It would group users into interest-based clusters
The use of browser cookies is a controversial practice, but a necessary one for many sites throughout the internet. However, Google is looking into one alternative that requires less personal user data while still holding significant promise for advertisers — an API called “Federated Learning of Cohorts.”
The goal of FLoC, in Google’s eyes, is to deliver advertising results “nearly as effective” as cookie-based approaches by grouping large groups of users into clusters called “cohorts.” These cohorts will organize people into ad groups based on interests they share.
Ideally, this would lead to less of your specific tastes and interests being shared with data companies — instead, you’re just another fish in the sea. That’s perhaps not a flattering comparison, but when it comes to web privacy, it’s far better than the alternative.
Google does acknowledge that users could have their cohort recorded and revealed if a site has gained enough personally-identifiable information about them, but without access to something like your IP address, it ideally won’t be enough to individually identify someone.
Even in cases where it might, Google hopes to implement a few design mitigations to obfuscate the practice, such as ensuring cohort sizes are “large enough that they are not useful for tracking.”
You can read a full breakdown of how FLoC will work on its dedicated GitHub page here. Whatever you might think of Google, it’s clear that the intentions, if nothing else, behind this project are good — or, at least, much better than what we have now.
Hopefully, the FLoC standard will be adopted by most of the advertising community. If it is, browser makers can begin phasing out third-party cookies entirely.