Related: #4, #2, #74
I'd like to point out that there should be ways for users or websites to prevent certain topics from being exposed.

Background

Some topics which are usually considered normal in most case may turn into "sensitive topics" depending on the place or situation, and plausible deniability on these topics may not work if websites or authorities consider it as an "acceptable risk."

For example, Topics like /Jobs & Education/Education/Homeschooling or /People & Society/Family & Relationships/Ancestry & Genealogy can be sensitive information if a user is living in a country where homeschooling is either illegal or commonly associated with certain religion, or where certain ethnicity or race is discriminated or repressed.

Likewise, /Sports/American Football, /Arts & Entertainment/Movies/Animated Films(which usually is associated with Japanese animation in East Asia) or even /Reference/Foreign Language Study (which is associated with English or French language in certain regions) can be used as a sign, signal or evidence that users are interested in American or Japanese culture, regardless of whether it's a randomly generated topic or not. Such information might be critical in a country which has hostile relation with aforementioned countries. In countries like North Korea, this is criminal offense punishable by death sentence.

While users can disable Topics api to address this issue as explained in #65, there are still remaining concerns to address. Such as,

• Websites may require their users to provide enough topics in order to enable some of their features, or to register on the ground it wants to ensure users are human, not bot.
• Users and websites may be legally compelled to provide information regarding topics data.
• Websites may desire not to collect data about certain topics in a region or country where they are considered as "sensitive", AND to provide targeted advertisements using Topics API.
  • Ideally, Websites should be able to plausibly claim that they did not collect such data not because of its sensitivity, but because it was not part of their whitelisted topics.

Proposal

1. As a general principle, users should be able to plausibly deny AND claim what they are and who they are. Both plausible deniability and plausibility regarding users' cohort or interest should be ensured.
2. Users and websites should be allowed to blacklist/whitelist certain topics. If a user blacklisted certain topics, They should not appear even as randomly generated topics.
3. Blacklist/whitelist should be allowed to be shared or set by browser extension.
4. Ideally, users should be able to "re-calculate" their topics data. This might be automated(providing configuration for adjusting calculation period might be a way) or done through browser extensions.
5. Ideally, topics data itself should be editable.
6. Ideally, Taxonomy of topics should be standardized and the curation process should be justifiable in a democratic and free society. and if a browser or browser extension sets custom, non-justifiable taxonomy of topics, it must be declared or acknowledged as inconsistent with the standard.