Diversity: Why open source needs to work on it in 2020

Commentary: Given how central open source has become to tech, we need open source communities to lead the way on diversity.

black woman employee speaking at diverse meeting

Image: Getty Images/iStockphoto

Sometimes the answer to a problem is hiding in plain sight. Take, for example, the problem of diversity in tech and, specifically, in open source software. 

If we look at US Bureau of Labor data, 21.2% of professional developers are female. According to a 2017 GitHub open source survey, however, 95% of respondents were male and just 3% were female (1% identified as non-binary). It turns out that 68% of the female respondents are "very interested" in contributing to open source, but are significantly less likely to do so than men (45% vs. 61%).

The reason? Well, men, to put it bluntly (or unwelcoming behavior, to add some color). For those who don't think diversity matters, this post won't convince you. But for those who would like to see open source development become more representative of the people who will use it, a 2019 DigitalOcean developer survey offers clues for improvement. 

Was blind but now I see

According to that survey, women are far more likely to say they'd be more likely to contribute to open source "if the open source community was more inclusive." That inclusivity is often measured by how welcoming open source is.

Figure A

9db4ac3f-e55a-4996-af76-c17c912007b6.jpg

Image: DigitalOcean

Notice anything interesting in the data above? Men are more likely to participate in open source, less likely to feel diversity is important (while simultaneously being more likely to think everything is A-OK), and also less likely to believe they need much to get going. 

SEE: Transgender employees in tech: Why this "progressive" industry has more work to do to achieve true gender inclusivity (cover story PDF) (TechRepublic)

This same thing is revealed in the GitHub survey, where men and women diverge on key areas like codes of conduct and welcoming communities (Figure B).

Figure B

bb99dd95-49de-4851-9c25-b36877c6fe74.jpg

Image: GitHub

There are hints that maybe things are getting better, with younger developers expressing a more positive experience with open source than older peers (Figure C).

Figure C

725bbc81-d9bf-4831-960a-263700d88c44.jpg

Image: DigitalOcean

But a certain blindness persists. Developers who are new to open source, whether young or female or both, want better resources to help them contribute, but those resources are most likely going to need to come from the older males who dominate open source. That group, in turn, doesn't really see a diversity problem and is less likely to be helpful. But it's that occluded vision that is really the problem. If men were more likely to discern the lack of diversity, and understand its implications, presumably the behavior that keeps would-be female contributors out would change.

So let's talk about why diversity matters.

We're not the same

First, let's acknowledge that open source software has never been more important. It is foundational to how great software gets written today. From Kubernetes to Android to Linux to [insert your open source project of choice], the world runs on open source. We would be out-of-our-minds crazy to not want the software that runs the world to be more representative of the world that runs it. 

This seems intuitively correct, and fortunately also has good data/science behind the intuition. There are scads of studies showing that the more diverse a company, the better its financial performance (including this one from McKinsey), and a great summary article in Harvard Business Review that finds "nonhomogeneous teams are simply smarter." Why? Because, as James Beswick has argued, a non-diverse development team leads to a "mono-chromatic, average approach to problem solving and an inability to think dynamically," which is why "Adding diversity is absolutely fundamental to the problem-solving process because we all have different life experiences, backgrounds and knowledge that combine to give us fresh insights and different approaches."

Open source development communities are working on some of the world's most difficult, important software. The people writing it should therefore reflect the difficult, important life experiences of a wider variety than we currently see.

SEE: Hiring kit: Chief diversity officer (TechRepublic Premium)

If we don't solve for this at the open source level, we risk making tech, overall, even worse.

Why? Given how dependent the world is becoming on open source, open source has become central to how developers build credibility and land jobs. In the GitHub survey, for example, roughly half of respondents said participation in open source development was important to helping them get their current jobs. We actually risk making tech's lack of diversity worse by keeping it difficult for women to contribute, because it will be that much harder for women to find good jobs. 

The sad irony is that this isn't a particularly hard hurdle to overcome. More than anything else, we need open source communities to not foster or encourage jerk behavior. According to the GitHub survey, while roughly 15% of respondents had directly experienced rudeness in their open source communities, nearly 45% had witnessed it. As a result, 21% of people who experienced or witnessed such behavior stopped contributing to the project. The percentage of women bowing out is almost certainly higher. 

SEE: Revised Linux Code of Conduct is now officially part of Linux (ZDNet)

As such, we need to set up clear guidelines (codes of conduct) to encourage and, to some extent, enforce welcoming behavior. We don't need quotas to brute force women into the open source workforce. Rather, we just need mechanisms in place that keep them from being shoved out.

As we do so, open source software will get even better, which, in turn, will make technology better, in general. 

Disclosure: I work for AWS, but the views expressed here are my own.

Also see