Topics API is not just a system for interest-based advertising.

This proposal is a general-purpose system for categorizing and ranking users based on web history. Many other parties besides advertisers might want to be able to classify users by interest, even imperfectly with random noise.

• Retail sites offering dynamic or personalized pricing (Cover dynamic pricing use case #34)
• Learning management systems classifying likely diligent and distracted students for recommending extra teacher attention
• Property management firms selecting likely noisy tenants for extra checks based on noisy or quiet interests

The extra, non-advertising uses that Topics API offers mean that users are less likely to accept it—a user who might have wanted personalized ads based on their interest in boating might not want to miss out on discount codes offered to less affluent non-boaters on unrelated retail sites. A user might want to get ads for new show dates for their favorite metal bands, but not find themselves denied a vacation rental in a quiet area.

Alternatives to Topics API that are more directly focused on advertising, and likely to be more acceptable to users in the long run, include FLEDGE interest groups and Seller-Defined Audiences.

This proposal is a general-purpose system for categorizing and ranking users based on web history. Many other parties besides advertisers might want to be able to classify users by interest, even imperfectly with random noise.

So you think the discussion would be less controversial if the derived interest would be bound to an advertising centric technical environment/purpose, like (just out of the top of my mind) injecting them as an interest signal into a FLEDGE style auction or more generally using them only in a purpose bound infrastructure based on TEEs (like a lot of other Privacy Sandbox Proposals) or MPC?

Yes, exactly. You don't want to turn on an option labeled "Topics of interest are based on your recent browsing history and are used by sites to show you personalized ads" and then Child Protective Services shows up because you went to your kid's school site with the wrong topics.

Besides limiting it to an ad environment, another alternative is to borrow the public review mechanism already proposed for First-Party Sets and use that for Topics API as well (#87). That way the browser could limit callers who were using it for other than the documented purposes.

Gotcha - so maybe something to consider for the Chrome team, would change the overall approach quite significantly thought (not in terms of deriving interest though)

The Topics API is certaintly not a perfect proposal and more work needs to be done but we appreciate the open nature of this initiative by Google which is positively different to what many other tech providers have done in the past. We also think that an outright rejection of such a proposal without suggesting any alternative solutions is not constructive. Users made it clear that they want content and advertising tailored to their interest and needs, as demonstrated by this survey (https://www.europarl.europa.eu/news/en/headlines/society/20170627STO78425/consumer-protection-how-parliament-safeguards-your-rights) from the European Parliament or this one from Optimizely (https://www.marketing-boerse.de/news/details/2212-studie-was-fuer-verbraucher-in-deutschland-2022-wichtig-ist/184325).

While I agree that we need to have a constructive discussion on the merit of interest-based advertisement and whether or not Topics API proposal should be accepted. However, I think there might be a more suitable venue for such purpose than here. If you're interested, I recommend you further follow up related discussions held during the initial design review.

Also I would like to emphasize that Topics API, if adopted and implemented, will be functioning as a general-purpose system; and also to point out that this nature of the API generates serious divergence of views among related parties, including stakeholders, users, regulators, vendors, concerning its far-reaching implications for privacy, competition policy as well as security, fairness, inclusivity. (#74, #76, #78, #84 etc.)

to @bvdw-ev
If data brokers and their satellites need lose 40 billion, so be it. If services available now has to rely on donations, paywalls, or downsize considerably, or even disappear, so be it.
If a 5cts donation is necessary but bothersome to create a transaction for it, you can consider mining 15 seconds (and thus paying through electricity bill by proxy in effect). Or creating a torrent-style internet where your users are required to participate in the bandwidth of your service for the duration of the stay.
But let it be clear, those businesses won't be regretted. They are parasiting society.
It's either:

• pay yourselves on our backs, using data trading that is essentially secret from the users, at least most certainly still quite obscure or unknown to most naive web users. Which is disingenuous.
• request direct money. Which is transparent.

Money doesn't lose itself, it's not energy. It's just account tracking, it doesn't work as a benchmark for society happiness or improvement. https://hbr.org/2019/10/gdp-is-not-a-measure-of-human-well-being
As Jean Marc Jancovici said, forgetting your light on in the toilet, only improves one thing: the GDP.
If that money isn't spent on that sector of the macroeconomy, it will be spent elsewhere, stocked, invested, or reimbursed (participating in deflation).
What do you want to do, fabricate artificial protections for preservation of systems in place, without thinking about why we do it, just because it's scary to let things go bankrupt and in the zone of discomfort? Or do you want to let society evolve, and follow its natural "Darwinism" through bankruptcy? Classical liberals were believers of the second. The first is more neo-liberal. Possibly even a sort of socialist-conservatism weird hybrid.

Related: #216: purpose specification: make the purpose explicit, or, if general-purpose, update to reflect the wider range of uses/abuses