Substack

Monday, December 20, 2021

Limits to technologies in public policy - policing edition

This blog has written to exercise caution in our expectations from the application of digital and other technologies in the public policy realm. This, this, this, this, and this are posts on the issue. This post is about the use of technology in policing to spot gun shots and prevent crimes. 

The Economist has an article assessing the impact of the microphones-based gunshot detection technology used by ShotSpotter which, since its inception 25 years back in Chicago, is now being in over 100 cities across US. For a technology which has been around long enough to have matured, the article paints a damning picture,

According to the Chicago Office of the Inspector General (OIG), a city watchdog, of just over 50,000 alerts generated by the system between January 2020 and May 2021, only about 2% led to the police stopping somebody. Only 0.4% of alerts resulted in arrests and even fewer the recovery of weapons. A separate study by the MacArthur Justice Centre, a civil-rights law firm affiliated with Northwestern University, found that in just 10% of cases generated by the system could police find evidence of a shooting, such as spent cartridge cases or bullet holes... But there is little independent evidence that it reduces crime overall. One study, published in April in the Journal of Urban Health, found that “implementing ShotSpotter technology has no significant impact on firearm-related homicides or arrest outcomes”. Microphones on street-lights are all very well, but they are no substitute for information gathered from humans.

Much the same could be said about the use of body cameras to deter aggressive behaviours by policemen. 

The point is that despite their irresistible appeal, such fancy technology solutions, with their logical neatness, are no substitute for human intelligence and other regular information collection systems. Worse still, such technologies engender two practical problems which might worsen the situation. 

One, there is a moral hazard from the introduction of such high profile technologies. It introduces a sense of complacency and slacking on other routine gun shot prevention measures within the system as a whole. This is amplified by the general sales pitch of the leadership which introduced the technology as something which would solve the persistent problem. 

Second, implementation of such new technologies take up significant administrative and supervisory bandwidth of officials. In capacity constrained systems, this displaces scarce administrative and supervisory bandwidth away from the arguably more important task of focusing on the routine policing measures to prevent gun shot incidents. 

These trends are likely to be more pronounced in weaker state capacity environments of developing country cities. 

Application of technologies have resonance in other areas like use of smart meters in billing electricity and water, Supervisory Control and Data Acquisition (SCADA) systems in managing water and electricity supply, GIS mapping in utilities network management and property tax assessments, and biometric devices and facial recognition in attendance monitoring in schools and hospitals. In each of these areas, the outcomes have proved far from satisfactory, even verging on failure. 

This is not to reject the use of technologies in these areas. It's about figuring out where and how to use them, based on the context. 

So, smart meters are great at monitoring supply on sub-station feeders and for high value high voltage consumers. SCADA systems should eschew supervisory control and be replaced with simple Data Acquisition Systems (DAS) that can provide information as decision-support. GIS mapping is useful in upstream network management. Simple software applications that can be deployed on smartphones of teachers and health staff are far more likely to bear results in attendance tracking than dedicated biometric devices or stand-alone cameras.

These are all cheaper second-best uses of the respective technologies. But they stand far better chance of success than their costlier first-best versions. 

There is a political economy to these technology choices. They serve the interests of both politicians and bureaucrats who can present them as solutions to persistent problems, and thereby both fend off criticism and also get acclaim from opinion makers. Also, the design of these choices are most often made at the behest of consultants, whose contextual understanding and practical knowledge is limited. More importantly, they are an integral part of the larger coalition that promotes these technologies. 

No comments: