Now this is a very interesting observation. Privacy, from an analytical standpoint, can simply be viewed as a lack of information. Even without 100% information, you can stil do useful calculations on the data. For example, if there is an access controlled region that you cannot scan but you can scan all regions around it, then you can still compute the size of the access controlled region and use that to hypothesize what range of activities can go on in that region.
In fact, this works with a lot of other measurements too. For example, privacy in family metrics. How many children does someone have? What are their individualized interests? If the known parents only have a small number of children, then the “access-controlled zone” is small and very easy to predict from the public data.
Also, this brings in a very interesting reality, that privacy isn’t really effective if only a small number of people opt in. It is only really effective if it applies to a large number of people across a large geographic area. “Strength in numbers.”
So that’s why Google knows everything about who you are and who your children are. Even if they don’t know everything, at least they know what they don’t know and will proceed to search for that information and attempt to reveal it.
But, there are more practical applications of using this understanding of privacy as merely a limitation in data collection.
- Bathroom resource metering, at a finer-grained level. Count door opens and closes, and based off of the time of the closed door, estimate the average resources consumed.