|
|
|
|
|
by gizmogwai
2998 days ago
|
|
That's more subtle, as individual anonymized data point can not identify an individual, but a linked set of those can, with a very high percentage of confidence.
Facebook can always play the card that each data point are anonymous, hence not part of your data, and keep everything. |
|
We know that facebook can use contact info to identify clusters, right? Suppose that I don't have facebook, but a bunch of my friends do, and facebook has all their contacts' phone numbers. They can infer the circle of friends of "whoever owns phone number X" using just this info. This is probably not the extent of a shadow-profile, but I think it's a plausible stab at the underlying mechanics (repeat again with email, with browser fingerprint, etc).
In order to "delete" my phone-shadow-profile, facebook needs to remove my phone number from their internal copies of my friend's contact lists. They also need to keep a copy of my number in a "do not shadow" list, or else they'll just add me back in next time they scrape my friends' contacts. Armed with a such a do-not-shadow list, there's no need to actually do the deletion, rather than simply marking the deletion.
So, assuming my speculations of the nature of shadow-profiles are about right, the only way to really avoid facebook having one is to register your fingerprints with them, so they can tell which parts of their data are you, in order to know not to use that data to draw inferences about you.
That, or give up on drawing inferences between disparate data-sources at all.