Daniel Castro is a senior analyst with the Information Technology and Innovation Foundation, director of the Center for Data Innovation and the author ofThe Rise of Data Poverty in America, which posits that policy-makers should begin a concerted effort to address “the social and economic inequalities that may result from a lack of collection or use of data about individuals or communities.” He and I were panelists last week at the Federal Trade Commission (FTC) Big Data Workshop. In this interview, I ask him for his impressions of the workshop and about his new paper.

Wolf: What do you think the evidence presented at the FTC workshop showed about discrimination and big data?  

Castro: Since the FTC workshop resulted from months of planning and provided a high-profile opportunity to highlight the most egregious uses of big data, I was expecting some fireworks. Instead, I barely saw a spark. While a number of panelists pointed to ways that individuals might use big data to discriminate against others, there were no concrete examples offered of where someone was actually using big data to directly harm consumers today. This observation tracks very closely to our review of the White House big data report, which found that of the 37 concerns mentioned in the report, all but two of them were purely speculative. Perhaps even more importantly, nobody at the FTC workshop (or, for that matter, in the White House big data report) presented an example of an individual suffering harm where there were no legal mechanisms in place to address the concern.

Wolf: Do you agree that big data can be used as a tool to identify and fight discrimination, as the FPF/ADL Report suggested?

Castro: Absolutely. Big data is an incredibly important tool to fight discrimination and promote equality, and we need more people to learn how to use this tool. In fact, one of the key takeaways from the FTC workshop was the need for government agencies and others working to combat discrimination to take a close look at how else they might use data to pursue this end. As the FPF/ADL report and others have shown, data is the key to identifying undesirable practices, such as inequalities in the workplace or unfair lending at financial institutions. Making more data available not only increases transparency, but it also promotes accountability. Using better data analytics, organizations can uncover both intentional and unintentional discriminatory behaviors and take steps to intervene. Data also unlocks the possibility of removing the human element from some decisions, which is often the cause of discriminatory behavior, and institutionalizing automated decisions that ensure equal treatment.

Wolf: Are you concerned that traditional application of the Fair Information Practice Principles (FIPPs) might actually deter the use of big data for positive uses, such as fighting public health problems or identifying/fighting discrimination?

Castro: Yes. The FIPPs were designed for locking up data, not for unleashing its potential, and many of the FIPPs are simply incompatible with innovation. For example, the principle “purpose specification” requires that organizations know how they will use data before they collect it. If organizations only use data to do with it what they have always done, then they are not innovating. Policy-makers need to remember that while the FIPPs are quite old, they were not carved in stone tablets and handed down from a mountain top. We need policies to keep up with technology, and we need a new conversation about creating 21^st-century principles for data sharing so that all these positive uses of big data do not get overlooked. Policy-makers should be leading the charge to tear down barriers to sharing data for beneficial purposes, whether it is to fight cancer or prevent racial profiling.

Wolf: Do you think the FTC workshop demonstrated any gaps in regulatory or legal protections against discrimination that might result from big data?

Castro: There do not appear to be major gaps in laws or regulations that would prevent existing protections, such as the laws enforced by the U.S. Equal Employment Opportunity Commission, from being applied to big data. For example, the Pregnancy Discrimination Act prohibits employers from discriminating on the basis of pregnancy. It does not matter whether employers learn of a pregnancy through water cooler gossip or predictive analytics; the law makes it clear that discriminatory actions are prohibited. Laws like this are particularly helpful to consumers because they provide technology-neutral guidance on what behaviors are permissible.

Wolf: You recently did a study on “data poverty.”  What was that about and what were your findings?

Castro: The Center for Data Innovation just released a report called The Rise of Data Poverty in America that shows how we are likely to see the emergence of a “data divide”; i.e., a gap between data-rich and data-poor communities. These gaps might even cluster geographically to create “data deserts,” or areas of the country characterized by a lack of access to high-quality data about individuals and communities. Given the increasing importance of data to decision-making by both government and the private sector, policy-makers should start thinking about how to address these gaps so as to ensure that all individuals share in the benefits of data-driven innovation.