A new study from University of Waterloo researchers shows developers perceived to be non-white are much less likely to see their proposed changes to open-source software projects accepted.
The study focuses on Github, an online hosting platform for software developers, and examined 37,700 projects and two million "pull requests," which are changes to projects proposed by other developers.
When approving or considering pull requests, users see only the name of contributors, so researchers used a tool called NamePrism to estimate the race and ethnicity of the 366,000 developers involved in the projects and then analyze the success the proposed changes had.
They found 70 per cent of contributions that were later integrated into the open-source software projects they analyzed were submitted by developers perceived to be white.
Developers who were seen as Asian, Hispanic or Black were responsible for less than 10 per cent of the contributions that were accepted on open-source software projects.
Researchers say their findings will help identify diversity problems in the industry, understand why they exist, and determine what interventions can help reduce and eliminate bias.
Github did not respond to a request for comment on the findings that were recently published in the IEEE Transactions on Software Engineering journal.
Mei Nagappan, one of the researchers and an associate professor at the University of Waterloo’s Cheriton School of Computer Science, said the difference in the number of non-white developers and white developers whose contributions are accepted is significant.
“The odds of (non-white developers’) contributions getting accepted are lower, but we don’t know why it may be lower,” he said.
To determine what’s behind the difference, Nagappan and fellow researchers Gema Rodríguez-Pérez and Reza Nadri would have to explore the data and study the issue more closely.
However, they hope their initial and eventual findings will help identify diversity problems in the industry, understand why they exist and determine what interventions can help reduce and eliminate bias.
Nagappan believes it’s important to research this area because it has seldom been explored, despite researchers examining how race and ethnicity impacts several other sectors and aspects of life.
Many studies in recent years, for example, have determined that resumés submitted with names perceived to be white have received more callbacks and job interviews.
“Biases have been noticed in other areas, but no study has looked into whether there are such biases based on perceived race or ethnicity ... in software in open source software development,” Nagappan said.
“Past work has looked at biases based on gender that can be perceived from names, but we wanted to look at race and ethnicity to see if something existed that exists in other fields.”
This report by The Canadian Press was first published July 29, 2021.
Tara Deschamps, The Canadian Press