As most of us know by now, Republicans in congress recently made it legal for internet service providers (ISPs) to distribute and sell information about us, their customers – including detailed internet usage data – without our permission. Much has been written about the potential dystopian marketing uses of this data, but I have seen little addressing something that could be even more chilling: the use of this data by law enforcement.
There is a recent but deep history of companies providing new data-driven tools to law enforcement. It can be hard to argue with these services on the surface, but repeatedly we have seen abuses which have led some companies to curtail the use of their services by the police. Companies have harvested the public timeline of Twitter, for instance, to provide tools for tracking lawful protestors. Facebook data is available through third parties that allow a level of deep search impossible on the platform itself. And while these services and the use of this data by law enforcement does occasionally result in a feel-good story on the news, the broadening of the use of these tools is concerning.
Thinking specifically about ISP customer data, we know that there will be data of interest to law enforcement. It is, after all, not uncommon for subpoenas and warrants to be issued for this data in certain types of cases. The idea that detailed demographic data coupled with explicit internet usage history could soon be available on every internet subscriber in the country, packaged into a neat, searchable database, must have some members of the authorities salivating. No longer will a case need to be made first, possibly to be backed up with a handful of potentially dubious web searches. Instead, algorithms can easily digest the data as a whole and spit out a list of people who “fit the pattern” of past offenders – maybe this is a list of users visiting known child pornography sites, or maybe it’s a list of us who regularly use encrypted communications, read hardware hacking tutorials, and search for “download Avengers movie” to see which service offers it cheapest.
We have already seen how problematic algorithmic analysis of data sets can be. They are biased, opaque, and lack nuance in their conclusions. Set loose on such a rich data set, it’s not hard to imagine the long list of potential “criminals” that can be manufactured from completely innocent, legal uses of a service they pay for. No warrant will be needed for this data. No oversight required for its use. The data brokers need not even be publicly revealed if their terms of service are properly written. And as laws regarding what is and is not legal or suspicious change, some of this data could damn us well into the future.
Congress has not just handed the marketers of the country with a gift, they have stocked the pond for some very rich fishing expeditions by law enforcement groups throughout the US.