Google implements data protection compliant BigData

Distributed computation renders central raw data storage superfluous

First of all: This is neither advertising for Google nor are all BigData data protection problems hereby solved. However, here we (at last) have a technical approach proving that BigData and data protection are not mutually exclusive.

In an excellently readable Blog article 1 , Google has introduced several new keyboard app features. On first sight, this is not sensational, but the title “Machine Learning without Centralized Training Data” is intriguing: The studies, methods and protocols presented here will allow BigData applications compliant with data protection!

It is specifically about predictive text when typing on the Smartphone. A function that data protection sensitive people often switch off because the prediction normally takes place in the Cloud. This means that everything that we type will be sent to the provider. Until now.

Google is already using the technology described in the Blog for its current keyboard (Gboard app). Google has hereby simply installed data protection – without any fuss. Not only as an end in itself, of course: Google in this way uses the computing power of all Smartphones to train its own neural network. This is fair because the user will benefit – improved predictive text.

And all this without Google ever receiving the typed content. And this also saves transmitted data volumes – as a fringe benefit. As part of the introduced communication protocol, even reverse-calculation of individual training updates sent by the Smartphone will be impossible: this (verifiably) ensures that only compressed data can be evaluated. Even taking this protocol by itself (also for other applications such as surveys) represents a true gain to data protection.

It will now be your and our task to continue along this path: Go mad. Flip out. Don’t try to solve new (BigData) problems using (only) old-school methods.

Demand such solutions from developers in your company. Report that alternatives exist. And that these ideas are not crazy, but are already found in products and running on billions of devices.

Enlighten the political arena that these are we should be developing instead of sacrificing to economic interests the basic rights we have struggled for over the years. At this point: Chapeau,

1 Anyone wishing to read Google’s Blog in German is welcome to visit – also excellent reading.