> Allow companies to collect data, but have federal agency oversight and requirements up to and including the right to be forgotten.
I agree with your overall sentiment, but I think the solution you propose is bad. It allows companies to collect data automatically behind users' backs, and then tries to address the extreme power imbalance that creates with regulations and escape hatches that users can take after that data has already been created. But ideally, we want to prevent problems from occurring at all, not just to try and fix them after the fact.
To your point about voluntary tracking, users may want to signal interest in some topics but not others. They may want to signal interest about some topics only in certain situations. They may want to signal interest about a topic to one site, but signal interest about another topic to a different site. They may want to "freeze" the topics that they signal interest about, so that they don't need to worry about accidentally revealing new preferences in the future, or to think about how browsing a set of sites will change the ads that they get. All of these are valid things for a user to want to do.
It is better if users have complete agency over what data is transmitted, and if companies do not fingerprint or correlate users behind their backs at all. Attempting to determine user cohorts clientside is the correct move, arguably the only part of FLoC that Google got right. The problem with FLoC is that even if cohorts are determined completely clientside, the user still does not have agency and transparency about those cohorts.
There is theoretically a version of FLoC that would be really good for privacy, and that would also allow the kind of optional tracking that some users want. Such a system would need to be easy for users to toggle on and off on the fly on a site-by-site basis. It would need to be transparent to users about what categories they're in, it would need to allow users to edit and delete their categories, to add new categories, and to turn off category aggregation. It would need to not send new categories to a website without the user's permission. It would need to block websites from determining whether they're being given an automatically generated category, or a user-defined category. And importantly, it would all need to be opt-in.
Right now, tracking is an invisible system that happens automatically, with (rare) escape hatches that the majority of users are not expected to pursue, and that give little to no granularity about how users can customize how they present to the world. This kind of tracking is inherently invasive. We should pursue a paradigm shift where users proactively volunteer information about how they want to present to the world in any given situation -- in other words, we should give users agency to choose their own identities online.
Most regulatory suggestions that I see online (including the one you give) consciously or unconsciously focus on preserving the invasive nature of tracking while adding safeguards to try and prevent advertisers from abusing the inherent power imbalance. Instead, that power imbalance ought to be totally flipped in favor of the user. We need to reject the underlying idea that preferences should be extracted from users, rather than something that users consciously choose to volunteer.
----
I've commented to the same effect a couple of times in the past:
- https://news.ycombinator.com/item?id=25906791
- https://news.ycombinator.com/item?id=26353494
- https://news.ycombinator.com/item?id=25907079
This isn't only true for advertising. Think about how little control users have over how personalized algorithmic content filtering/ordering works on sites like Youtube, and you'll see similar conceptual problems with how the systems are designed. But it's really apparent in advertising, moreso than in other contexts.
I want to fight back against policy changes that (in my mind) cement and normalize the current philosophical approach to advertising. It's not OK to be tracked without your permission, and regulations or safeguards and escape hatches around tracking aren't enough to make it OK. The system would still be fundamentally broken.