Yesterday, we learned that
Google has killed off one of its Privacy Sandbox initiative called Federated Learning of Cohorts (FLoC), and is instead replacing it with "Topics". I read the research paper behind FLoC last year and
published my thoughts on how clustering users could preserve individual privacy. At the same time, I emphasized that in order for the endeavor to be successful, "Google will need to have strong governance and security procedures in place which ensure that users cannot be de-identified by combining their data with other signals". It's clear that
Google was not able to satisfactorily resolve these concerns around monopolizing individual data, hence the latest pivot to Topics.
I plan to publish an explainer on Google's Topics API in layman terms later in the week too, but for this piece, I just want to explore the idea of ensuring individual privacy while still showing personalized ads. Google's Privacy Sandbox plans to tackle this problem using privacy-preserving mechanisms, with its latest endeavors in the area being FLoC and Topics, and while they may appear to be a technically better implementation than what we have right now, I believe the actual uphill battle is shaping public perception, not the technology.
Google could announce one technology or API after the next, but the fact of the matter is that it has to regain public trust and win support from competitors for its efforts to be successful.
Let's take Topics as an example. Throughout its
blog post, Google continues to emphasize that it will collect your interests (or topics) from a rolling time window of three weeks and then show only one interest per week to a website and its advertising partners. All your topics will be stored...............
