> With that in mind, there's not really any kind of reasonable way to hide what devices are connecting to what tower. This is where location information comes from. It's really an inescapable part of how cellular devices work.
The comment from mirimir elsewhere in this thread reminds me that this is kind of an overstatement, or at least that the details are complicated.
You could gain a lot of privacy in this regard by separating mobile data services from telephony and identity. For example, you could imagine paying for mobile data anonymously, either using existing prepaid mobile data services or using a hypothetical future service with blinded payment tokens and only extremely-ephemeral device identifiers. Then you could imagine getting all of your identity and communications services from someone totally independent of your mobile data provider. If the mobile data provider cooperated with your application-layer communications service provider more than you wanted, you could try to create your own service instead, or try to route at least the messaging setup part of the process via Tor or other proxies, so that the mobile data operator and application-layer intermediaries didn't even know about the connection.
I'm not sure anyone other than privacy advocates would consider this progress relative to the current situation; certainly carriers and governments would like it a lot less (it might already be illegal in some jurisdictions in various ways), and most users would probably find that it increased cost while decreasing reliability and usability of some services. It could also make it harder to use the network to investigate or deter device theft, as well as harder to investigate application-layer fraud and account hijacking.
To be clear about what could change, it's true that towers will always know which devices are connecting to them at a particular moment, but this could in principle be separated from billing, identity, and any kind of persistent identifier. So they don't have to know that a particular device is being used by you, or that a particular device is being used by the same person who uses a particular application-layer identity.
The other problem in trying to get there right now without carriers' and governments' cooperation is that, since you can't rotate hardware identifiers on GSM interfaces, a carrier can see your movement patterns for the lifetime of your use of a particular device, and can probably determine that those movement patterns are similar enough to another device's movements that they're probably used by the same person.
Sometime I hope to write a long article on possible non-metadata-collecting mobile communications futures. It's a really interesting topic.
The comment from mirimir elsewhere in this thread reminds me that this is kind of an overstatement, or at least that the details are complicated.
You could gain a lot of privacy in this regard by separating mobile data services from telephony and identity. For example, you could imagine paying for mobile data anonymously, either using existing prepaid mobile data services or using a hypothetical future service with blinded payment tokens and only extremely-ephemeral device identifiers. Then you could imagine getting all of your identity and communications services from someone totally independent of your mobile data provider. If the mobile data provider cooperated with your application-layer communications service provider more than you wanted, you could try to create your own service instead, or try to route at least the messaging setup part of the process via Tor or other proxies, so that the mobile data operator and application-layer intermediaries didn't even know about the connection.
I'm not sure anyone other than privacy advocates would consider this progress relative to the current situation; certainly carriers and governments would like it a lot less (it might already be illegal in some jurisdictions in various ways), and most users would probably find that it increased cost while decreasing reliability and usability of some services. It could also make it harder to use the network to investigate or deter device theft, as well as harder to investigate application-layer fraud and account hijacking.
To be clear about what could change, it's true that towers will always know which devices are connecting to them at a particular moment, but this could in principle be separated from billing, identity, and any kind of persistent identifier. So they don't have to know that a particular device is being used by you, or that a particular device is being used by the same person who uses a particular application-layer identity.
The other problem in trying to get there right now without carriers' and governments' cooperation is that, since you can't rotate hardware identifiers on GSM interfaces, a carrier can see your movement patterns for the lifetime of your use of a particular device, and can probably determine that those movement patterns are similar enough to another device's movements that they're probably used by the same person.
Sometime I hope to write a long article on possible non-metadata-collecting mobile communications futures. It's a really interesting topic.