Breaking up Silicon Valley's white boy's club, one interview at a time

Stephanie Lampkin, a petite black woman, was once told during a job interview that her background wasn’t “technical” enough for software engineering jobs. She was told this despite a software engineering degree from Stanford University and stints working for Microsoft and Deloitte.

“So I made an app,” she quipped while presenting at an inclusive innovation showcase at the Massachusetts Institute of Technology last month.

Specifically, she created Blendoor, a platform that matches up companies and potential workers, Tinder-style, and scrubs the name and photo from a job candidate’s résumé. If employers who subscribe to the service like what they see otherwise, they can request more information or set up an interview. It’s geared at bolstering the chances of qualified minority candidates to break into homogeneous industries, inspired by numerous studies, like this one, showing that résumés with names like Jamal and Lakisha were far less likely to lead to in-person interviews than those with names like Emily or Greg.

Blendoor, which launched in beta at South by Southwest in March, and is being tested by the likes of Twitter, Google, and Airbnb, is part of a bumper crop of startups jockeying to be the go-to tool for making hiring more democratic.

They may tackle the problem in many different ways, but the operating principle is the same: Making structural tweaks to the screening process for job candidates is a more effective way of ensuring a diverse workforce. Placing the onus on even the most well-intentioned hiring managers to overcome their own ingrained biases or, even more likely, avoid falling back on already-established social connections to make a hire, is less so.

Whether or not an objective, skills-based hiring process can work for non-tech careers, or change the calculus for higher-level jobs that require more nebulous qualities like “leadership,” and “creativity” remains to be seen. But at the very least, supporters say, the approach goes a long way toward fixing the broken process that initially introduces employers to potential employees.

“I don’t know if we should get rid of them entirely, but résumés as a first-pass filter should be completely done away with,” Aline Lerner, creator of Interviewing.io, says.

'Blind auditions' for tech jobs

Virtually nonexistent two years ago, the market for startups that match companies with qualified candidates via a blind screening process is growing by the minute. GapJumpers, an emerging leader, compares its method to the “blind auditions” on the TV singing competition “The Voice.” A candidate completes a series of tasks related to a job and is given an anonymous scorecard, which serves as her first introduction to the hiring company. GapJumpers has matched candidates with tech, financial, and media firms and counts hiring managers from Google and Dolby among its users.

Ms. Lerner created her job-matching and interview prep platform, Interviewing.io, after a few stints in recruitment for the tech industry.  

“When I was hiring there was a strong preference for a very specific type of candidate,” she says. Too often, she saw worthy candidates overlooked because firms wanted hires “who went to the same five schools or worked for the same five companies.”

Interviewing.io lets hiring companies and potential employees chat anonymously, sharing ideas and solving coding problems, before any personal information about the candidate is revealed. “We don’t care who you are as long as you can code, and every interaction is blind,” Ms. Lerner says. “Interviewees can do practice rounds for free, and earn credentials based on their performance. When companies come to us, we send them the best performing candidates.”

The platform completed its first round of venture capitalism funding last year and counts a pilot program with Yelp among its partnerships.

Evidence that making candidates more anonymous yield more inclusive results is well documented. The most famous case comes from American symphony orchestras. In the middle of the 20th century, several major orchestras began auditioning musicians blind, concealing their identities by placing curtains between them and the selection committees. As a result, the percentage of female musicians in five of the country’s leading orchestras jumped from single digits in the 1970s to 21 percent in 1992. Today, orchestras in St. Louis, Chicago, and New York, among others, are nearly half women.

“Changing environments can make it easier for us to get things right, instead of changing mindsets,” says Iris Bohnet, a professor of public policy at the Harvard Kennedy School of Government and author of the book “What Works: Gender Equality by Design.”

Before blind auditions, conductors typically handpicked their musicians. But past research shows that even expert predictions of a candidate’s future performance in a particular role can be unreliable, bolstering the argument that selection processes, for jobs or otherwise, are more effective the more objective they are.   

One example: In the face of a statewide shortage of physicians, Texas passed a law in 1979 requiring the University of Texas at Houston to expand its medical school class from 150 students to 200. That forced the admissions committee to admit dozens of initially rejected students.

Sensing a rare opportunity, the University of Texas deployed a team of researchers to track how the rejected students’ performance in medical school and postgraduate work stacked up with students who had been admitted in the first place. It turned out to be just about the same.

That prompted a closer look at the school’s admissions criteria, which included a candidate interview with a members of the faculty and admissions committee. Academically, rejected students didn’t do significantly better or worse than their accepted counterparts.

But they tanked the interviews. That meant that the faculty members’ intuition that a candidate wouldn’t do well in medical school overrode more objective indications to the contrary. And it was dead wrong. So wrong, in fact, that the researchers suggested that the interview component be replaced with a lottery system for otherwise comparable candidates.

“If they took out the interviews, there was a correlation with future performance,” says Ms. Bohnet. “The interviews were just noise.”

Maybe if you change your name?

Hiring platforms like GapJumpers are an outgrowth of a tech industry that has earned its reputation as a white boys’ club. In 2014, in an effort to be more accountable, Google, Amazon, LinkedIn, and others of the biggest names in tech began releasing their employee demographics to the public. Women make up roughly 30 percent of the workforce at those companies (compared with about 59 percent of the US labor force), and less than a quarter of leadership positions. Ethnically, even the most diverse tech giants had percentages of Hispanic and African-American workers in single digits.

“I have a hard time believing most companies who claim to be equal opportunity, fair, unbiased ... and then ask for a résumé and a handful of references,” says Tim Yocum, a support and operations manager for the database company Compose.io. “Neither has ever adequately provided a solid view of a person's ability to succeed at a job.”  

Compose, which launched in 2011 and was acquired by IBM last year, first began hiring the traditional way: “Collect résumés, look through them like baseball cards, and call back the candidates with the right combination of stats and pedigree,” he says. "This information can often trigger unconscious bias. We knew it was happening even as we tried to avoid it.”

The startup soon switched to a system that focused more on work samples – giving candidates a job-related problem to see how well they solved it – and stopped looking at résumés and names. Candidates who score well then join the company for a work day.

“Our team has grown from a stereotypical startup to a really diverse crew,” Mr. Yocum says. “We've hired a lot more women than I think we'd have otherwise.” The company's operations and support group is now 45 percent women.

Like Lerner, the founders of GapJumpers were inspired by their own experiences in the stubbornly insular environs of Silicon Valley. “In college, I was struggling to get interviews for an internship, and the suggestion was made that I try with an abbreviation of my last name,” says GapJumpers co-founder Petar Vujosevic. “They said, ‘Maybe this will get you through the door, and then no doubt they’ll get you a job.' You go, 'OK, maybe I should play that game,' but we think people can play on a more level playing field.”   

Lerner says the culture of Silicon Valley is so closed off, and the language of interviews so technical and specific, that it can be difficult for any outsider to crack. She saw people as qualified as federal defense contractors and veterans of major movie studios get passed over for coding jobs because “they didn’t come from sexy backgrounds, or know the hot language.”

Another obstacle, Mr. Vujosevic says, is that managers who need to hire someone in a pinch are prone to fall back on preexisting relationships (between 70 and 80 percent of jobs are landed through networking), which can be a limiting factor to achieving a more inclusive workforce. “If I need to hire five engineers tomorrow, the consequence is I’m going to go where I know,” he says. “But we offer a way for companies to interrupt that implicit bias, and for hiring managers to be more open.”

Is it working?

It’s early yet, but both Vujosevic at GapJumpers and Lerner at Interviewing.io say they have had some early success getting more unlikely candidates to the interview stage, particularly women and those with less glamorous educational backgrounds. GapJumpers says women make up nearly 60 percent of the top-scoring performers in the blind auditions it hosts are women, as were 68 percent of those ultimately hired. In a survey released in May of last year, the platform saw a 15 percent increase in community college graduates who got an in-person interview, compared with those who went through a more traditional application process.

Additionally, Lerner says, a more anonymous process exposes companies to candidates without any traditional education.

Andrew Deal, a freelance software engineer and Interviewing.io user, learned to code mainly through online courses on Coursera and EDx. Before a friend referred him to the service, he says, companies often “rejected his application outright."

He’s still searching for a job, but says Interviewing.io has helped him land several interviews, both through connecting him with employers and giving him a space to practice technical interviewing. “The platform has streamlined my application to companies that otherwise would have taken weeks,” he says.

The Monitor is launching an in-depth focus on solutions to inequality. Want to read more?