Image via Gild
Tech companies have a diversity problem, especially when it comes to hiring software engineers. Software may be eating the world, but for the most part that software is created by white males, despite the fact that there are many qualified engineers in the marketplace who are neither.
A series of self-reinforcing cycles reinforces the status quo. Companies tend to hire alumnae from certain universities and workers from particular companies; they also rely heavily on referrals from existing employees. The result is a workforce with a lot of faces that look similar.
It’s the kind of problem you might expect could be solved by, well, software. Human biases in hiring, innocent and otherwise, can be corrected by an approach that ranks candidates based on the quality of their body of publicly visible work — or so the thinking goes. That’s what Gild does. It’s one of a few up-and-coming companies that has sought to give its customers — some 300 companies at last count — a leg up in finding qualified software developers.
The San Francisco-based company has been growing fast and recently closed a $13.5 million Series B found of venture capital funding led by Menlo Ventures that brought its total capital raised to $27 million. Prior investors include Draper Nexus, Baseline Ventures, Globespan Capital, SAP Ventures and Correlation Ventures.
In an interview with Re/code, Gild CEO Sheeroy Desai and chief scientist Vivienne Ming talked about how the company is starting to help its customers grapple with the difficulties of building a more diverse work force.
Re/code: Gild has just closed a big funding round, and you’re working with companies as large and varied as Microsoft, VMware and Progressive Insurance. Can we start with an explanation of what’s driving your business?
Desai: Number one is the sheer shortage of software engineers. Every CEO I talk to talks about this problem. What we’re seeing playing out now is exactly what Marc Andreessen wrote about: Software is eating the world. You can look at any and all business models, and software is taking them over. As that happens, the world ends up needing more software engineers. And so that acute shortage makes us highly topical right now as companies look for any advantage they can get in hiring them. Also, companies have realized that they need to walk away from their own internal biases in hiring practices. They need different strategies.
When you talk about biases, what do you mean exactly?
We’re a business that was founded specifically to encourage hiring based on merit. But it’s been frustrating to see that for the most part, mostly in tech, there’s a tendency to hire people who went to certain schools and people who have worked at certain companies like Google or Facebook. It’s highly frustrating because we know the data and we’ve looked at it, and there are unbelievable people who don’t necessarily have the pedigree or the credentials, but whose work shows them to be incredibly skilled software engineers. There are a number of software engineers who, for one reason or another, didn’t even graduate from high school, but they’re still incredibly skilled. But they would easily get filtered out of any company’s screening process.
Do you then find that companies are becoming more open to breaking away from these biases?
It’s a mix. There are some companies that are starting to say they want to change how they hire. Google has been saying that for a year or so now. Their practice may vary from what they say, but at least they are publicly saying it. We work with some customers who are actively trying to change and telling us they don’t want to use their old policies anymore. And there are some that are sticking with them because they say they work. In those cases, I think reality will catch up with them because the supply of programmers who mirror their requirements will be exhausted or they will just become remarkably expensive. That’s the other problem, because when you place a high value on those credentials, you end up overvaluing someone who may not be worth the money and undervaluing someone who may not have the credentials but who has the skills.
So you’re saying there are a lot of overpaid software engineers who may not be so great?
Desai: Compared to the number of highly qualified software engineers who are underpaid, yes. In some ways we’re trying to get the market to clear efficiently.
So we’ve arrived at this moment where we’re having a broader conversation in Silicon Valley about class, social upheaval and overall economic viability. Is your data showing a meaningful shift in how companies are acting in their hiring practices, or anything relative to outcomes?
Desai: I think meritocracy and diversity are closely linked. They are not the same thing, but they are closely linked. And I think one of the things we’re starting to see are numbers being publicly reported by some large companies. Google has done it, Yahoo and Facebook have done it. So has LinkedIn. Pretty much the story is the same at all of them. And when it comes to software engineers the story is pretty abysmal. The vast majority are white males. Are there more software engineers who are white males? Yes, it’s true, but it’s pretty striking what is going on in these places.
Vivienne Ming: There was a great line I read in a blog post about this recently saying that if spam filters sorted mail the same way Silicon Valley sorts engineers, you’d only ever get mail from your college roommates, and you’d never know that anything was wrong with that.
Desai: And there’s one thing that tends to reinforce the hiring bias, and we see this again and again at different companies, especially tech companies: We ask them [for] the number one way they find their employees, and they say it’s their employee referral program. People refer who they know, and that means that if your software team is already white and male, guess who they’re going to refer? And what companies don’t realize is that this is bias, and when you institutionalize it, it is discrimination. It’s unintentional. And I don’t want to accuse any companies of wanting to discriminate against certain ethnicities or anything like that. I don’t think that’s how it starts. The intention may be pure, but the result is still the same.
So where do you see Gild fitting into this?
Desai: We’re not going to solve this problem by driving diversity. What we’re about is the argument that the decision-making process in hiring software engineers should be based on merit and the body of work that you’ve done, rather than your educational credentials and who you know.
If a company wants to address this directly, it would seem they could search for potential employees in such a way that they’re blind to gender and where they went to school and so on. Can you search for employees based purely on merit, on what code they’ve posted to GitHub and so on?
Ming: Can we do it? Yes. For research purposes, I’ve built filters that run through our database and pick out information that is indicative. There was a high-profile project done at a university in Switzerland where the gender and ethnicity markers were stripped off the applications of graduate students and suddenly they found they had admitted a huge number of Turkish and female students in the following year’s class. People have a good reason not to put this stuff on their resumes, so it takes a certain amount of work to build a system that can figure it out. So when I say I’ve done it, it’s not to the degree of accuracy that we’re going to promote this and make it available. The point is that when someone does a search and asks, for example, “Who are the best Java developers in New York City?” we present an unbiased list, or as unbiased as we can. … I think what you’ll find is that the number of women and people of different ethnicities that will appear in our Top 100 list is significantly different than what you’d find in a poll of the 100 best developers in New York.
And yet from what you’re telling me, most companies would probably feel more comfortable hiring programmers based on who their employees recommend. If you take cost considerations out of the discussion for a moment, is that necessarily such a bad thing for a company?
Ming: There’s been a lot of research into this issue of positive discrimination that [shows] you tend to attract and refer people who are like yourself. There’s a lot of research that says that internal referrals work because they’re more likely to get people in the door. It’s a great solution to your short-term needs. The problem is that the research says it’s a terrible long-term solution because you end up with creative stagnation.
It sounds like there’s some fundamental tension between the short-term need to fill an open position and the long-term intent.
Ming: There is. That’s why we say some of our customers are schizophrenic. Their leadership is concerned with the long-term view. Meanwhile, people on the front lines doing the hiring are more concerned with keeping their jobs and meeting their numbers and filling the immediate need. I won’t name names, but we met a very prominent technology company on site, and they took us through their hiring process step by step and apologized for every step. They were doing all the things that we’ve been talking about, but at the end of it, they said, “We don’t know what else we should be doing.” … There’s a recognition of the problem at the leadership levels, but people feel trapped because they need 30 programmers today.