Fact-Checking WPD Chief and Deputy Chief on ShotSpotter Connect

On Friday, February 12th, 2021, WPD Chief Steve Sargent and Deputy Chief Paul Saucier discussed ShotSpotter Connect on Hank Stolz’s “Talk of the Commonwealth’’ radio show. Throughout the interview, both members of the WPD misrepresented, misinterpreted, and oversimplified the facts surrounding ShotSpotter Connect and predictive policing. We have decided to fact-check the WPD’s claims about ShotSpotter Connect because the City Council as a whole has done very little fact-finding on its own. One City Councilor even blamed community groups for spreading misinformation.1 To dispel this notion, we are providing Worcester residents and officials with more research so they can have more informed conversations. All facts provided by DefundWPD are cited and linked in the footnotes.

1.

The Claim: ShotSpotter Connect predicts future locations of crime. (Chief Sargent, 1:09 - 1:28)

The Facts: There is no evidence suggesting that ShotSpotter Connect, nor any other predictive policing tool, will predict where crimes will occur. According to a product FAQ page released by ShotSpotter, the company does claim that Connect is able to “detect areas at highest risk for crime” and “ensure patrols are placed at the right place before crime occurs to more effectively deter crime.”2 However, there is no publicly available source supporting these claims, nor one detailing the accuracy of ShotSpotter Connect. This trend is not unique to Connect: one publication from the group “Upturn”3 found that several predictive policing companies failed to publicly share how they measure accuracy of their products, including Connect’s predecessor, HunchLab4.

2.

The Claim: ShotSpotter Connect provides real-time crime information. (Chief Sargent, 2:17 - 3:32)

The Facts: Sargent argues ShotSpotter Connect is a valuable tool because it offers data “in real time,” allowing officers to address crime proactively rather than retroactively. However, this claim works off assumptions that are not necessarily true. Firstly, the “real-time data” provided to officers are data-driven guesses about where crime may occur, not where it will occur or are actively occurring. As Deputy Chief Saucier describes it later in the interview, “[Connect] alerts commanders every shift – not every few days – of the probability of where a risk of a crime is occurring” (9:06). The lack of certainty may be in part because crime, as with any human behavior, can be difficult to predict. In fact, the ongoing debate is so inconclusive, two5 separate6 articles from the same author in the same publication offer two different findings. Secondly, Sargent’s statement implies that if an officer is able to reach a potential crime scene before the crime occurs they will prevent the crime from occurring altogether. This is a faulty premise, as described in item 7 below.

3.

The Claim: The Worcester community wants ShotSpotter Connect. (Chief Sargent, 3:46 - 3:58)

The Facts: Sargent claims that the community “is always asking” for more police in neighborhoods. While there are certainly community members who make that request, many are members of our city’s crime watch groups whose demographics have been shown to not represent Worcester’s diversity. There are many others who request the opposite, or request very specific forms of police intervention such as foot patrol, such as residents in Councilor Rivera’s district. Unfortunately, Connect is a “city-wide initiative” according to Deputy Chief Saucier. For any residents who are not comfortable with this program being implemented (and, based on public comments received during the January 19th City Council Meeting and the February 1st Public Safety Subcommittee Meeting, there are many residents who are not), their concerns will not be taken into consideration if the Connect program is adopted. It is also worth noting that many members of local crime watch groups were sent misinformation about ShotSpotter, and when called upon during Council meetings noted that they did not feel prepared to speak when confronted with the additional information provided by members of the public who had already spoken.

4.

The Claim: The opioid epidemic, homelessness, and COVID have made this a very trying year. ShotSpotter Connect is a way to make our community safer and better. (Chief Sargent, 6:26 - 6:41)

The Facts: The opioid epidemic, COVID-19, and people facing loss of housing are all examples of the ongoing crises in Worcester that deserve our attention and require resources. However, ShotSpotter Connect will not provide those resources, and would cost the city $63,750 this year and nearly $50,000 every subsequent year. If one of Sargent’s top priorities is to find safer community practices in order to address the ongoing struggles he’s listed here, he would at the very least find the funding within the existing $53 million dollar police budget rather than redirecting more dollars away from other city departments and services.

5.

The Claim: ShotSpotter Connect is a successful program. (Chief Sargent, 6:44 - 6:51, Deputy Chief Saucier, 11:33 - 12:05)

The Facts: There is little publicly available information to support the claim that ShotSpotter is “very successful.” As previously noted, there is no available evidence regarding the accuracy of ShotSpotter Connect, or how effective it has been at reducing crime. This is at least in part due to the fact that the Connect program is still in its early stages with too few cities implementing the program for the company to offer a report of the success rates. Of the 5 cities Chief Sargent listed in his report on ShotSpotter to the City Manager7, not one has publicly stated that Connect definitively led to a reduction in crime. Similar programs with longer histories, such as IBM’s Crime Insight and Prevention software, have also struggled to produce quantifiable accuracy and effectiveness rates8. Saying this program is “very good” and “very successful” is simply conjecture. The WPD has been upholding Wilmington, NC as a model city for its adoption of Connect. According to a local article, Wilmington, NC implemented ShotSpotter Connect in 2019. The following year, the city saw a 120% increase in homicides9. This highlights the possibility that crime may be decreased in areas officers are sent to by the Connect algorithm, but overall crime rates in a city can remain the same or even increase.

6.

The Claim: ShotSpotter Connect uses objective data sources. (Deputy Chief Saucier, 8:52 - 9:03)

The Facts: The use of objective data such as “weather patterns, time of day, and whether or not school is in session” does not negate the additional use of crime data that, historically, has been skewed. This is particularly true when we do not know how the Connect algorithm weighs different variables when calculating a risk score or “forecast.” Predictive policing programs of any kind have the potential to exacerbate existing racial biases present in any city or police department. WPD misses the fundamental point of why so many community organizations seek reform: it’s because over decades the WPD has proven that its crime data cannot be trusted. From falsified arrest reports10 to civil rights abuses11 to racially motivated attacks and arrests12, WPD’s historic crime data is tainted with millions of dollars worth of lawsuits13 14, Justice Department audits, opacity, and unaccountability15. In short, the data they want to feed into this algorithm cannot be trusted to be objective.

Using historical crime data that is overly representative of neighborhoods that are predominately Black, brown, or low-income neighborhoods will continue to perpetuate a cycle. The Connect algorithm will direct officers to these neighborhoods, because it’s given crime data that overly represents these neighborhoods. Officers will continue to be more present, and are thus more likely to witness misdemeanor activity in these neighborhoods versus neighborhoods they’re less likely to be directed towards (i.e. affluent, predominantly white neighborhoods). Chief Sargent may claim that the goal is not to make arrests, but that does not mean no arrests will be made.

7.

The Claim: Additional police presence deters crime. (Deputy Chief Saucier, 9:32 - 9:56)

The Facts: It has not been proven that police presence deters crime. There have been studies that indicate increased police presence may help deter crime in a particular area, though the type of patrol matters16 17. This is also a claim touted on ShotSpotter’s website18. However, the hypothesis that visible police presence deters crime is a highly contested one that has been and continues to be heavily researched. Several studies indicate that visible police presence has little impact on crime rates in an area19 20.

Further, prevention and deterrence may be similar but are not synonymous. Even in situations where crime in one area may be deterred by police presence, that does not mean it has been prevented altogether. One study found that, although the presence of foot-patrol was beneficial in reducing crime immediately, the effects of patrol diminished rapidly after officers left21. If WPD officers were to spend 15 minutes in a “hot spot” as the ShotSpotter algorithm recommends, research suggests any benefits from this would dissipate after they left.

8.

The Claim: Upgrading ShotSpotter to ShotSpotter Connect is worth the money, because Worcester’s current ShotSpotter Flex program contributed to getting guns off the streets. (Deputy Chief Saucier, 12:57 - 13:31)

The Facts: Saucier compares Connect to ShotSpotter’s acoustic gunshot detection program, ShotSpotter Flex. He conflates the two programs and seemingly uses the success of Flex to explain the need for Connect, posing the rhetorical question “what does that cost the community?” This is an emotional appeal that is misleading to listeners. Connect did not play a role in getting guns off the streets – Saucier’s example occurred due to ShotSpotter Flex alone. Further, Worcester isn’t required to get Connect in order to expand the Flex program. The city would receive a discount on a mile of the gun detection program, but after accounting for the cost of Connect, the savings is only $16,25022. If the amount of savings that comes with the bundle deal is crucial, the funding for this program can come out of the $53 million WPD budget, rather than the City Manager’s emergency contingency fund originally planned for saving jobs and providing raises to city employees.

9.

The Claim: ShotSpotter Connect will not lead to over-policing. (Deputy Chief Saucier, 17:48 - 18:27)

The Facts: The term “over-policing” has been used in several ways in the literature on policing and predictive policing, and doesn’t necessarily mean “a hundred officers” in one neighborhood, as Deputy Chief Saucier puts it. Over-policing can be a statistical term that means police are in one area or interacting with a subcommunity disproportionately. It is possible for only one or a small number of officers to engage in the over-policing of a neighborhood if they are spending more time in that neighborhood than they are in other areas of the city, particularly if those areas are already over-represented in historical police data23. Even if the intention is not to profile select communities, if police presence is determined by the Connect algorithm after it has been given biased crime data, officers will still be over-policing certain areas.

10.

The Claim: Using location-based data in predictive policing is not harmful, and community groups concerned over ShotSpotter don’t understand the difference between location-based data and person-based data. (Chief Sargent, 6:51 - 7:03, Deputy Chief Saucier, 10:07 - 10:21, 18:28 - 18:51)

The Facts: There is a difference between person-based and location-based data, and ShotSpotter connect is the latter. However, despite Sargent, Saucier, and ShotSpotter’s attempts to discredit our criticisms and shy away from the term, location-based data algorithms are still a form of predictive policing. Predictive policing has been defined as “the use of historical data to create a spatiotemporal forecast of areas of criminality or crime hotspots that will be the basis for police resource allocation decisions with the expectation that having officers at the proposed place and time will deter or detect criminal activity.”24 This is precisely how Deputy Chief Saucier has described ShotSpotter Connect.

There are risks in using person-based data programs like PredPol, and one could argue they are perhaps more severe than the risks of using location-based data programs – no community group has claimed anything to the contrary. However, the danger of a program like PredPol does not negate the dangers of other programs and of predictive policing in general. Concerns associated with location-based data programs like Connect include: the fact that neighborhoods are often segregated25, including in Worcester26 27, that crime data is often over-representative of Black, brown, and low-income individuals28 in part because what is considered crime is racially biased29, that there is a lack of empirical evidence to support the effectiveness and refute the potential harms of Connect, the fact that crime may be unpredictable, and concerns of implementing a new program before we as a city fully understand the risks. In 2020, Judge Stephanie Thacker on the Fourth Circuit Court of Appeals found that “‘hot spot policing,’ which utilizes historical crime data to predict future crime hot spots, can be infected with years of racial bias…Predictive policing will inherently reinforce and perpetrate this structural racism.” She further criticized predictive policing technologies as “little more than racial profiling writ large.”30 Hot spot policing has also received heavy scrutiny by organizations like the RAND Corporation31.

There is a robust body of literature and public discourse to support residents and community groups’ concerns about adopting ShotSpotter Connect. We encourage city officials and WPD officers to engage with that body of work in-depth, rather than shoehorn unproven technology into our communities.


  1. Facebook comment ↩︎

  2. ShotSpotter Connect Product FAQ, p. 1 ↩︎

  3. Predictive Policing Systems: how they work, the data they use, and who uses them ↩︎

  4. ShotSpotter announces acquisition of HunchLab to springboard into AI-driven analysis and predictive policing ↩︎

  5. Computers may not improve predictions of repeat criminals ↩︎

  6. AI bests humans at prediction repeat offenders among criminals ↩︎

  7. Chief Sargent’s Letter to City Manager Augustus, January 6th, 2021, p. 1 ↩︎

  8. IBM SPSS Crime Prediction and Prevention ↩︎

  9. Wilmington homicides more than doubled in 2020? What fueled the increase? ↩︎

  10. Video of Beer Garden arrest at odds with officer’s account ↩︎

  11. ACLU Massachusetts: Cocroft v. Smith ↩︎

  12. 62-year-old grandmother sues Worcester Police Officer James Powers, city; claims racially motivated ‘police brutality’ ↩︎

  13. City actions best defense in lawsuit ↩︎

  14. Two Black former Worcester police officers win another legal battle in decades-old discrimination case that could cost city millions ↩︎

  15. Watchdog Report: Lawyer calls for probe of Worcester police credibility ↩︎

  16. The Philadelphia Foot Patrol Experiment: A Randomized Controlled Trial of Police Patrol Effectiveness in Violent Crime Hotspots ↩︎

  17. PROTOCOL: Increased Police Patrol Presence Effects on Crime and Disorder ↩︎

  18. ShotSpotter Homepage ↩︎

  19. More cops don’t mean less crime, law enforcement experts say ↩︎

  20. Police Officer Staffing: Analyzing the Commonly Held Belief that More Cops Equals Less Crime ↩︎

  21. Foot Patrol in Violent Crime Hot Spots: The Longitudinal Impact of Deterrence and Posttreatment Effects of Displacement ↩︎

  22. Chief Sargent’s Letter to City Manager Augustus, January 6th, 2021, p. 3 ↩︎

  23. To predict and serve?, p. 18 ↩︎

  24. Proactive Policing: Effects on Crime and Communities (2018), p. 129 ↩︎

  25. Metro areas are still racially segregated ↩︎

  26. Mapping Segregation ↩︎

  27. 2010 Census Tracts for Worcester, Massachusetts ↩︎

  28. An Unjust Burden: The Disparate Treatment of Black Americans in the Criminal Justice System ↩︎

  29. Proactive Policing: Effects on Crime and and Communities (2018), p. 251-251 ↩︎

  30. United States of America vs. Billy Curry, Jr. ↩︎

  31. Predictive Policing: The Role of Crime Forecasting in Law Enforcement Operations (2013), p. 20 ↩︎