I think the scariest part of is in the first part of this; the collection of data was voluntarily crowdsourced by repo men all acting as independent actors. While right now there is still a centralized corporation that can be targeted to stop the practice, I can see a future where this stuff becomes decentralized in the same lieu as software piracy sites and their infrastructure is done today, maintained by "helpful" repomen/investigators that each have a pervasive interest in maintaining such a resource.
The fundamental problem is that technology has made collecting this information too cheap. Instead of manual entry and uploads, it's automated with essentially zero marginal cost over and above putting the cameras on cars or buildings.
The solution is to make license plate collection expensive again - allow people to install license plate covers that people can manually lift to read the plate and automatically retract when the car is started.
I'm a huge proponent of this core idea -- that networked technologies have changed the cost structure that underpins our 20th century implicit understandings about privacy, and that one stopgap approach is to rework the economics of surveillance in a way that re-approximates the cost structure under which our legal and moral consensus was developed.
For example, it was never possible to read, secretly and automatically, in bulk, everyone's correspondence in the 20th century. With email, it's a commonplace that governments do it routinely. Encryption doesn't make it impossible to figure out what someone's corresponding about, it just makes it more expensive (as the investigator may need to devote resources to shoulder-surfing or other means instead of auto-collection). It used to be fairly expensive to have to go to the post office and steam open an envelope, copy it, and seal it back up without detection.
Likewise for various other surveillance technologies, such as the ALPR/ANPRs. The people who (still) write our laws were raised watching TV cops listen for a radio APB on a license plate. For their intuitions about this matter to be valid, it should be roughly as costly in manpower, time, and attention to do a "live alert" on a license plate today as it was for those TV cops.
A pen register used to be an actual box you put at the central office on a specific bad guy of interest, not an excuse to sweep up and cross-correlate every dialed number ever. Again, several orders of magnitude cost difference here, in a way where the difference of degree (cost) becomes a difference of kind.
The big danger IMO is when you have well-meaning legislators and jurists opining about network-enabled privacy intrusions using intuitions and examples formed with 20th century cost expectations.
Eventually those expectations will need to reflect current technology, but in the meantime, anything that can be done to raise the cost of pervasive surveillance is IMO a good thing as it buys time for legal and moral intuitions (much slower rate of change) to adapt.
Already beat you: local laws that require a valid plate to be visible at all times when parked on public streets. (Private drives/businesses subject to jurisdiction.)
There aren't just repo men scanning in my local area; police cruisers are equipped with automatic scanners, there's numerous scanners posted on traffic poles (with silly blue and red flashing lights as a CYA) throughout the metro area, etc.
Maybe states could offer CAPTCHA license plates that would be easily readable to a human police officer or other first responder with a legitimate need, but unreadable by license plate OCR.
Or license plates could become programable e-ink displays that display a new hash every X minutes so that a person with a legitimate need could rapidly identify the vehicle but which would defeat bulk passive data collection.
On the other hand, wouldn't organizations collecting this information go quickly further for emissions other than visible light? A car's occupants are likely carrying around devices emitting GSM, Bluetooth and Wifi radio signals, as well as onboard systems. The model for this is Google's collection of household radio emissions with the StreetView surveillance patrols or the gargoyles in Snow Crash.
>Or license plates could become programable e-ink displays that display a new hash every X minutes
Once a week would likely be enough. The point is that you can no longer associate "Plate ABCDEF was at this location at this time" with "Car owned by person X was at this location at this time" without making a lookup in a database that's owned by the government.
> The solution is to make license plate collection expensive again - allow people to install license plate covers that people can manually lift to read the plate and automatically retract when the car is started.
The public at large would never go for any proposal of removing the requirement for two-ton murder weapons and get-away rides to show identifying marks at all times.
Yes, and such 'crowdsourced' decentralized sites will suffer the same bit-rot as all the other ones, and we'll see all sorts of outrages caused because 5 years ago or whatever a bad actor lived at such and such an address, but the "helpful" repomen/investigators never updated things, and we'll that's just too bad and if you shut down our database everyone will steal a car and won't someone think of the children and (this is the only part I completely, 100% guarantee) no "helpful" person will take a single shred of responsibility for any collateral damage.
