Re: [w3ctag/design-reviews] Review request for TURTLEDOVE (Issue #723)

Hello!  Glad that we've made it to the head of your queue.

> it looks like plenty of privacy properties of Turtledove depend strongly on fenced-frames.
> This is therefore a crucial dependency. However, this is not mentioned in the explainer.

@lknik is correct.  [The original TURTLEDOVE explainer mentioned an “opaque iframe”](https://github.com/WICG/turtledove/blob/main/Original-TURTLEDOVE.md#rendering-the-winning-ad) which refers to the concept that was renamed to Fenced Frames.  **I would recommend reviewing [the FLEDGE explainer](https://github.com/WICG/turtledove/blob/main/FLEDGE.md) instead of the original TURTLEDOVE explainer**, as the FLEDGE explainer is more up to date and contains more concrete details.  The first “F” and last “E” in FLEDGE stand for First Experiment, as such FLEDGE specifies a first venture into ads serving in the TURTLEDOVE family.

> can you give a brief update of where the work stands 

Chrome has been conducting an experiment to test FLEDGE for almost a year ([announcement](https://groups.google.com/a/chromium.org/g/blink-dev/c/0VmMSsDWsFg/m/_0T5qleqCgAJ) & [details](https://github.com/WICG/turtledove/blob/main/Proposed_First_FLEDGE_OT_Details.md)).  It’s currently available for testing [to 6% of Chrome stable channel users](https://groups.google.com/a/chromium.org/g/fledge-api-announce/c/Mi1rMwfhwkg/m/WTaH3iwSAQAJ).  We’ve put a ton of work into FLEDGE over the last couple of years.  [The WICG Turtledove Github repository](https://github.com/WICG/turtledove) and [biweekly conference calls](https://github.com/WICG/turtledove/issues/88) remain extremely active.  We’ve started [drafting a spec](https://wicg.github.io/turtledove/) though it’s only partially complete.  As [privacysandbox.com/timeline](https://privacysandbox.com/open-web/#the-privacy-sandbox-timeline) shows, we’re roughly 4 months away from our planned general availability of FLEDGE.

> where TAG feedback can be most useful?

TAG feedback on all aspects of [the FLEDGE explainer](https://github.com/WICG/turtledove/blob/main/FLEDGE.md) would be useful.

The explainer calls out certain behaviors as _temporary_.  These are design trade-offs that we have chosen in the interest of having an API available in Chrome and adopted by ad techs as soon as possible, since the API is a key piece of our effort to remove third-party cookies.  The most useful feedback would probably be focused on the long-term goal, rather than the interim compromises that we've made.  We remain happy to have discussions about any part of the design, though.

> Can you also let us know how this fits together with Topics API and other Privacy Sandbox? 

FLEDGE and Topics are the two Privacy Sandbox APIs for selecting relevant ads.  FLEDGE is meant to serve [remarketing and custom audience use cases](https://developer.chrome.com/docs/privacy-sandbox/fledge/#what) while Topics is meant to enable [interest-based advertising (IBA)](https://developer.chrome.com/docs/privacy-sandbox/topics/overview/#what-is-the-topics-api) use cases. Topics was also designed as an easy-to-use API that could be adopted by the ad tech industry without major upheaval.  FLEDGE is more ambitious, focusing on enabling on-device ad selection using information not shared with servers, and as a result it's forced to take on substantial new complexity around ad rendering and reporting.

Achieving FLEDGE’s long term user privacy and utility goals oftentimes depends on some other Privacy Sandbox APIs, namely:

- Fenced Frames API to prevent joining the ad and the page showing the ad, and
- Private Aggregation API to facilitate measuring aggregate, cross-site data in a privacy preserving manner.


Previously you had pointed out that there were a number of competing proposals in this area.  Some proposals, like Criteo's "SPARROW" or RTB House's "Product-Level Turtledove", were made in response to the original Turtledove explainer, and have been incorporated into the design of FLEDGE.  To my knowledge Microsoft's PARAKEET was the only other proposal in experimentation.  The FLEDGE and PARAKEET teams ran a joint session at TPAC 2022 capping months of progressive API convergence.  We have not ended up fully aligned quite yet, but PARAKEET has since [paused their meetings for a few months](https://github.com/WICG/privacy-preserving-ads/issues/3#issue-819187901) so I think a TAG review of FLEDGE at this time makes sense as it’s the one proposal in this area under active experimentation and active discussion.

-- 
Reply to this email directly or view it on GitHub:
https://github.com/w3ctag/design-reviews/issues/723#issuecomment-1481081377
You are receiving this because you are subscribed to this thread.

Message ID: <w3ctag/design-reviews/issues/723/1481081377@github.com>