Facebook patents tracking method using dust on camera lenses
As of early 2018, Facebook's friends recommendations (People You May Know) are based on the address books users give them. However, Facebook has been filing patent applications for a new generation of technologies for collecting more information about its users and matching them more accurately. One, filed in 2014, describes technology to discern whether two people might know each other from smartphone data including location, accelerometer, and gyroscope readings, which show how often two phones are co-located and whether they are facing each other or walking together. The company says it has not used this technology except for a brief test in 2015.
Another patent, filed in 2015, describes connecting people by comparing the metadata associated with the photographs they upload, including times, dates, file names, and artefacts such as dust or scratches on the lens. In other patents, Facebook has privileged recommendations for people with few friends, for those who don't use Facebook much, for ranking friend suggestions by hometown, mutual friends, or interests, and for using external feeds and third-party databases to identify and suggest relatives.
The company says it isn't using any of these; that it frequently patents technology it never implements; and that the existence of these patents does not indicate its future plans. Those applying for patents are not required to show that the technology works. However, Facebook values user connections highly because they increase engagement (and therefore advertising consumption). It does use other technology it began patenting in 2012 to build shadow profiles to connect users by associating the contact details they upload. If all these techniques were deployed, Facebook could implement a comprehensive tracking and surveillance system.
Writer: Kashmir Hill and Surya Mattu