This is entirely correct actually. Number of items on day n = (13 - n)*n

Looks like they are using a Prio based protocol. If they are using Prio2+, I think this article is likely overblown. EDIT: I mixed up my sources - Mozilla tested Prio for telemetry collection. They are using a system called IPA for ads, and I don’t know whether there are formal guarantees for this system

Does Firefox explain what measures they’ve taken to protect their aggregation servers? If so, this is a perfectly fine and practical method for privacy preservation

I do wonder how many are bots. Most of Twitter is bots anyway