Google has provided new information about the disrupted development of FLoC (Federated Learning of Cohorts) that it hopes to use as a replacement for cookies. Google has chosen to end the project and offers a replacement: Topics.
The search giant’s first attempt to replace the third-party cookie with its own technology was met with fierce opposition from some, mistrust from others, and very little positive response. Originally, Google committed in early 2021 to ending support for third-party cookies in its Chrome browser in 2022. At the time, Google intended to replace cookies with a new technology that the company said was much more anonymous and still delivered 95% conversion rates for every advertising dollar spent.
Obviously, things did not go as the company had hoped. Google finally ended development of FLoC in July 2021, around the same time the company announced that Chrome would continue to support third-party cookies until at least mid-2023. So far, the company has been tight-lipped about how it intends to move forward with its cookie replacement plans.
Dubbed simply “Topics,” Google’s new proposal aims to track users anonymously using a new API designed to meet Google’s four main privacy goals:
- The technology must make it “difficult to re-identify a significant number of users on sites using only the API”.
- It should offer a viable replacement for “a subset of third-party cookie capabilities”.
- Any data recorded must be “less personally sensitive” than what is collected today.
- The API must be understandable to users and transparent in its intentions.
Google believes that its Topics API meets all of these criteria while providing the data that targeted ads need to continue operating at a similar level to their current cookie-based business.
In addition to posting a GitHub entry revealing the technical details of Topics, Ben Galbraith, Head of Google’s Privacy Sandbox, also held a press briefing where he revealed additional settings. In particular, he explained that Topics will initially attempt to track user behavior across 300 to 350 specific interests. These domains are based on the IAB’s Audience Taxonomy, which has a much more comprehensive list of around 1,500 interests to track.
Google’s post on GitHub says this is an early design, hinting that these roughly 350 domains could be expanded in the future. According to Galbraith, if so, they won’t extend to what Google calls “sensitive topics,” which include things like the user’s race and gender.
In practice, the Topics API allows the user’s browser to share three of its centers of interest detected when the user visits a site using targeted advertising. The API will randomly select these three topics from the first five it detects. A topic will be chosen from the top five for each of the previous three weeks to provide a more accurate, but still anonymous, picture of the user’s recent online browsing history. Google wants users to be able to engage personally as well, stating that they will be able to opt out of tracking their specific interests while still being able to review the interests that have been chosen for them at any time.
This level of transparency and user control addresses two of the main issues raised by critics of the FloC project: it was deemed too opaque and added too much data that could be used by fingerprinting technologies to the system. . The promise to avoid “sensitive” topics also addresses a problem with the FloC system, which automatically created cohorts around topics such as gender and race.
Google plans to start testing Topics with third parties later this quarter. Whether this technology will fare better than FLoC or whether Google will once again be forced to continue accepting third-party cookies in its Chrome browser for years to come remains to be seen.
Source: “ZDNet.com”