by
Panelists:
Elizabeth Joh, UC Davis School of Law Professor
Thomas Abt, Harvard Kennedy School Adjunct Lecturer and Program in Criminal Justice Senior Research Fellow
Vivek Krishnamurthy, Harvard Law School Cyberlaw Clinical Instructor
By Ana Choi
On Monday, the Harvard Law and Policy Review kicked off its annual symposium featuring the articles in our forthcoming Volume 10.1, with this year’s theme being “Policing in America 50 Years After Miranda.” The opening panel featured Elizabeth Joh, who spoke about her article The New Surveillance Discretion: Automated Suspicion, Big Data, & Policing, and Thomas Abt and Vivek Krishnamurthy, who provided responses to the article.
Prof. Joh started out by discussing why we should be paying more attention to big data in policing. Big data is drastically changing the nature of police surveillance discretion by increasing surveillance power and lowering costs, and yet the law is not paying much attention to this development. With big data policing on the cusp of wider adoption, there are important questions of accountability and transparency that must be considered. Prof. Joh then went on to highlight three specific examples of big data being used in policing. First, she talked about automatic license plate reader (ALPR) technology and gave the example of the Los Angeles City Council’s recent proposal to scan the license plates of cars driving in neighborhoods with high prostitution activity and mailing “Dear John” letters to the houses of these car owners. Second, she discussed social media analysis and how the police department of Fresno, CA is using software to identify red flags in a person’s online activity and calculate their ‘threat score.’ Third, Prof. Joh turned to social network analysis, with the most prominent example being the Chicago Police Department’s creation of a ‘heat list’; this is a list of people who have been determined to be prone to violence—either as a perpetrator or a victim—based on who their acquaintances are.
After Prof. Joh’s overview of her article, Mr. Abt gave a response. The first point he made was that there is often a disconnect between what people see on the outside and what people see on the inside, in terms of police capabilities. He observed that the added burdens of counterterrorism activity that police have been responsible for since 2001, combined with the fact that the recession took a huge toll on law enforcement budgets, have resulted in the police being asked to do more and more with less and less. In his view, the type of sophisticated and powerful techniques of big data policing discussed by Prof. Joh are largely theoretical or anecdotal at this point. Second, Mr. Abt asserted that even if it were true that the police currently have this kind of capabilities, that would not necessarily be a negative thing. He discussed how the Chicago PD’s ‘heat list’ is used not just to scare people, but rather to provide resources and prevent individuals from falling into trouble. Also, he offered the possibility of big data policing as a remedy to the problems of overpolicing. There is a lot of criticism of policing strategies such as New York’s stop-and-frisk policy because they impose heavy costs on innocent citizens for the sake of ferreting out the few actual criminals; Mr. Abt believes that with the effective use of data, we can solve this problem by enabling police to differentiate between harmless citizens and criminals ex-ante.
Mr. Krishnamurthy also gave a response, during which he analyzed the issue of big data and policing from several different angles. First, he noted that while big data technology often actually reduces police discretion by forcing police officers to justify their decisions on what the data shows, there are other potential risks we need to think about such as bad faith on the part of the programmer who is designing these systems, unintentional disparate impact on certain groups, and errors resulting from low quality data. Second, he discussed the possibility of big data policing leading to system-wide benefits in two ways: by pushing strategic decisions about policing to a higher level on the chain of command and thereby bringing more deliberation and scrutiny to these decisions, and by equalizing treatment of criminals so that people who commit the same types of crimes will be subject to the same types of consequences. Third, Mr. Krishnamurthy asserted that the ends being served by big data tools are something we need to think about and determine, not something that is a given; we tend to assume that big data is something that is solely used to facilitate police effectiveness, but we might also imagine a big data tool that is designed to force police to face their implicit biases. Finally, he highlighted the need to talk about privacy at a high level of law and policy: what are the permissible uses of all this data that is being produced, and what kinds of consequences should attach to impermissible uses?
Prof. Joh concluded the panel discussion by providing brief follow-ups to Mr. Abt and Mr. Krishnamurthy’s responses. With regard to the question posed by Mr. Abt—what is the actual harm caused by big data policing?—she focused on the issues of legitimacy and transparency. Even if surveillance technology is being use for good, productive ends, it’s still a problem if the public is not informed about what kind of information is being collected and how it can be used against them. Prof. Joh also acknowledged that there definitely are benefits to big data policing and that it’s true that big data can reduce police discretion at the ground level, but cautioned that there are still hidden sources of discretion—e.g. input data, how the algorithm is designed, etc.—that we need to think about. Prof. Joh’s final point was: the surveillance question is over in a sense; you cannot problematize it anymore. The real question going forward, then, is who has control over the data, how is it going to be shared, and how is it going to be accessed?