MIT Professor: Don't Fall For These Big Data Booby Traps

During the second and final keynote address of the Santa Clara, Calif.-hosted event, Crawford, who is also a principal researcher at Microsoft, gave a room full of big data gurus a wake-up call by addressing some of the inherent shortcomings of big data analytics.

"Big data could represent a fundamental shift in the way we understand knowledge," Crawford told the crowd. "But there is a trap, which is that sometimes, when we are working with big data, we get the impression that we can see the world from a 30,000-foot view with total clarity and objectivity."

[Related: 10 Big Data Startups You Shouldn't Overlook ]

Crawford outlined a number of real-world examples to prove how this sense of objectivity can be false, emphasizing that, while big data may give the illusion of being all-inclusive, there are often critical subsets of data hiding "in its shadows."

Sponsored post

Crawford used a recent project conducted by the city of Boston as an example. The initiative, dubbed "Street Bump," is meant to leverage data from drivers' smartphones to help detect where potholes are on the city's roads. Drivers place their smartphones somewhere in their car's interior -- on the dashboard, a seat, or in a cup holder -- and when a pothole-induced "bump" is detected, that data is sent to the city, so the pothole can be identified and fixed.

While great in theory, the project has one major flaw, Crawford said: It only captures data from parts of the city where there is a high population of smartphones. And as smartphone usage is predominantly higher in the wealthier parts of the city, the lower-income areas of Boston are somewhat left in the dust. What's more, areas of Boston with large elderly populations -- which are also less likely to own smartphones -- are left to fend off those pesky potholes on their own.

"So if you think about how this might be used to fix roads, we might see a future where the wealthy areas with young people get more attention and resources, unlike the areas with older citizens, who might get fewer resources," Crawford said. "So if you're off the map, this could have some really material consequences for social inequity."

Crawford didn't completely discount the benefits of big data; the analytical implications of the technology are huge, she said, and will no doubt push the boundaries of research. But she urged the audience to be weary of that false sense of objectivity big data can bring, and to balance big data with traditional computing and analytics methods as well, in order to draw more accurate conclusions.

"One of the ways to move forward and address some of the weaknesses I've talked about today with big data, is thinking about how we might bring big data together with small data -- computational social science along with traditional qualitative methods," Crawford said. "We know data insights can be found at all levels, and by bringing together a range of tools we get a much more three-dimensional image of what we are looking at and the questions we're trying to ask."