Originally published in the Washington Post on 8/21/2018
What’s the real value of a college education? For some people, that is calculated by the earning power that comes with a particular college degree. For others, it’s more about the satisfaction of a particular career, or about the social skills or intellectual rewards gained from the college experience. In every case, it is a life-altering decision with little data to inform it. Less data, in fact, than many other types of major long-term expenses and commitments, such as purchasing a home or buying a car.
Prospective students and their parents deserve a better look at what they are signing on for, and they could get exactly that with either of two bipartisan bills in the Senate. The Wyden-Rubio Student Right to Know Before You Go Act and the College Transparency Act would require colleges to collect and share data about such things as college costs, post-graduation salaries, the percentage of students that go on to graduate school, and how long it takes to graduate. The initiatives would give prospective students a great deal of information that could help many decide what college to attend.
No one thinks that’s a bad idea, but many people—privacy advocates, congressional Republicans, and private colleges—point out that the collection and dissemination of such student-level data could violate a ban written into the 2008 reauthorization of the Higher Education Act. Consequently, the bills have been stalled in the Senate, caught in the tension between two constituencies that mean well. One wants to improve consumer awareness and one wants to protect the privacy of individuals in an age when there isn’t much privacy left. It’s an increasingly familiar conflict between privacy and the public good, and because of a little-used cryptographic technology that has been used to solve similar clashes, it need not exist.
Two years ago, well-meaning people in Boston wanted to know the pay disparity between women and men. The Boston Women’s Workforce Council recruited 69 employers, all of whom believed that publishing the difference in wages could help redress a long-standing social injustice, but none of whom wanted anyone else to see their payroll data.
The effort, trapped in the same kind of limbo as the Senate bills, found a way out because of a complex algorithm conceived by theoretical mathematicians 30 years ago. Called secure multiparty computation, it could be used to answer questions about data from many sources without exposing data from one source to another.
Using software developed by my colleagues at Boston University’s Hariri Institute for Computing and Computational Science & Engineering, the Boston companies submitted encrypted data to a web browser, at which point it was randomly subdivided, then reconstituted with keys that only work when used collectively by all parties. No party ever glimpsed data from another, but everyone learned of the disparity in pay: Women at the Boston companies were paid 77 cents for every dollar paid to men.
It was bad news, but its consequences were good. Emboldened by the knowledge that the technique could guarantee the privacy of data, more companies joined subsequent efforts, and in 2017, a follow-up study included 166,705 employees in 114 companies constituting 16 percent of the Boston area workforce. That study found that white women earned only 75 cents for every dollar paid to a white man, and Asian women, black women, and Hispanic women made 71, 52, and 49 cents respectively for every dollar made by a white man.
Boston University researchers are helping the Greater Boston Chamber of Commerce learn how often employers subcontract work to minority-owned businesses, and they are enabling a nonprofit online sexual assault–reporting platform to identify repeat perpetrators. The same technology has been put to other uses. It facilitates trades of electric power, credit ratings, location services, and email encryption, enabling those applications to get beyond inevitable conflicts between privacy and public good.
It can do the same for the legislative efforts to tell students what they need to know. In fact, the most recent version of the Wyden-Rubio bill includes a requirement that the collectors of data use secure multiparty computation or another technology that offers greater privacy and security. That requirement has already pushed the bill closer to approval: the National Association of Independent Colleges and Universities, which had opposed earlier versions, has expressed a willingness to talk about it. Other opponents should take note, because the technology that put us into the privacy versus public good predicament can get us out.
Deciding which college to attend is too important and too expensive a task to undertake with limited information about its likely outcome. Consumers and higher education leaders should support legislation that can give students and their parents greater confidence they are making the right decision. Otherwise, students and families will be left in the dark on one of the biggest financial and personal decisions of their lives.
Azer Bestavros is the founding director of Boston University’s Hariri Institute for Computing. He is a researcher and professor in the Computer Science Department in the College of Arts and Sciences.