By the end of 2021 Google Third Party Cookies wants to abolish for advertising purposes. But while Apple and the Mozilla Foundation simply curtail the possibilities of advertising, that’s what Google is not an option. The large advertising konzern and coarse browser manufacturers in the world is true, in the context of his "Privacy sandbox"-Project alternatives to create today’s tracking-based advertising system.
The Group presented the latest proposal at the end of October: Federated Learning of Cohorts – short "Floc". The nuclear ideas of the new advertising system is to have a significant part of personalization to take place directly in the browser. Over techniques of machine learning the browser should automatically form an interest profile, the user "Cohort" – or: "Flock" – Composits.
To understand the system, you first have to look at the status quo. Advertising is barely booked on websites, but individually played every user due to the interest profile. This is made possible by programmatic real-time marketplaces, where advertising placements are still auctioned during the charging time in the browser.
Group tracking
To set up the necessary interest profiles, Third Party Cookies are currently indispensable. If a user visits about a technique website, the browser gives the information by cookie on often dozens of advertising service providers. For thousands of such signals, detailed user profiles are generated, for which, in addition to data such as age and gender of the user, also sensitive information such as diseases or religious convictions can be recorded. Add to that data handlers who combine profile information from numerous sources and unite them with the cookie profiles.
In view of the increasing resistance to the all-encompassing targeting, Google now wants to relocate the interest profiles into the browser. Chrome continuously calculates a kind of hash value from the user’s browser history and summarizes all surfers with the same value to a cohort. The cohort formation should prevent you from dragging backlusses over individual users. Attached is a minimum grape of 5000 users, which are sufficiently similar. above "Federated Learning" If the user profiles are to be adjusted together, without the raw data relying the device of the user.
Important also: Every user is only a member of a cohort. Thus, advertisers and Adtech service providers can not retrieve individual attributes such as age or gender and integrate them into their own databases.
Skeptical advertising industry
Instead of retrieving exact user profiles, the advertisers had to satisfy themselves with the knowledge of the whole cohort. For advertisers, Google’s initiative is not feeling well. However, given the increasing anti-tracking measures, a partial orientation of data is preferable to the complete loss of personalization, made Chetna Bindra, the stated product manager in Google, opposite C’t.
So that the new system works, Google needs to convince many competitors and partners from change. Therefore, Google provides its proposals on GitHub for public discussion and takes part in the standardization organization W3C on a specialist dialogue, which now has many associations and companies from the advertising industry.
But here are rough reservations. The advertising industry does not want to be booted by industry hardwoods such as Google, Apple or Facebook, made the BVDW vice-prassident Thomas Dohr opposite C’t. Who enforces with what suggest, must show 2021.
This article comes from C’t 24/2020.