Big Data and the Underground Railroad

n the fall of 1769, Thomas Jefferson lost a slave. His name was Sandy, and he was a runaway. Sandy was “about 35 years of age.” He worked as a shoemaker. Jefferson described him as “artful and knavish.” He was also “something of a horse jockey.”

Jefferson criticized slavery. Yet when he signed the Declaration of Independence in 1776, Jefferson owned almost 200 human beings. When Sandy went missing, he owned about 20; losing even one was significant. So Jefferson used the best available technology to find Sandy: the newspaper ad.

Sandy was caught and later sold for 100 pounds. Around the turn of the century, however, things slowly started to change. A secret network was built to help people like Sandy. Over time, tens of thousands of runaway slaves would escape bondage on the Underground Railroad.

How many of them would have made it in the age of big data?

There is a booming debate around what big data means for vulnerable communities. Industry groups argue, in good faith, that it will be a tool for empowering the disadvantaged. Others are skeptical. Algorithms have learned that workers with longer commutes quit their jobs sooner. Is it fair to turn away job applicants with long commutes if that disproportionately hurts blacks and Latinos? Is it legal for a company to assign you a credit score based on the creditworthiness of your neighbors? Are big data algorithms as neutral and accurate as they seem—and if they’re not, are our discrimination laws up to the challenge? ...