Just saw this presentation, which was much easier to understand -
https://docs.google.com/presentation/d/ ... -P-TY/edit
This entire system seems to rely on trusting adtech servers with users' privacy: IIUC it leaves it entirely up to adtech servers to A) actually apply the data-scrambling measures outlined in the presentation, B) do a good enough job at it that the data really can't be unscrambled, C) not go back on (A) and (B) later, and D) not get hacked (or planted) by malvertisers trying to subvert the system.
Am I reading it right that the only privacy implemented on the user's device(s) is that websites can't enumerate existing match keys?
This has me scared to update to Firefox 98+ and not completely sure I should even be trusting the 97.0.1 I'm using now. Is there any indication of when/if Mozilla is going to implement IPA in Firefox? And if they are what options people building FF from source would have?