Why Can’t Silicon Valley Solve Its Diversity Problem?
Earlier this year, Facebook released its diversity numbers—which were only incrementally higher than they had been a year before, despite several internal initiatives—in a blog post that outlined what is usually known as the “pipeline” problem. “Appropriate representation in technology . . . will depend upon more people having the opportunity to gain necessary skills through the public education system,” Maxine Williams, the company’s global director of diversity, wrote. Later that day, an essay, “Invisible Talent,” circulated across my social-media feeds and Slack back channels. “I haven’t even started my first full-time job yet and I’m already so tired of feeling erased and mistreated by the tech industry,” it read.
Kaya Thomas, who wrote the essay, is a Staten Island native who is studying computer science at Dartmouth College. A black, female engineer who has interned for Intuit and Apple, she is no stranger to Silicon Valley or its particular brand of bias. “After reading the Facebook thing, I did feel crazy,” she recently told me over the phone. “Are you kidding me? There’s no talent? Who am I? Am I invisible? Do I not exist? I felt out of my mind.” Black and Latino workers fill only about five per cent of technical roles—meaning roles that require programming chops, whether that means writing software, designing interfaces, or analyzing datasets––in the tech industry, though they make up about eighteen per cent of computer-science graduates each year. According to Thomas, Silicon Valley is losing black and Latino employees either because of prejudice, or because technical recruiters aren’t looking in the right places. “A lot of students from underrepresented backgrounds, who will get their degrees in computer science and want to work in the tech industry, are not working in Silicon Valley,” she said. “They’re engineers in finance or at Walmart, or Target, or Coca-Cola. They’re not in the software industry. It’s because the network is so small and homogenous that if you don’t have access to it, you can’t get in the door.”
Technical positions might seem to be exactly the kind of jobs that it would be easy to fill from a diverse pool of applicants. The skills involved, such as logical and quantitative reasoning, are found in other fields, from finance to medicine to electrical engineering. Free online courses and developer boot camps abound. Yet, as we see again and again, these are the positions with the fewest women and people of color. This might be because most recruiters lack the skills to evaluate code (even those who can are often stymied by nondisclosure agreements from candidates’ past companies), and so end up relying only on candidates’ traditional credentials: a computer-science degree from a notable university, a previous position at Google or Facebook. Companies’ interview methods can also be misguided. Many employers, attempting to assess raw engineering talent, use controversial “white-board” interviews, in which candidates are asked to solve computer-science problems (“How would you triangulate a polygon?”) or brainteasers (“How many Ping-Pong balls fit in an airplane?”). Many people feel that these give about as good a sense of how well an engineer will write code as would a game of table tennis with the C.E.O.
Then there is the incestuousness of Silicon Valley’s hiring process. Many startups turn to their own employees for assistance with hiring, offering generous bonuses for successful referrals. A former colleague of mine made a modest second income referring friends to our company for jobs. (I was one of these friends, and came with an eight-thousand-dollar dowry.) Meanwhile, the social network Refer Me Please exists solely to connect job seekers with employees who will be compensated to refer them. Though companies do tend to screen for the ever-elusive “culture fit”—a friend of mine spent a week on a cruise ship with her potential employers, during their annual retreat—many tech workers are eager for that culture to change. “I look to see: Is it completely a sea of white people? Is it completely a sea of twentysomethings?” Haley Smith, a senior software engineer at Slack, told me. She relies on a “whisper network” of women engineers to evaluate potential workplaces.
The tech industry is metrics-obsessed, always optimizing, and eager to find technical fixes for inefficiencies. But the conversation about diversity in tech is also a conversation about social change—about economic inequality, access to education, and the latent racism and sexism of an industry that prides itself on building the future. It’s about the social insularity of Silicon Valley, and the insistence on clustering along a forty-mile strip of land, despite the ballooning international reach of technology products. Building momentum for social change is nuanced and painful and slow; it’s tempting to look for shortcuts.
“The brogrammer culture, the hoodies, the flip-flops, the ‘Revenge of the Nerds’ type of vibe,” Stephanie Lampkin, the C.E.O. and founder of Blendoor, a “blind recruiting” app, said over the phone earlier this month, “I think is a significant contribution to why a lot of women and minorities have been pushed out.” Blendoor is just one of the startups that have emerged over the last three years to tackle the recruitment problem the way they know best: through software. Lampkin, who is black, has an élite résumé: degrees from Stanford and M.I.T., a five-year stint at Microsoft as a technical account manager. Blendoor, which launched in 2014, looks like a hybrid of LinkedIn and Tinder, but with a twist: résumés, scrubbed of personally identifying details (photos, names, and graduation years, which can inspire racial, gender, and age bias) are presented to employers, who swipe right when they like the credentials they see. Still, she admits, even the blind-résumé model isn’t perfect.
Lampkin also acknowledges that a blind audition won’t necessarily secure a job offer; bypassing the résumé screen can only mitigate bias at the beginning of the hiring process. Similarly, Gapjumpers adapts the “blind audition” model for tech workers, who anonymously complete “challenges” to test competency, while Textio, a text-analysis tool, helps identify coded language in job descriptions. (Words like “adaptable” and “creative” attract more women; “ambitious” and “assertive” appeal more to men.) Tech companies use Textio to adjust their job listings, with the hope that linguistic tweaks will result in a more varied applicant pool.
Most of the startups that companies hope will address the diversity problem prefer to think of themselves as levelling the playing field more broadly. “Diversity often means women and underrepresented minorities,” Aline Lerner, the founder and C.E.O. of Interviewing.io, an anonymous-interview platform for tech positions, told me, as we sat in a conference room at a perky WeWork in downtown San Francisco, where she rents an office. “To us, it’s bigger than that. To us it’s anybody that wouldn’t be able to get a technical interview at a top company because they don’t have a specific kind of pedigree,” she said. Harjeet Taggar, the C.E.O. and co-founder of Triplebyte, a hiring platform for engineers that features a standardized, anonymous programming test, called for “a diversity of background . . . and a result of that should be a diversity of just, like, thought.” A wide range of backgrounds, he said, would encourage “companies [to] push their products in a direction that’s going to appeal to more people, because the people working at the companies are themselves more representative of the broader population.” Triplebyte regularly sees engineers who would be passed over in a résumé pile: a GameStop manager; a self-taught parent reëntering the job market. “Our ideal user is someone who’s working in Citrix in Florida, someone in the I.T. department,” he said, “or someone working for the last eight years as an I.T. support member of a tech company I’ve never heard of in Milwaukee.” In other words, diversity, in tech, is seen by some as relative—though most people agree that it is necessary for the survival of the industry.
Still, Taggar admits, most of the people who use Triplebyte to find jobs are men. The same goes for Interviewing.io. Chris Lyon, a recruiter at the gaming company Twitch, which has hired several engineers through Interviewing.io, told me, “I have not yet seen any women.” Lyon has asked Lerner to recruit more women, women of color, and black and Latino engineers to her platform. But both Interviewing.io and Triplebyte depend on word of mouth and coverage on tech-oriented Web sites like TechCrunch and Hacker News, which tend to bring in a self-selected user base. There is something matryoshka-like about recruiters at tech companies asking their recruiting platform to do broader recruiting. At some point down the line, someone will be doing the grunt work of running marketing and sourcing campaigns, or showing up in person at college job fairs.
For Kaya Thomas, one measure of success in diversification will be the number of people from underrepresented communities in top positions. Today, only about five per cent of employees in executive roles at tech companies are black and Latino. In addition, black and Latino tech workers are more likely than others to leave the industry within a decade; nearly half of women, of all races, also leave within ten years. (The rate of attrition for men, on the same time scale, is seventeen per cent.) Merit-based technical-hiring tools open doors for people who would previously have been shut out. It’s on the rest of the industry to do the introspective, inefficient human work of figuring out how to keep them.