We know what automatic license plate readers are good for: collecting massive amounts (billions of records) of plate/location data housed by private companies and accessed by law enforcement for indefinite periods of time. What we don't know is how effective ALPRs are at fighting/investigating crime.
George Joseph at Citylab has done some digging into the effectiveness of license plate readers and hasn't found much that justifies the expense, much less the constant compilation of plate info.
Last month, the Bay Area’s UASI released ALPR data from the Central Marin Police Authority showing that only .02% of the nearly 4 million license plates tracked over October of 2015 through April of this year resulted in matches to any police “hot list” databases. The data indicate that zero “known or suspected terrorists” have been tracked using ALPRs, and that only a handful of other matches related to other hot-list criteria.
Why the mention of "terrorists?" Well, like most other high-tech law enforcement gear, the funding and deployment of these tools relies heavily on a narrative that never pans out: the neverending War on Terror. UASI stands for "Urban Areas Security Initiative" -- a DHS grant program meant to better equip law enforcement for handling terrorism/terrorists. To secure grants to pay for ALPRs, Stingrays, 1033 program supplies, etc., all law enforcement has to do is insert "because terrorism" somewhere in the requisition form. Existential angst -- and every government agency's natural desire to stay well-funded -- takes care of the rest.
But in reality, ALPRs aren't catching terrorists. They're not even catching dangerous criminals. Use of ALPRs has increased dramatically over the past decade, but there's not much to show for it other than millions upon millions of "non-hit" snapshots. Instead of bringing down terrorists, drug traffickers, auto theft rings, and kidnappers, ALPRs are being used to troll low-income neighborhoods. (Oakland, CA: Picture/data by Dave Maass, Jeremy Gillula, and EFF.)