Charlie Harrison

Thanks @AlexandreGilotte ! Yes I agree the ML use-case is a clear case where it seems useful to contribute to many aggregate keys at once. I am happy to consider...

We discussed this in the call yesterday ([minutes](https://github.com/WICG/conversion-measurement-api/blob/main/meetings/2021-10-18-minutes.md)). I raised two concerns with supporting events contributing to many buckets (~hundreds): 1. Performance. We need to make sure that aggregate designs...

1. I think MPC will be the biggest factor, although it seems even in naive non-MPC implementations there will be some kind of factor for the # of buckets contributed...

> Got it! This definitely makes sense. Here, Linf is used to control the L2 sensitivity for Gaussian noise we applied. In other words, if Linf=1, the L2 sensitivity is...

Revisiting this, I am wondering if it is feasible for parties to advertise via some global configuration what kind of sensitivity bounding they are interested in. This would have to...

Parties: reporting origins Users: devices / browsers Essentially I am thinking of a speculative new mechanism where e.g. criteo.com hosts a file saying "Please bound my contributions such that the...

cc @mthiesse. The browser is configurable by the calling app. In the example code we just use the default browser. 1. CCT works just like a Chrome tab so it...

Thanks so much @shigeki ! This is a great write-up. I'm still fully digesting, but I have a few quick comments: **3.1 Reporting Loss** Thanks for providing this data. We...

> For example, when ad-tech find a report of which impressiondata timestamp is 2day-(1hour+10min) at the time of receiving, they can look for and extract conversion log from around 70mins...

Thanks for filing Lamrani. I agree doing ROI optimization is hard in the event-level API. I think the current best thing you could do is use a combination of event-level...