In his essay, “Mining Student Data Could Save Lives”, Michael Morris argues that universities should implement data mining algorithms to detect patterns in student activities on their networks (e.g. any activity occurring on the WiFi network, school computers, or communications through university email accounts). According to Morris, the implementation of these algorithms could potentially prevent violent acts from occurring on campus; after all, almost every large-scale act of campus violence has been preceded by warning signs which, if recognized before the incident, would have indicated that an act of violence was imminent and could have been prevented.

Indeed, there is something compelling about Morris’ argument. A student who purchased high-powered firearms on the school network sending emails on a clearly detailing plans to perpetrate an act of violence  clearly warrants a breach of individual privacy to ensure the safety of the campus community. However, very few scenarios are this clear cut since most evidence would not be as damning as an explicit description of a violent act sent on a university email.

This raises the issue of false positives, one that is inherent in all data mining algorithms. In the article, Morris specifically cites the example of banks using data mining to detect stolen credit cards. And while these algorithms are good at detecting stolen cards, they are equally adept at generating false positives, deactivating cards after valid transactions that were deemed suspicious. Similarly, algorithms designed to monitor communications on university networks would need to be extremely sensitive even to small red flags in order to effectively prevent violent acts. However, designing the algorithm in this way would lead to false positives being regularly detected, incriminating students who had no violent intentions simply for their normal browsing activities and communications with others. If even one student is called in to “have a conversation” because of something the algorithm detected, it has already failed to do its job at the cost of individual freedom and privacy.

In principle, Morris’ idea is persuasive. The perfect data mining algorithm would be ideal for stopping campus violence without the need for extreme invasion of privacy or the generation of false positives. However, the implementation of a data mining algorithm in our complex world would require sacrificing students’ digital privacy for little to no benefit.