How about this: e-ink plates plus public-key cryptography.
Your actual license number or other identifier, plus a time-based nonce, is encrypted with the DOL's public key. The displayed value changes as the nonce changes according to its schedule, so third-party observers can't correlate the displayed value across time.
If you get in a hit-and-run and note the displayed plate, the DOL has the other half of the keypair, the time, and the derivation function for the nonce, so can translate the displayed value to the actual owner.
Not being able to publish a single stable value in amber alert cases would be a bit of a regression, but you could still publish what a value would be at a particular time interval.
Doesn't do anything about governmental abuses of ALPR data but could be effective at cutting out corporate abuses. I'm probably missing something but it doesn't seem to increase info leakage w.r.t the status quo either - you'd theoretically be able to figure out when a particular image of a plate was taken, but that source would almost definitely be timestamped anyway.
e: I don't think "just ban private ALPR" is a solution; it's simply way too easy to do with COTS+FOSS and way too hard to enforce against.
How about this: e-ink plates plus public-key cryptography.
You’re suggesting that a solid plate of metal that can sit, neglected, out in the weather for multiple years without much visual wear, and when damaged by the car wash can just be bent back to shape, and replace that with your delicate little piece of electronics and software? And pile on some PKI to boot?
I’m seriously on the fence in deciding if this comment is trolling me, or if this is what late-stage HN looks like. :-)
It was definitely at least partially tongue-in-cheek ^_^
I bet you could figure out the physical aspects. E-ink tech itself has come a long way in the last few years following some patent expirations, and the electronics stuff is basically just a yubikey JB welded to a license plate frame. The cost per unit would be pretty low at scale, so just replacing borked units seems pretty doable.
Imo, a bigger problem is competent implementation. Yeah sure, the DOL is gonna run a bunch of PKI infrastructure and not mess that up. At least in my region, just keeping a largely static website up seems to be a struggle.
> I don't think "just ban private ALPR" is a solution; it's simply way too easy to do with COTS+FOSS and way too hard to enforce against.
Which leads to another issue, that local governments have contracted these corporations to do just this. From red light cameras to suvellience cams, police don't actually store this data themselves, private companies do the bulk of the work here.
It makes sense as a technical solution to the problem of not being tracked / identified by NGOs. Could work but seems a little complicated, and unlikely to end there; car, location & face recognition could achieve the same ends, by and large. (China already does this). However the bigger issue to resolve is the goals and legality. We haven’t yet established that being able to identify someone in public is bad, or conversely that being able to travel anonymously is a goal we want, right?
> car, location & face recognition could achieve the same ends, by and large. (China already does this).
Yeah, I actually started out writing that comment about how license plates are probably unnecessary given the volume of other forms of location data accessible to LE but the peak HN strat was more fun to think about.
> We haven’t yet established that being able to identify someone in public is bad, or conversely that being able to travel anonymously is a goal we want, right?
I don't have full answers here, but I think it's worth considering the modes of enforcement enabled by this change. Despite there being no de jure change in privacy protections while in public, there's been a de facto change from that kind of data only being accessible in cases of specific, targeted investigations to that kind of data being accessible to automated dragnet enforcement. Targeted investigations are inherently limited in scale and there's (at least theoretically) a nexus between the investigation and some kind of probable cause, but dragnet enforcement generally disregards fourth-amendment protections. The Carpenter decision theoretically offers some protection against this, but parallel construction is trivial enough that I'm not exactly resting easy.
So, I think it is possible to be against ALPRs without necessarily being for wholly anonymous travel in public - it's an issue of probable cause and avoiding the fruit of the poisoned tree, not one of absolute lawlessness. My (admittedly silly) suggestion is also problematic because it doesn't address this concern at all. My real feelings are a lot closer to 'calvinmorrison, but I acknowledge that "just get rid of license plates" isn't exactly a winning proposition to the average voter.
Your actual license number or other identifier, plus a time-based nonce, is encrypted with the DOL's public key. The displayed value changes as the nonce changes according to its schedule, so third-party observers can't correlate the displayed value across time.
If you get in a hit-and-run and note the displayed plate, the DOL has the other half of the keypair, the time, and the derivation function for the nonce, so can translate the displayed value to the actual owner.
Not being able to publish a single stable value in amber alert cases would be a bit of a regression, but you could still publish what a value would be at a particular time interval.
Doesn't do anything about governmental abuses of ALPR data but could be effective at cutting out corporate abuses. I'm probably missing something but it doesn't seem to increase info leakage w.r.t the status quo either - you'd theoretically be able to figure out when a particular image of a plate was taken, but that source would almost definitely be timestamped anyway.
e: I don't think "just ban private ALPR" is a solution; it's simply way too easy to do with COTS+FOSS and way too hard to enforce against.