Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Agenda Request - Accountability designs for aggregated attribution #162

Closed
martinthomson opened this issue Nov 29, 2023 · 4 comments
Closed
Assignees

Comments

@martinthomson
Copy link
Contributor

Agenda+: Discuss accountability designs for aggregated designs

One aspect that was considered a significant factor in the design of Apple's proposed design for aggregated attribution measurement (what I'm calling AAAAA here), was that it gave people the ability to observe what was going on. After some discussion, it became clear that this wasn't necessarily a feature that would be presented through the user interface of browsers. Still, the ability to observe the system in operation was considered a useful tool in ensuring that the system as a whole was trustworthy. That is, while we might not expect most people to investigate what is going on, the option to do so was perhaps important in choosing a design.

After doing some investigation into the subject, I (with substantial assistance from @benjaminsavage and @bmcase) have reached some conclusions, which are presented in this document. In this document, I look at the accountability options available in a setting where attribution occurs on-device and compare those with IPA, where attribution happens off-device.

The conclusions I reach is that while there are some minor differences, each of the proposals offers a fairly similar level of accountability guarantees. Most of the places where there are shortfalls in accountability correspond to aspects of the design where we consider flexibility and usability for attribution to be very important, so there are very few options to make a substantive improvement in terms of explaining to a regular person what is going on.

The baseline information that we can share is, however, fairly good and it looks like all of the options we are considering have similar sorts of options. We should have ways of presenting people with information on how attribution might affect their privacy.

I'd like some time to discuss these findings and to discuss whether there are any opportunities we might take to improve the transparency of the different designs in operation.

Time

One hour, ideally. We could probably manage in 40 minutes if time is short. This is a fairly substantial topic, so while the document is short, there are lots of details to work through.

Links

The bulk of what we'll be discussing is in this document.

@martinthomson martinthomson added the agenda+ Request to add this issue to the agenda of our next telcon or F2F label Nov 29, 2023
@bmayd
Copy link

bmayd commented Nov 29, 2023

Apple's proposed design for aggregated attribution measurement (what I'm calling AAAAA here)...

Maybe we can refer to it as A5?

@AramZS
Copy link
Contributor

AramZS commented Feb 26, 2024

Added to the upcoming meeting

@csharrison
Copy link
Collaborator

We discussed this in our last meeting and the conclusions were documented in patcg/docs-and-reports#55. Was there something else to discuss?

@npdoty
Copy link

npdoty commented Feb 27, 2024

I've sketched some initial principles regarding accountability here:
npdoty/patcg-docs@b3b64c9

Not sure if Martin or others have more discussion or proposals ready for this agenda item today.

@AramZS AramZS removed the agenda+ Request to add this issue to the agenda of our next telcon or F2F label Feb 28, 2024
@AramZS AramZS closed this as completed Apr 1, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

6 participants