More actions
Main article: Ratings system
The concept of privacy is central to how the ratings system will be designed. Privacy settings will allow users to change the levels of privacy at either the personal (subjective) or community levels. Within the subjective level, individuals control their own settings and how they interact with their contacts in a peer-to-peer paradigm. At the community level, members who have joined voluntarily will decide the settings for decision making, discussion, and debate in a presumably direct-democratic manner.
Given the influence of John Rawls on the thinking behind freedom in a ratings-based society, we might preface some of these ideas with how Rawls may have viewed privacy, since he did not address it directly. A look at how culture and privacy might work within the ratings system. We might also address the subject of privacy and fraud since they usually tend to work against each other. The ratings system will have built-in protections against fraud and fake identities, adjustable by users/communities.
Privacy is closely tied to the concept of identity and especially how fake identities play with the ratings system. We have additional concerns with identity in a community-based ratings system and yet more concerns where we postulate the existence of AI-based identities. However, the issue of identity trust is mitigated in the subjective ratings system. Needless to say, privacy and the prevention of fraud naturally work against each other which leads to the concept of a reasonable balance between the two once technology-based mitigation strategies have been exhausted.
We also have issues with privacy in ratings aggregation. Aggregate ratings are certainly made available to the requestor and perhaps even the public. But even if each individual rating is private there are ways of guessing the individual values from the aggregate, thus compromising the privacy of each contributor. Several technical protocols can be advanced for dealing with this possibility. These include schemes such as differential privacy, secure multiparty computation, and homomorphic encryption along with the subject of collusion between parties to reveal an individual contributor's information. With this analysis, it is useful to compare and contrast the subjective and community ratings system in light of privacy.