FLoC does reduce tracking: instead of many third parties building a thorough picture of your activity server-side, your page-level activity stays in the browser. Which already needs to know that, so it can maintain your history and turn links purple. Only the aggregated "cohort" is available to sites.
(Disclosure: I work on ads at Google, speaking only for myself)
When one of those cohorts is narrow enough to reveal something specific and highly private (like a specific disease / medical / legal / societally high interest issue) - what then? Sites get to learn that you are associated with that very private concern.
No. I am against any and every form of personalized AND aggregated categorization, because until the associations created are no longer used as criteria by others, no form of tracking is neutral.
Cookies AND FLoC and any other replacement all need to stop.
And those who want to characterize others need to stop, because the information WILL be abused.
If this were to succeed, then not too far out any website a user visits on the Internet can basically reverse engineer to a high degree (and at least approximate) the sites a Chrome user was on.
Additionally, combined with just IP address its probably a 99+% precise user identifier.
> It uses SimHash, so any website a user visits on the Internet can reverse engineer the sites a Chrome user was on.
How so? A user's cohort is shared by many other users, and for any individual site you have visited it is very likely there will be another user with an identical cohort who has not visited that site. This means you cannot tell exactly which sites any individual has visited.
> combined with just IP address its probably a 99+% precise user identifier.
IP address alone is already a massive fingerprinting leak, and needs to be addressed if browsers are to prevent cross-site tracking. Chrome is also working on this: https://github.com/bslassey/ip-blindness
Why would anyone care about that, as opposed to, you know, reducing the amount of tracking?