ALPRs | LARE

The Issue: Flock Automated License Plate Readers (ALPRs) are being proposed in Los Altos. We want to make sure that the City Council is clear on what problem we want to solve by using ALPRs, whether ALPRs have been shown to be effective in solving that problem, how we measure its effectiveness, and whether the benefits in using ALPRs outweigh the cost.

What problem are we trying to solve?

Chief Averiett has proposed ALPRs to “increase our property crime solvability rates by 10%.” So from 4% to 14%. The California DoJ defines a “cleared crime” as one for which an arrest has been made, or the culprit identified but cannot be brought before a judge. This is a worthy goal, as our clearance rate for the past 6 years has been around 4%, which is extremely low.

What is the proposed solution?

The Police Department would like to hire a private company called Flock to install and run ALPR cameras. The pilot would have up to 25 fixed cameras, placed around Los Altos.. Every car that passes by a camera would be photographed, and automatically placed into a database. The artificial intelligence system is supposed to identify the type of car, the color of the car, and whether it has roof racks or bumper stickers or other features. The system will only take pictures of the back of the car, so that the driver cannot be identified.

Once all of this information is stored, there are two ways the police can see it. The first is automatic. The Flock software would match detected license plates against “hotlists” such as the California list of stolen vehicles (SVS). If a license plate on any of those lists is seen by the Flock cameras, a “hit” occurs and an immediate notification goes to the police, so that they can send an officer.

The second way the police can see the data is by querying the database as they are investigating a crime. If the police know that a certain crime happened at a certain time, they can ask the database to tell them about all of the cars that drove by that location around that time. Or, they can search for a particular vehicle or license plate number. These searches are logged and can only be done with a reason typed in.

What is the monetary cost?

Each Flock camera costs $2500/year. Our proposed system would be $75,000/year plus a one-time setup cost. In addition, there would be a cost for police department staff time to operate the system. Los Altos Hills reported a cost of their 40-unit Flock system as $110,000 in FY22-23, and an additional staffing cost of approximately $114 000, so it is safe to assume that our system would be over $100,000 when staffing cost is included.

Conclusion: The ALPR system is expensive, so we may achieve better results in solving property crime cases by hiring a detective.

Do ALPRs increase solvability of property crimes?

There are conflicting results. One study that was done on ALPRs and vehicle theft in Arizona showed that ALPRs did help in returning stolen vehicles and led to more arrests[2], but made no difference in reducing vehicle theft. Another study on the efficacy of ALPRs showed lower vehicle theft after ALPRs, but only a weak correlation between vehicle recoveries and plate “hits”[3]. In both of these cases however, stolen vehicle recovery does not equate to case clearance.

In LAH, ALPRs did not clear a single residential burglary case.[See LAH data below]

We cannot rely on anecdotal data from various police departments that highlight examples of thieves caught for their crimes because there is also anecdotal data in the news that highlight examples of wrongful arrests, and situations with police drawing guns on innocent victims due to incorrect information from ALPRs and hotlists.

Conclusion: Not enough data to support or refute that ALPRs increase clearance of property crimes.

Are ALPRs safe?

There are real concerns over personal privacy, data misuse, and data breaches. ALPRs have been used by police in illegal, biased, and questionable ways - to mine data and impose fines[30], spy on Muslim worshippers[31], track protesters[21][22], and share data with ICE[23][32]. There have also been large data breaches associated with ALPRs[33]. Many of these egregious occurrences have been with the Vigilant ALPR system; Flock has much better policies around data retention and data sharing, but strict local policies should be instituted to protect against deliberate misuse and because company policies could change at any time.

The other factor to consider for ALPR safety is their error rate. Flock safety touts a 93% read accuracy rate, but in the field, studies show much lower accuracy with the hotlists. A study in The Police Chief magazine found that 37% of fixed-position ALPR hits were erroneous.12 Errors can happen in several ways. The system can misread a license plate, mistaking a 7 for a 2, for example.13 Or the system can read the plate correctly, but the hotlist can contain erroneous information; for example, a rental car was automatically reported stolen when it was returned late, and the license number was never cleared from the stolen vehicle list. We don’t have good data on how that 37% error rate breaks down between these kinds of errors. But it is these errors that can lead to life-threatening encounters of innocent civilians with the police and result in significant litigation[16][17][18]. Therefore, good policies to manually verify license plate numbers and handle ALPR hits as prone to error are crucial.

Conclusion: ALPRs have extremely high error rates in the field, which can make them ineffective as a policing tool. Strict policies on ALPR usage are critical to ensure privacy and safety for the community.

What are surrounding communities doing?

Los Altos Hills has had a Flock ALPR system of 40 cameras since April, 2021. Their ALPR system was renewed in 2022, and again in 2023. They fall under the Santa Clara County Sheriff’s Office (SCCSO), and do not share data with any other agencies. On Feb. 16, 2023, the LAH City Council looked at a 1-year review of data from their ALPR system. Some key highlights are included below, but the full report can be found in the Feb. 16 Agenda Packet: https://losaltoshillsca.portal.civicclerk.com/event/2742/files

Saratoga had 7 Flock cameras installed in August 2022. They also fall under SCCSO, and do not share data with any other agencies.

Mountain View is investigating ALPRs, but don’t have any proposal yet according to a call with their PD.

Palo Alto has had a single ALPR mounted on a patrol car for the past 10 years. They will be putting forth a proposal to their City Council on April 3rd for about 20 fixed ALPRs. They are a bit vague in their goals, but feel it would be a deterrence to crime in general and allow for better investigation after crimes have occurred.

Sunnyvale does not have any fixed ALPRs, although they have some on newer PD vehicles.

There is a private Flock system that has been in operation in Los Altos with 2 cameras for the past 8 months. This has cleared 0 cases so far.

Conclusion: Contrary to Flock’s claim, some surrounding communities still are not using ALPRs. LAH, which had the highest number of Flocks, has had 0 burglary cases cleared by ALPRs. Mountain View does not have ALPRs, but was not listed in the staff report as one of the three cities in Santa Clara who do not have them.

What is our position?

LARE does not support the proposed 1 year trial of 20 Flock ALPR cameras. Instead we support hiring a detective to increase clearance of property crimes.

We do not have any data that shows that ALPRs are an effective tool for property crime clearance. The system and its operation will be expensive, and it does not seem prudent to spend so much money on unproven systems – especially when there are clear priorities, like mental health responders and criminal analysts, that are currently unfunded.

Beyond these fundamental concerns, there are a number of concerning practical matters to the proposal as of 2/28/2023.

The endorsement letter provided by Chief McCully of Atherton PD for the 2/28 City Council meeting does not have a single data point, and instead gives vague reassurances of ALPR effectiveness. We would like to see the metrics that were used to measure the effectiveness of ALPRs in Atherton.
We also do not believe that the submitted policy for 2/28 is sufficient to prevent misuse of the system nor does it have sufficient oversight. It is a standard template policy created by Lexipol.
- We were only supposed to subscribe to the Stolen Vehicle System (SVS), Amber Alert and Silver Alert hotlists. As written in the policy, the hotlists are at the discretion of the ALPR administrator. This is not acceptable. City Council should have final say and the public should have input on which hotlists we are subscribed to.
- ALPR data should only be stored for 30 days unless it is part of an ongoing investigation or required by law. This should apply to downloaded data as well. The proposed policy states that downloaded data can be stored for up to 1 year.
- The policy leaves too many loopholes through the use of “exigent circumstances” or “when practicable” that render specific policy points unenforceable, including the requirement to doublecheck license plates manually before taking action, which would help mitigate against wrongful action by the police.
- This policy is claimed to have been vetted by the ACLU. Upon digging deeper, we have found no evidence of that. Alameda actually does have a policy based on ACLU recommendations, which looks quite different than the proposed policy.
A 3rd party Independent Police Auditor (IPA) should conduct a yearly ALPR audit, by taking a sampling of the search data, downloaded data, and privacy procedures to verify adherence to policy and prevent misuse. This would free up the PD’s valuable time. There are some PDs already utilizing IPAs for ALPR audits, including Menlo Park, Shakopee (MN), and Austin (TX). There are many PDs utilizing IPAs for general purpose police audits, including Palo Alto.

Conclusion: We do not feel that the benefits of this ALPR proposal outweigh the costs - both fiscal and civil liberties - hence we can not support it. If ALPRs were to be implemented in Los Altos, our proposed policy does not safeguard the public against misuse nor have enough accountability.