Facing down employment discrimination with an algorithm

Getting a good job, getting promoted, getting a raise. These things aren’t easy. In many cases, it may be even harder if you’re not a straight, white, highly-educated man.

Vivienne Ming knows this. And at Gild she sought to correct the problem with the help of a massive data set and a proprietary algorithm.

Ming, who spoke with MIT Sloan students Oct. 13 as part of the school’s Innovative Leadership Series, is today the co-founder of Socos, where she has created a machine learning-driven app to help improve a child’s life outcomes.

At Gild from 2012 to 2014, she worked to remove bias from the hiring process. Her example for the audience: Jade Dominguez, who taught himself to program Ruby and began seeking a job in Silicon Valley.

“They give you five seconds,” Ming said of recruiters and hiring managers. “Your name, your school, and your last job. His last name was Dominguez. Let’s be blunt. That is not the right name to get a job in the United States. School: nothing. Last jobs: nothing. Into the shredder.”

Yet Gild’s algorithm — the company builds profiles of candidates that pull from up to 55,000 data points — determined Dominguez was the “second best Ruby developer in Los Angeles,” Ming said. The company needed a developer. It hired Dominguez.

Gild’s data-driven talent acquisition model, Ming said, seeks to answer the question most recruiters ask: “Are you the right person for this particular job?” But it attempts to do that in a way that both strips bias from the hiring process and makes use of more, better predictive information. Is a candidate creative, adaptive, a problem-solver, and good with personal relationships? What sort of work have they done?

Even in a business environment that claims to promote meritocracy, bias remains rampant. According to research by MIT Sloan professor Emilio Castilla, believing you have a culture of meritocracy can make matters even worse.

A tax on being different
In a 2014 experiment, José Zamora changed the name on his résumé to Joe. The offers started rolling in.

To Ming, Zamora’s experience revealed what she called a “tax on being different.” Work she did with Gild’s data showed that someone named José would need a Master’s degree or higher to qualify for the same Silicon Valley software engineering jobs that someone named Joe would qualify for with no education. That sort of discrimination has lasting effects throughout a person’s lifetime, as “discrimination comes with compound interest,” Ming said.

The result of such discrimination is reflected in lifetime earnings. The effective “tax” on gay men in the United Kingdom is about £54,000 pounds, Ming said. For women in technology jobs in the United States, it’s $100,000 to $300,000. For men named José: $750,000.

How do people pay that “tax?” To reach equal heights, people in disadvantaged groups must attend more prestigious schools, achieve higher degrees, and have more exceptional experience and a more eye-catching list of previous employers.

“Working extra hard to grow is great,” Ming said. “Working extra hard just to prove who you are? No. That’s crap.”

A life of substance
Ming left Gild in 2014. She holds nine different positions today, according to her LinkedIn profile. Her career is not defined by a single job title or even a single industry. Theoretical neuroscientist. Entrepreneur. Hacker. Executive. Advisor. Data scientist. They all apply.

For Ming, though, it’s not about titles or even about what exactly you do. It’s about why you do it. Her work is aimed at “maximizing human potential” in “tangible, measurable ways.” She spoke, as well, of her father’s charge to “live a life of substance” and of the need for creative and career rebirth.

As parting advice, Ming encouraged students to see themselves as living and dying many times over, each time starting their “next life,” the next phase of their career, one for which they will develop new skills for a new industry in which they will possibly start a new company.

“This is me right now. I’m kind of done with startups. I’m writing six books. Never written one before. I’ve been asked to run for office. All these sorts of things,” Ming said. “So I’m trying to figure out what actually is next. And that’s great. In fact, I would say, pick yourself at your peak. Die. Take it to the next level. Poetry and hoverbikes and physics and philosophy. These are your lifetimes and I beg everyone in this audience to use them.”