For how long should Automated License Plate Reader data be kept by police?
For how long should Automated License Plate Reader (ALPR) data be kept by police?
This is a slightly different question than "Should ALPR data be maintained/kept by police at all?" I'm operating under the assumption that we should be maintaining a database of this publicly-visible data. Therefore, for me, it becomes a question about length & duration. This database is accessed via a variety of search tools & analytical queries, to help identify & locate criminal offenders.
If you don't believe the ALPR data should be maintained at all, you should probably scroll on by...
For decades, ALPR vendors have allowed government customers to determine the length of data retention. Only with a relatively new company has a vendor decided to determine that length of data retention. This company originally defaulted to 30 days; yesterday, they dropped the default to seven (7) days.
I've held police assignments in homicide investigations, burglary surveillance, carjacking interdiction. And one (1) thing is crystal clear to me: Seven (7) days is an absolute joke!
I've worked homicides with decaying bodies found weeks after the murder. ALPR data would have been extremely helpful had the retention been longer than 30 days. Cases that we solved over weeks could have been solved in days. Or hours.
Other homicides have not been charged due to crime lab & forensic analysis delays & purged ALPR data. I'm confident we could have secured charges against offenders had ALPR data been available six (6) or nine (9) months after the murder.
I've worked multi-week robbery patterns, where breaking leads came in six (6) or eight (8) weeks after the first robbery. And thankfully, an ALPR system with extended retention was able to confirm & lead to additional proof of pre-robbery surveillance by the robbery crew on their target.
I do analytical work on gun crimes, where NIBIN (ATF shell casing matching system) leads connect to cases that are years old. Imagine a gun used in a shooting or homicide being recovered months or years later during an unrelated arrest or search warrant -- & that gun's tool marks matching those casings recovered from an old scene.
I work intelligence on home burglaries. We develop suspects or recover jewelry/proceeds months later... and need license plate data from those months ago. This is how we link so many crimes together -- by comparing old data with new data. Also: Not all home burglaries are discovered right away; and not all of them are reported to police. We've met with victims months after a home burglary, after we've received geo-temporal data from other sources.
In each of the above real-life circumstances, there were hundreds of searches & analytical queries. Hundreds of searches where detectives or analysts scroll thousands of images & plate scans to find the needle in a haystack. They scroll past thousands of innocent people's cars, until they find that one (1) image of the guilty offender's car. They pull thousands, perhaps tens-of-thousands of scans, and run complicated algorithmic queries... to find their suspects. This is how the workflows go.
While not necessarily the topic of this post, these linkages & clues from ALPR data aren't always near the scene of the crime. So to retain images nearest a crime isn't the best practice. (Regional data-sharing is going to be a future topic.) The leads generated or evidence gathered in these cases are often coming from far away & a long time ago.
These are my experiences. They're not one-offs. They're not anomalies. They're not "just in case." Nor "this one time.." This is how so many criminal investigations unfold. Slowly, methodically, analytically, precisely. And with as much data to exclude innocent persons as possible.
The work is tedious. Because it has to be. The answers to justice are hidden among thousands & thousands of bits of data.
Back to the question:
For how long should ALPR data be kept by police?
There is no objective answer. Only opinions. Some opinions based on experience; I hope mine is among them. Other opinions are based on business stratgy, or business continuity, or reputation management, or crisis response. Vendors should not be dictating policy; it should be determined by legislation, executive policy, & judicial case law.
The answer is not zero (0) minutes, which includes the suggestions by the "hit-only" proponents. (Perhaps "hit-only" deployment is a topic for another day? Basically, data is purged if not associated with some hotfile.)
The answer is not three (3) days, as suggested by some proposed Illinois legislation. This means a crime on Friday assigned to a detective on Monday already has no ALPR leads...
The answer is not seven (7) days, as defaulted by the current vendor-in-crisis. People hold weekly routines; these guys couldn't even extend it to eight (8) days. Total lack of investigative experience is influencing this placating nonsense!
It's not even the 30 days some have accepted before. I've had to close too many decent criminal cases because valuable data was purged. I probably had five (5) cases this week that needed data from more than 30 days ago. This week. Just my team. (Like I said, these aren't anomaly cases.)
I'm starting the bidding at three (3) years.
But I'm also going to advocate for tight controls, audit trails, verified search purposes, & strict penalties for misuse. It shouldn't be burdensome for police officers to access publicly-available (per Court decisions) information. It also shouldn't be freely available to police officers without justification. I'll perhaps even acquiesce a few extra button clicks & verifications to extend search responses beyond lower default retention allowances. But three (3) years is reasonable for addressing the numerous delays, lead generation, & procedural schedules that I included above.
Your opinion might differ. I hope it comes with decent experience working with ALPR systems.
***
I'm drafting a model policy on how ALPR data could & should be accessed & retained. It's a work in progress. I decided to tip my hand, just a bit, in this post. The policy is more nuanced than this.
If you're wondering what sort of mistakes ALPR systems make, click here. It includes the importance of police officers verifying all sorts of pieces of data before taking enforcement action. I've been writing a lot on this technology...
***
Follow Lou on LinkedIn, & also the LinkedIn page for The Illinois Model. ***
Comments
Post a Comment