Everyone leaves telemetry breadcrumbs everywhere. Almost everyone radiates a ton of different kinds of data, RF, optical, etc fingerprints that are incidentally picked up everywhere they go. In fact, there have been interesting papers on tracking entities that don’t leave any breadcrumbs because they visibly perturb the entities that do! Existing in the world has side effects that are picked up by all kinds of systems and it is analytically possible, with sufficient large quantities of data, to start attributing those side effects to specific entities. That’s the game.
The caveat is that it requires large quantities of data. Many, many petabytes is table stakes.
What you are not grokking is that this can be done with data that is not designed to track anyone but which can reconstruct identity with sophisticated analysis across many innocuous data sources. This has been one of the unexpected headaches of GDPR; you can often identify people with boring industrial sensors that were designed for industrial purposes with sufficiently sophisticated analytics. The fact that this is possible has made this data (legally) PII.
A person that literally doesn’t own a mobile phone is still trackable. The whole “burner phone” was always a movie trope, that hasn’t worked for decades if ever.
The ability to do graph reconstruction across dirty, noisy spatiotemporal events that may have nothing to do with the target in question is robust. It isn’t trivial, but at this point it is a mature field of endeavour.
>Everyone leaves telemetry breadcrumbs everywhere. Almost everyone radiates a ton of different kinds of data, RF, optical, etc fingerprints that are incidentally picked up everywhere they go.
What does this mean in concrete terms? "optical, etc fingerprints" just sounds like a fancy way of saying "facial recognition". I'm also not too sure what "RF" translates to. Is it just wifi/bluetooth/celluar sniffers?
Everyone leaves telemetry breadcrumbs everywhere. Almost everyone radiates a ton of different kinds of data, RF, optical, etc fingerprints that are incidentally picked up everywhere they go. In fact, there have been interesting papers on tracking entities that don’t leave any breadcrumbs because they visibly perturb the entities that do! Existing in the world has side effects that are picked up by all kinds of systems and it is analytically possible, with sufficient large quantities of data, to start attributing those side effects to specific entities. That’s the game.
The caveat is that it requires large quantities of data. Many, many petabytes is table stakes.
What you are not grokking is that this can be done with data that is not designed to track anyone but which can reconstruct identity with sophisticated analysis across many innocuous data sources. This has been one of the unexpected headaches of GDPR; you can often identify people with boring industrial sensors that were designed for industrial purposes with sufficiently sophisticated analytics. The fact that this is possible has made this data (legally) PII.
A person that literally doesn’t own a mobile phone is still trackable. The whole “burner phone” was always a movie trope, that hasn’t worked for decades if ever.
The ability to do graph reconstruction across dirty, noisy spatiotemporal events that may have nothing to do with the target in question is robust. It isn’t trivial, but at this point it is a mature field of endeavour.