Concerns concerning racial bias in open-source software have been raised by a recent research.
According to a recently released study, software professionals with names that seem white may succeed on GitHub more than developers whose names are viewed as Black, Hispanic, or Asian-Pacific Islander.
The results, which were earlier this year published in IEEE Transactions on Software Engineering, highlight significant issues regarding the effects of a lack of diversity on GitHub and in the open-source software community at large.
Over 2 million contributions, or “pull requests,” made by 365,607 developers were examined by University of Waterloo researchers. The researchers discovered that being regarded as white on succeed on GitHub often boosts a developer’s chances of getting their ideas accepted. This finding was made using an AI technique called NamePrism that examines people’s names for their perceived race and ethnicity. It raises those odds by 6 to 10% compared to developers who are seen as Hispanic or Asian-Pacific Islander.
“This is the only place where a true meritocracy could theoretically exist. There is no human presence in open-source software. You probably haven’t encountered them or formed an opinion of them. You at least recognize their name “Mei Nagappan, a co-author of the paper and an assistant professor of computer science at the University of Waterloo, stated.
Given the impact open-source communities like succeed on GitHub have on product development, it is alarming that racial bias may still exist in this setting, according to Nagappan. Software will be created by and for a fairly homogeneous people if we don’t pay attention to different viewpoints, he warned.
Additionally, as GitHub has evolved into a sort of portfolio for software professionals, this prejudice can negatively impact the careers of developers. “As a newbie, you may transfer that into a successful career at a company if your contributions are accepted to even one of the big projects,” Nagappan added.
In response to Protocol’s request for comment from GitHub, Nagappan stated that the research’s objective is to solve issues in the open-source community as a whole rather than focusing specifically on GitHub. GitHub developers who are viewed as women have lower acceptance rates, according to earlier study, according to Nagappan. The country of origin of developers has also been proven to affect acceptance rates.
He acknowledges that estimating a person’s race and ethnicity isn’t entirely accurate using the NamePrism method his team utilized. Researchers only designated a developer’s race or ethnicity when they were confident in the tool. They labeled the developer’s alleged race as “unknown” for everyone else.
The majority of developers who post ideas on GitHub and the majority of people who comment on those ideas have names that the Waterloo researchers believed were white, despite the fact that they avoided assigning a particular cause to this phenomena of racial bias on GitHub. Furthermore, they discovered that when the persons responding to pull requests are from the same racial or ethnic group, developers who are thought to be Black, Hispanic, or Asian-Pacific Islander are more likely to have their pull requests accepted.
The researchers propose that GitHub use a single or double-blind framework, similar to how research is evaluated in the field of academia, to address this possible bias. A different suggestion would require numerous reviewers to evaluate a specific contribution, ensuring that biases of any one person are eliminated.
The issue of how racial stereotypes affect individuals’ online interactions is not particular to GitHub. In order to examine how racial discrimination emerges on the platform, particularly the part that names play in distorting other users’ perspectives, Airbnb created Project Lighthouse last year.
