Re: [admin] Consensus Call: Adopting Privacy-Preserving Attribution into PATCG (#26)

@dmarti I'm still struggling with taking those principles and turning them into something concrete.  Or at least, any more concrete than we have done so far.

The proposal explicitly has a weakness when it comes to faked impressions, which might enable more conversion stealing than is ideal.  That's an acknowledged shortcoming, but one that we have signaled an intent to address.

We have concrete proposals in the form of a "late binding" mode, in which impressions are collected separate from conversions.  This turns out to be considerably more expensive to operate, to the point that we aren't confident that all browsers will be able to offer an aggregation service that works for both advertisers and browsers.

Meta and Mozilla have investigated the use of three-party MPC and shown that a system based on that is likely feasible, despite the increase in volume that results.  A TEE can probably made do no matter what we do.  However, we haven't worked out the details to the point that we're confident in specifying that just yet.

If you have something more concrete to contribute, such as alternative approaches or tweaks to the existing one, that would be wonderful.  Is there something else?

-- 
GitHub Notification of comment by martinthomson
Please view or discuss this issue at https://github.com/patcg/admin/issues/26#issuecomment-2398789158 using your GitHub account


-- 
Sent via github-notify-ml as configured in https://github.com/w3c/github-notify-ml-config

Received on Tuesday, 8 October 2024 04:13:41 UTC