Advertisement

‘The Marriage Pact’ and the risks we take with data

I teach machine learning at Duke University. Scientists in my field are increasingly concerned about the ethics of data and its uses. So, when my students told me about a new dating service making the rounds of college campuses, I realized that something was terribly wrong.

This new dating service is called The Marriage Pact, and it has the potential to ruin the futures of college students across the country. The matchmaker app originated at Stanford University and has spread swiftly on many college campuses. Large fractions of the undergraduate populations (often over half the students or more) at Stanford, Northwestern, Cornell, UVA, Columbia, Wesleyan, Tufts, Yale and Duke have reportedly filled out its survey in hopes of finding the perfect match for life. The database owned by the Marriage Pact is unquestionably huge, and it’s risky for anyone in it.

The Marriage Pact questionnaire gets pretty personal. It asks about gender identity, sexual orientation, political affiliation, whether it’s OK if your spouse does softer drugs, or harder drugs, how long you wait to have sex when you start seeing someone, whether you prefer kinky sex, whether you often wish you were someone else, whether your looks are the most important part of your identity, and whether expensive dates would be more fun. The Marriage Pact stores all that information with each student’s name and contact information. And they aren’t the only game in town. It has rival called “DataMatch” that was founded at Harvard with the same idea.

As of Feb. 2, more than 4,200 Duke students, including 1,967 straight men, 2,075 straight women, 144 gay men, 65 gay women, 53 bisexual men and 260 bisexual women filled out the survey, according to a recent news article in the Duke Chronicle.

My students, who have come of age in an era when it’s normal to share every last bit of personal information, are buzzing about this new matchmaking service. I fear they — and college students across the country — don’t see the risks.

The algorithms used by the Marriage Pact are very simple — they just match each person’s record to the closest other record, according to a weighted sum of agreement on the questions. It’s the data that makes this setup dangerous, not the algorithms.

The Marriage Pact’s website assures users repeatedly that their personal information is not for sale. That’s fine, but what if the database becomes a target for hackers? At least tens of thousands of college students have contributed personal information to this highly organized database. The recent hacks of Equifax and SolarWinds illustrate that no company’s data security is ironclad.

What is the true value of these data? I would argue that for hackers, they are possibly worth more than healthcare and financial data combined. How much money could hackers extort from young people to keep data about their sex lives, sexual preferences, and drug abuse safe? When two Duke students asked the Marriage Pact how they store their data, the response was reportedly just “[W]e do our best to apply the most rigorous industry-standard practices around data handling,” without mentioning whether their practices actually meet the industry standard for cybersecurity practices.

I realize that at this point, we all freely volunteer personal information in myriad ways — from our clothing preferences while shopping online to our literal location tracked by our smartphones — and we should always be wary of these disclosures. But college students seem particularly unfazed when doling out their personal information. In this case, it could bring potentially serious consequences.

By the way, my students who used the service said the results were lousy. Not a great return for handing your secrets to a stranger with no guarantee they will be protected.

Cynthia Rudin is a professor of computer science, electrical and computer engineering, and statistical science at Duke University.