I have a theory on how we can make our internet more private without trying to find the perfect solution.
So first let’s look at how tracking software works today. The main purpose is to gather as much data as possible in order to use the law of large numbers as an advantage to “predict” the future actors of an agent within a system. Using this assumption we can look at what is the purpose of big data.
The idea behind big data is that given enough data, we can discern non-random patterns to gain insights about an agents future actions and past actions. However there’s a distinct difference between these two. On one hand past actions are deterministic, assuming they were properly recorded, and therefore reveal correlateable information. However, with future actions there’s no detereminism to the actions that will be taken in the future because of the inherent dynamic nature of the system.
However, in both cases, there’s a commonality which is that we’re using past actions to identity non-random patterns. This non-randomness is inherently what allows us to “track” users today and predict future behavior as well as deduce insights about past behaviors. In principle, the idea is that given a set of actions, is it possible deduce patterns of behavior from that set of actions that are non random?
Now let’s take this idea and flip it on it’s head. If the assumption of “tracking” is that a set of behaviors need to found that are non-random then the definition of “private” would be for the set of actions to not reveal non-random patterns. Put another way, privacy is made possible when all actions appear to be random. This allows the correlation of information to remain hidden in plain sight.
So working off this idea of randomness, is it possible introduce enough noise into the system such that mixing the noisy actions with legitimate actions to make the entire set appear random? I believe this is possible, and the idea is worth exploring as we work to build an internet that is more private.
Some follow up questions, worth exploring as well.
- Is there a way that to blind my personal data in such a way that it’s completely random, but when combined with other’s data it’s non-random?
- If it is possible to extrapolate non-randomness at a macro level but not at a micro level, is it still an invasion of privacy? (This question comes with a personal bias that microlevel tracking is an invasion of privay)
- What are the boundaries of this privacy?