers needs have a few lines, but you add a lot more detail and considerations
... that we might consider and at least link from
... do we have existing proposals or issues that reflect these needs
... let's build out the list of needs and think through how we meet them

Don: would it be helpful to migrate the whole publishers use cases doc into this whole section
... or should I add links to cases within that table

wendell: I will jump queue to say I love the single doc format
... if there is no sprawl
... it has been super useful in industry conversations to say what will be allowed
... and to not have conversations about separate sites and documents missing

Wendy: there is some mechanical question on what makes sense for the development
... hearing one voice for it's useful to get all of info into one place, if that is possible
... now go to Michael

Kleber: thank you
... Thank you very much, Don and Paul for bringing this list
... it is a great list of publisher use cases
... I have been badgering Paul to bring these forward for a while
... you did a nice job of expressing them very well
... I think majority of these we have addressed previously and are being addressed in the "bird" proposals
... makes me feel good we have already discussed a lot on how to meet
... the one that sticks out to me as one we have not talked a lot about before
... that may prove problematic from privacy POV
... is number four about competitive exclusions
... if you have one ad being served through contextual, and another served through a TD or other IG type mechanism
... that tries to made ad isolated from surrounding page
... way we have talked so far
... if contextual ad is chosen first,
... when chosen, quite easy to send contextual signals into the auction that picks the IG based ad
... that says ads from this company not allowed on this page
... but if it happens in the reverse order, like Coke ad inside a fenced frame
... and not supposed to know who ad is
... and not have Pepsi on that page
... seems tricky to figure out
... that's the one that makes me the most nervous about this wish list
... otherwise, rest of list feels comfortable

Paul: thanks, Michael
... we know some of these were talked about before, but wanted to put them into a single place
... I like what Don and Wendell said about getting these docs together
... competitive situation is good point
... a lot of complicated situation
... can you have a Coke and Pepsi ad on a very long page
... and how and when a given ad slot renders on page matters

Don: there are other use cases where the
... publisher or site serving top level page
... needs some info on what is inside the iFrame
... the brand safety one
... if you don't know when you are running a Pepsi ad, then how you also
... can show that you are meeting your brand safety commitments

Kleber: absolutely; we have talked about that before
... the Criteo proposal addressed that
... to what extent....allowing for brand safety commitments
... that's a topic we know we must support

<Zakim> jrosewell, you wanted to congratulate Don, Paul and CafeMedia for talking on behalf of their customers

James: quickly thank Don, Paul and CafeMedia explaining on behalf of their customers
... important to hear from smaller customers
... it's a helpful step forward, thank you

Wendell: I would like to speak a bit about some of the parallels we are seeing in the publisher side and the media side
... and previous discussions on the demand side when we were running right media
... repeating same history here
... it's all good when relevant sides come together to express their interests
... we found it is not poss to unify all these concerns in a single role or mechanism
... the way market and mechansim design work mean that cannot happen
... these are adversarial relationships; demand, supply, and incentives among service providers
... we will get to a point that finding a way that is neutral enough
... so that supply side and demand side can live with it
... media will be operated by one of two companies
... I like that there is documented set of info for demand and supply side
... demand may fail...no way to codify in computer science
... remediate at level not usual with print media analogy
... I would be concerned that we are trying to produce extreme continuity with way industry works today
... don't know that is entirely possible
... need to figure out limit for continuity
... safe harbor API was good start
... and now creeping in of business use cases around the edge
... may get us to a more complicated solution
... would like to see us get to implementable solutions in 2021

<scribe> ...done on simple media and we can build forward from there

UNKNOWN_SPEAKER: after 30 years on the web
... start simply is a great place to be

Wendy: Don, do you want to respond?

Don: yes, I understand the desire to get something simple up and working
... problem with some approaches to simplification is that they would squeeze out some players in the ad ecosystem
... once you have a system that is consolidated, it is harder to re-diversify
... if there is going to be simplification, it needs to be balanced out in interest of small sites
... with interests of particular platforms or processes

Hong: thank you, Paul and Don, thank you for the excellent presentation
... and to Michael for explaining coverage of use cases
... if there is intention to simplify or some sort of gap
... and functionality will not be addressed
... some kind of plan during these trials
... if released, in coming Chrome browser
... if there is going to be simplification, or some sort of gap, would help publishers to understand impact on current operations

Wendy: hearing interest in testing
... an opportunities to experiment; thank you
... Michael, do you want to respond?

Kleber: In Chrome we are very sensitive to idea that we need something to start experimenting with and use quickly
... instead of waiting until next year when we have to rely on things
... Wendell expressed why the reasons well

<dinesh> +1 to experiment and test early

Kleber: and desire to get something out quickly and add things as we go; that resonates with me
... that is the direction we want to push in Chrome; make something available in 2021
... even though it won't have all the features we want for utility of ecosystem, and won't have all privacy information for flow of into
... we don't yet have a plan for 2021 but I am working onit
... we want to present the simplest thing that could possibly work

bmay: regarding conversation not allowing two competitors' ads to show on same page, and not showing what is in a fenced frame
... can there be a taxonomy to not show who advertiser was
... and allow publisher to now dynamics on page without knowing specifics about it

Kleber: sort of things we need to think about
... when you open up an info channel with FF about result of auction
... you run the risk of some clever engineering opening up a way to track people
... info leak

<Zakim> wseltzer, you wanted to discuss "threat modeling"

Kleber: getting to detail for privacy preserving might be hard

Wendy: I queued up
... to say this was the kind of discussion of details I had in mind when I put "threat modelling" into the agenda modification
... thinking through the second level
... and what happens when someone uses the interface, mechanism design
... how do parties with varying intersets use designs like this
... how do things we are designing be used and abused

<alextcone> +q

Wendy: opps to experiment in the wild

James: quick comment
... from earlier conversation comment
... time is important, but so is a level playing field
... better to take the time not to rush things
... we need to be careful about the rolling mandate and balance those things

<Jordan> IAB Tech Lab offers the "ad product taxonomy" (see https://iabtechlab.com/blog/tech-lab-adds-new-ad-product-taxonomy-to-portfolio/), a standardized nomenclature for describing the product or service being advertised within a creative unit.

Wendell: quick answer to the media control question
... companion ads or exclusion
... parable is media operator creates media and offers it for sale
... if that [missed]
... that will be a problem for the publishers
... lots of extramural stuff, outside of computer science
... obvious ones are no guns, no drugs
... but other things media opps care about
... sitting in post-its on wall of editorial staff
... will be part of mechanism design, incentive program that supply side has radically different concerns that demand side
... need to find neutral ground
... both sides may be a little unhappy, not just privacy concerns, but other business concerns

Paul: that is a good point
... there will be things that will make people unhappy
... so maybe that is right place to be if everyone is a little unhappy
... talking with publishers, if these use cases are not supported

<gendler> Jordan, the IAB taxonomy has some proven issue with it, and I think while we're discussing privacy problems with taxonomies, the IAB Tech Lab solution, to me, is one to use as a "how not to do" https://techcrunch.com/2019/01/27/google-and-iab-ad-category-lists-show-massive-leakage-of-highly-intimate-data-gdpr-complaint-claims/

Paul: it is important to understand the must haves, nice to haves, etc.; we need to dig into this ourselves

Don: to your point, Wendy, about threat modeling
... point to get clarity on
... which is whether the definition of privacy includes concealing the ad that ran on a site
... from the owner of the site
... thinking back to my own experience of my own experience from Linux Journal
... and an engaged tech audience and their concerns about certain advertisers
... it's a labor of love for editor of site content
... and important for them to see which ads they are running
... we need to define privacy for next gen ad standards
... does that mean privacy for advertiser not visible to editor of site they are appearing on

Wendy: I will say that sounds like a distinct issue
... from the user privacy
... on the user privacy side, I hear us trying to think about
... how we protect the user from unwanted info leakage
... including if you know all the ads chosen by various means to show to given user, would you learn things about that user
... that user considers sensitive

Kleber: respond to that
... goal is not to keep the publisher from knowing what advertisers are appearing on that site
... that is fine; good info flow
... aggregate info flow, across all the visitors

<Jordan> Gendler, that article was from nearly 2 years ago. We've updated best practices and the taxonomies since then.

Kleber: what we want to avoid is what advertiser is using a mechanism like TD to target a particular user
... and not on a user by user basis cross site info
... that is the crucial privacy preservation we are trying to do

Don: We need to enable seeing every ad on site, but we don't share ad/user pairs back to the publisher
... that seems reasonable

Kleber: yes, that is exactly right

Alex: hi
... I just wanted to note based on Wendell's comments about moving quickly
... I get that as a product person
... role of privacy taking on role of lots of current product in industry
... has to take on good a job or better as current products out there
... has to do just as good or better if you are going to change
... what Don has laid out is real
... dismissing them at all as 'hard to achieve' is fair
... but if you are demonstrating with privacy sandbox
... and take on role of a lot of product in industry
... you have to take on role of many constituents
... by taking on privacy sandbox role, you will have more constituents than any adtech product than Google ad stack
... not get to that later

<jrosewell> +1 alex cone

Alex: if I were your leadership team, I would say you cannot have that attitude

<joshua_koran> @Alex - your question raises whether Privacy Sandbox is meant to replace open marketplace solutions by giving marketer and publisher choices OR removing their choices by substituting its own solution

Brad: want to thank Don for this issue list
... and most of concerns could be solved outside of a standards body if we agree the auction could be controlled by a publisher

<jrosewell> These changes alter a market if not done properly. Changing markets is not the role of a technical standards body.

Brad: we tried to address this with our PARRROT proposal
... outline in a single standard may be challenging to tackle all these issues

Aram: I was going to add
... the PARRROT proposal seems to hit some of these problems headon
... would be great to merge this into the existing use case document
... we have use cases, more detailed proposals
... maybe split into smaller document with headers
... say PARRROT addresses this, TD does that

<robin> +1 to AramZS

Aram: poss tech proposals that we want to spend more time on

Wendy: great, that is a key feature of the use cases document, identifying what we addressed
... what we are trying to address and where there are gaps

Brad: reply to Alex's comment
... might have been a misframing of what Kleber said
... goal is to get MVP versions of APIs out for testing while we still have 3rd party cookies around
... so we can build out around
... not something that is permanent

<dinesh> +1 to AramZS to map tech proposals to use cases as there are several proposals and several use cases

Garrett: just a more general comment
... what makes a virtual F2F was having a slide deck to refer to

<brodriguez> +100000

Garrett: having some text slides would be more effective to crystalize what people are saying
... I would propose that this be norm
... and let conversation drift back

Wendy: I will take that on as chair on how to make our discussions the most effective

<gendler> Jordan, that's a fair point. I have not seen many updates that gave me hope on the subject, but I see that you released a new updated in October. I'll hold further thoughts until I take a read on your latest update.

Alex: responding to Brad
... totally, MVPs are great
... when it's net new funcitonality and whether market participants can choose to participate
... but in this case, don't have that choice
... privacy sandbox is inserting itself into product roles in the industry
... so it has to be as good
... and worth it to step into this role
... in stepping into this role

<robertblanck> +1 to Aram

Alex: you are going to be stepping into a function that has a lot of competing products today
... so you need to focus time and energy of one team on a product stack
... if you want product to be good
... I understand Kleber is talking about MVP
... most not looking at 90+ numbers
... something that gets out in a mix where people have to adopt

<kleber> qq+

Alex: that this is just a classic MVP, you are stepping into the happy path

<phileligio_epc> +1 to AramZS, alextcone

Alex: you have to do ads good or better with the new product

<Zakim> kleber, you wanted to react to alextcone

Kleber: let me jump in
... the MVP thing I just mentioned
... this is Chrome wants to ship something as a way for people to start trying out these ideas while 3rd party cookies are still around
... so people can see what system will look like while cookies are still around
... it won't have all the properties, more like what Wendell said, that it can address the main use cases and we will learn what needs to be added

Aram: two thing

<alextcone> The willful ignoring of the point that in 2021-2 that you have to adopt a product is quite similar to the US president's last 4 years of press secretaries.

<gendler> +1 to Aram on slides

Aram: I don't think we need slides as long as they have corresponding markdown files
... worry about timings
... write overly complex text files

<robin> +1 to AramZS, standards are too complex for slides, they need writing

Aram: as much as I love slide shows...just something to keep in mind
... I would rather have these MVPs while there are 3rd party cookies available
... it gives us strong arguments on where we need this tech to go and how to sustain our businesses
... illusionary if we think adtech business will operate the same way as it is now
... don't think most users want it to be the same
... we are here because the current adtech flow does not satisfly our users, the people interacting with our sites on the web
... there are far more of them [users] than adtech companies
... I think there needs to be some pull back from that assumption
... we are not building the same exact thing that already exists

<gendler> +1 to Aram on assumption evaluation

Aram: most of don't want to do that
... just keep in mind as we go through those proposals

James: so much to say
... we have expanded out to talk about things that talk about market conditions, which is not for standards tech conversation
... [missed]
... going beyond incubation
... consider at which point these things move into W3C WGs and we have the proper processes
... so we have actual standards and not de facto standards
... a lot to consider around messaging; what is a draft, an MVP

<gendler> -1 to James, market dynamics are part of the world of tech, standards affect those dynamics, so it is important to concern ourselves with the nature of market dynamics when designing standards.

James: not things to solve now, but things we need to be very conscious of

<AramZS> If it helps, there are some *great* systems for generating revealjs slides from markdown. If we had a formal markdown format in terms of document structure we could prob automate that slide generation from headers or something like that

Wendy: everything we are discussing at this point is drafts and incubation; nothing is formal WG yet

bmay: whether we have slides or whatever we have, I would like to request that we get material ahead of the meeting

Wendy: let's try to do that, yes
... review material and agendas out in advance

<kleber> Note that Don sent the link last night (https://github.com/dmarti/in-browser-auction-publisher-issues)

Wendy: put links in irc as we go through explainer
... I hear that those request

<kleber> (web-advertising issue #97)

Wendy: I will take it to the list to continue requesting agenda items and see if we can them queued up for discussion [sooner]
... this is not the last discussion on publisher use cases
... we need to adjourn here
... see you next week
... thanks very much

<wseltzer> [adjourned]

This is scribe.perl Revision of Date 
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: Irssi_ISO8601_Log_Text_Format (score 1.00)

Succeeded: s/non open-web/non-platform open-web/
Succeeded: s/adtech DXT/ads.txt/
WARNING: Bad s/// command: s/ads.txt
Succeeded: s/if we/not as simple as, if we/
Succeeded: s/@/allocation/
Succeeded: s/@/wendell/
Succeeded: s/how/how and when/
Succeeded: s/@ are servers/incentives among service/
Succeeded: s/unconcerned/concerned/
Succeeded: s/and [missed]/you run the risk of some clever engineering opening up a way to track people/
Succeeded: s/thread/threat/
Succeeded: s/@/role of lots of current product/
Succeeded: s/APIs/MVP versions of APIs/
Succeeded: s/@:/bmay:/
Present: aschlosser_ gendler Karen bmay blassey lbasdevant arnaud_blanchard ionel dialtone xaxisx jrosewell kleber AramZS hong pbannist seanbedford br-rtbhouse mlerra dinesh pl_mrcy bleparmentier wbaker apascoe dkwestbr tomkershaw
No ScribeNick specified.  Guessing ScribeNick: Karen
Found Scribe: Karen Myers
Agenda: https://lists.w3.org/Archives/Public/public-web-adv/2020Nov/0012.html

WARNING: No date found!  Assuming today.  (Hint: Specify
the W3C IRC log URL, and the date will be determined from that.)
Or specify the date like this:
<dbooth> Date: 12 Sep 2002

People with action items: 

WARNING: IRC log location not specified!  (You can ignore this 
warning if you do not want the generated minutes to contain 
a link to the original IRC log.)