Re: [web-adv] minutes and next meeting agenda 12 January

Hello Wendy, everyone,

We wanted to put on the agenda, either for today or the next call,* our
proposal for a cohort bidding experiment.*

As we are moving towards the end of third party cookie in Chrome, many
proposals have been published in the context of the Privacy Sandbox and
many unknowns remain about their impact on the ad-funded open web
ecosystem. As it has been stated several times by the authors of the
proposals and other people at Google (see TURTLEDOVE introduction
<https://github.com/WICG/turtledove#Introduction> or Google's Chetna
Bindra's interview
<https://www.thedrum.com/news/2020/11/17/amid-id-gridlock-google-indicates-no-hard-deadline-chrome-cookie-changes>),
the replacement solution needs to be "private" enough, while at the same
time preserve the ecosystem. On top of the framework choices, each of them
includes various parameters, whose values must be fixed (though the order
of magnitude is already known for some of them). The minimal size of
cohorts, the level of noise/aggregation in the reporting, etc, are examples
of such parameters, for which different values would vastly impact the end
results of a chosen framework.

In the last year of discussions we have seen a varied array of opinions,
ranging from "it is apocalyptic" to "it's fine, people", to "as a user,
xyz". These points of view derived from internal analysis, to knowledge of
how the web works and/or how it is funded, to gut feeling, and probably
sometimes, from overly positive thinking or ungrounded belief.

I think we can all agree that the best way to ground these opinions in
facts is to start testing the new framework(s) to come. This, however, is
easier said than done and raises many questions about what should be tested
and how. Indeed, this change is so massive and has so many ramifications in
the various operations that testing right away a finalized version (no
final design is available as of today anyway) seems ill-advised. Getting
ready for the test would be extremely costly for participants, the points
of failures would be numerous and the results hard to analyze.

A step-by-step process where each module of this new puzzle is added and
tested each at a time seems to much better fit better our shared need to
understand each element in-depth and the interactions between them, and
takes into account the on-going design process of the Privacy Sandbox.
This would also allow currently discussed proposals to get necessary
feedbacks and change specific design choices if the impact of a specific
part of the machine looks insurmountable.

As it has been at the very core of all proposals until today, we propose to
start by testing the impact of cohorts. In this article, we motivate this
choice and propose a draft test design. We obviously had to find a bird
name for it, so here is TEETAR
<https://github.com/criteo/privacy/tree/main/TEETAR>.

Best
Arnaud

Le lun. 11 janv. 2021 à 20:29, Wendy Seltzer <wseltzer@w3.org> a écrit :

> Hi Web-Adv,
>
> Minutes from our 5 January meeting are available at
> https://www.w3.org/2021/01/05-web-adv-minutes.html
> We discussed the Augury API proposal, leaving Scaup for tomorrow's call.
>
> Next Meeting: Tuesday, 12 January, 11am Eastern (1600 UTC).
>
> Draft agenda:
> agenda+ Agenda-curation, introductions
> agenda+ Scaup API [Gang Wang]
>    - The Scaup API for similar audience, aka lookalike targeting
>    https://github.com/google/ads-privacy/tree/master/proposals/scaup
> agenda+ Roundups/updates request
>    - I've heard requests to get updates on proposals we've previously
> discussed. Anyone have updates to offer or specifics to request?
> agenda+ Dashboard highlights?
>    https://w3c.github.io/web-advertising/dashboard/
> agenda+ AOB
>
>
> Please share additional agenda requests with agenda+, ideally with a
> link to an issue or other background.
>
> Call-in info at:
> https://lists.w3.org/Archives/Member/internal-web-adv/2020Apr/0000.html
> (let me know if you need me to send it to you directly).
>
> irc for real-time queueing and minutes is http://irc.w3.org #web-adv
> Github repo: https://github.com/w3c/web-advertising
>
>
> Thanks,
> --Wendy
> --
> Wendy Seltzer -- wseltzer@w3.org +1.617.715.4883 (office)
> Strategy Lead and Counsel, World Wide Web Consortium (W3C)
> https://wendy.seltzer.org/        +1.617.863.0613 (mobile)
>
>
>
>