With “Privacy Sandbox”, Google wants to save its advertising business while preserving your privacy
The publisher of Chrome wants to create a technology to make advertising targeting online while protecting the personal data of users. This quest for the Grail is based on a complex set of mathematical techniques.
The giant of Mountain View went in search of the five-legged sheep. In several blog posts, Google gives details of its new concept of “Privacy Sandbox”, a set of web technologies capable of delivering to advertisers useful information on the personality of visitors to a site without revealing their identity and their lives private. In other words, the “Privacy Sandbox” will make it possible to know who one is dealing with without having to identify it in a unique way. A subtle nuance that – one suspects – is quite complex to achieve.
To get there, Google is exploring different tracks, some of which are now detailed on GitHub pages. A first principle is the aggregation of data of Internet users. Rather than creating a unique advertising identifier for each user that will be associated with a list of interests, it would be enough to create group identifiers called “flock” (Federate Learning Cohorts Key).
Clusters of people and query ceilings
These clusters of people would be precise enough to allow a useful advertising targeting (example: “people loving death metal, sweet drinks and holidays in the mountains”), but they would also be large enough for each user to be drowned in an anonymous mass. To achieve a satisfactory level of anonymity, there should be relatively few clusters and at least several thousand people.
Another idea is to introduce a ceiling of data requests for each user and each website. Called “Privacy Budget”, this technology would be able to limit the amount of technical data that a browser would communicate to a site, so that the latter is unable to identify the user by fingerprinting. This technique, remember, allows advertisers to identify Internet users by softly collecting technical features of the browser and terminal. To limit requests, the Privacy Budget would rely on cryptographic tools such as “k-anonymity” or “differential privacy”. The first is to drown someone’s data in a set of people. The second makes it possible to anonymize a set of data by introducing mathematical noise.
Weapons to fight against advertising fraud
Obviously, Google also wants to help advertisers. He wants to rely on a technique called “Privacy Pass” to separate real users of clickers robots and, thus, improve the fight against advertising fraud. The technique of the “Privacy Pass” makes it possible to give a privilege of access to a user without this one being forced to reveal his identity. It relies, for this, on algorithms called “zero-knowledge proof”.
For now, nothing is really engraved in the marble concerning the “Privacy Sandbox”. But one thing is certain: Google takes this new initiative very seriously. So much so that he would like to make it a universal standard of the web. Why ? To save his business. Mozilla and Apple are continuing to tighten their personal data protection policy, and Internet users are becoming more and more receptive to this topic.
But Google believes that their approaches are too radical and ultimately inefficient. Confronted with a massive blockage of their cookies, advertisers are increasingly turning to opaque techniques such as fingerprinting. With Privacy Sandbox, Google wants to offer a third way, more respectful of personal data, but also profitable to advertisers. A ridge line that may be hard to keep.