Does ShotSpotter Even Work?
Does ShotSpotter Even Work?
Is the system of microphones that Cambridge is installing effective in detecting gunshots?
Cambridge is installing ShotSpotter, a federally-funded system of microphones intended to detect gunshots. Part 1 of this series examined how ShotSpotter's lobbying and grant-writing emphasis leads to a deployment of their system in a city that has virtually no gun crime. Part 2, below, asks if ShotSpotter even works.
Popular culture is full of images of magical technology solving crimes. ShotSpotter evokes those images with publicity photos of its "Incident Review Center" with banks of monitors in a dimly lit room, conjuring a technical solution that will lead to reduced violence and solve crimes. The reality is a little more complicated.
Take Suffolk County, NY's experience. Suffolk's ShotSpotter system suffered so many false alarms that the company had to adjust its algorithms to be more discriminating. Even after the adjustment, a report (pdf) by Suffolk County police to the County legislature stated that, over an eight month period, only 6.5% of ShotSpotter's 212 activations could be confirmed as an actual gunshot. Over 30% were confirmed as false alarms. The remaining events remain undetermined.
Suffolk County was not alone in finding that ShotSpotter did not live up to its promises. Trenton, New Jersey has had a checkered history with ShotSpotter. A ShotSpotter system was installed in 2009. In 2012, the city rejected a federally funded $300,000 expansion after the system failed to detect a fatal gunshot and, according to ShotSpotter themselves, resulted in over a thousand false reports in 2011. Recently, with promises that the technology has improved, Trenton chose to implement a ShotSpotter expansion. Despite the use of ShotSpotter, Trenton's murder rate is "surging." In Newark, NJ, after promises by then Mayor Corey Booker that ShotSpotter would solve crimes, the police note that "when [they] get there, usually the perp is not in the area." This has led to calls for more surveillance cameras so that, along with the sound of the gunshots, there would be images, as well. That is how it is being used in some parts of Boston, with ShotSpotter directing surveillance cameras to turn towards the location of possible gunfire.
Troy NY's police department had enough of ShotSpotter, discontinuing its use at the end of 2012. The system was missing actual gunshots and falsely reporting other noises as shots. According to The Troy Record, when Troy police were polled at roll call about ShotSpotter, they were unanimous in calling it ineffective. Troy will sell their system as surplus.
There have been no fully independent studies of ShotSpotter and its effectiveness. In 2011, ShotSpotter commissioned CSG Analysis to perform an evaluation (PDF) of ShotSpotter. The study, paid for by ShotSpotter but, according to the analysts, conducted independently, was too small to allow for quantitative analysis. Instead, the analysts looked for "themes." A prominent theme was that "false system activations" drove police dispatchers "bananas." There were so many false alarms that police officers stopped keeping track of them, rendering any statistical data from their systems meaningless.
To their credit, the Cambridge Police have downplayed the effectiveness of ShotSpotter, referring to it as just one of the many tools they'll now have at their disposal. But in other jurisdictions, when police or city officials talk of reduced response time or better tactical awareness, it is important to remember that there's simply no objective evidence to support those assertions.
Perhaps the best measure of confidence in the system comes from ShotSpotter themselves, in the form of a clause in its contract with the City of Boston. In the words of the contract:
Any and all warranties, express or implied, of
fitness for high risk purposes requiring fail-safe
performance are hereby expressly disclaimed.
The message can't be more clear: when officers are at risk, don't depend on ShotSpotter.
This work by Saul Tannenbaum is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.