Open source effort seeks to build tech diversity into code
A CNCF initiative to build tech diversity directly into open source code and documentation takes shape as the industry faces pressure to become more inclusive.
Amid a cultural reckoning about race in the United States and a global pandemic in 2020, the tech industry has been challenged to broaden its diversity, down to the level of open source software code.
Tech diversity isn't just a matter of platitudes in marketing, according to industry experts -- a lack of diversity in the industry's workforce will affect the bottom line at many if not most businesses as digital transformation takes hold. Enterprises building software stand to lose money without better tech diversity not just because of potentially poor brand reputation, but because without diverse developers, what they sell won't be relevant to all their potential customers.
Christopher Lafayette
"Here's what's at stake -- if we continue to dodge the elephant in the room, and don't work with or hire from communities of color, it won't just disrupt business," said Christopher Lafayette, a technologist who consults with enterprises and speaks publicly on issues of tech diversity and founder of the Black Technology Mentorship Program. "It will affect how businesses compete and do business globally. You have to hire the people you also want to buy your products."
The solutions to a lack of tech diversity aren't going to come through PR-driven hiring initiatives, either -- practices and language must also change, down to the level of open source software code, according to members of the new Inclusive Naming Initiative (INI) project within the Cloud Native Computing Foundation (CNCF).
Priyanka Sharma
"When we look at our codebases, our ethos is not always reflected," said Priyanka Sharma, CNCF general manager and co-founder of the open source initiative at the group's public kickoff meeting during KubeCon in November. "Terminology such as master/slave and whitelist/blacklist still exist in our codebases today, and that is not OK. That's what our collaboration seeks to fix."
The initiative first arose in June 2020, after footage of George Floyd's killing by a police officer circulated in an online video and captured public consciousness throughout the U.S. In the aftermath of that incident, projects and companies, including the Linux kernel, as well as IT vendor Red Hat, began or accelerated projects to remove problematic language from open source code and documentation. Eventually, the CNCF initiative formed so that these efforts wouldn't overlap, conflict with each other or duplicate work.
"What forced the forming of this initiative is that we didn't want to replace terms like master/slave a million times over," said Celeste Horgan, senior technical writer at the CNCF, during a public meeting of the group at KubeCon + CloudNativeCon Europe this month. "Projects were all rushing to replace these terms ... but they weren't necessarily moving in a consistent direction."
Open source initiative digs into diversity details
While the group's kickoff in November was mostly a brainstorming session, an organizational structure has coalesced since then, with project leads and specific goals, organized into subgroups the project refers to as workstreams.
Some INI workstreams are domain-specific, such as the Company workstream, which advises open source projects and maintainers doing inclusive naming work. Another domain-specific workstream will coordinate inclusive naming efforts with broader standards bodies such as the IEEE.
Others are cross-cutting efforts, such as the Language workstream, which researches and recommends specific inclusive language for use in code and documentation, and has already created an initial word replacement list. Early participants in INI include tech vendors Akamai, BMC, Cisco, IBM, Red Hat, VMware and Splunk.
Language changes are a lot further-reaching than people assume.
Celeste HorganSenior technical writer, CNCF
The INI hasn't published a roadmap with timing for specific workstreams yet, but the project's overall goal is to develop a certification program for CNCF projects by the end of 2021. Like the Kubernetes conformance program under CNCF, this program will affirm that project codebases have been updated with inclusive language as defined by the INI.
Leaders of the initiative said there aren't any plans to make inclusive naming certification a requirement for CNCF projects. They also warned that the work will be painstaking.
"Language changes are a lot further-reaching than people assume," Horgan said during this month's INI meeting. "People sometimes propose these changes without necessarily thinking through the process of what it's going to mean for maintainers, or companies."
Relatively small changes to the names used by software code can ripple through multiple projects dependent on each other, in ways that can have unintended consequences, INI leaders said. Backward compatibility and allowing time for migration to new versions are also important considerations for the project. And that's assuming all maintainers agree about the need for inclusive language.
"There is absolutely pushback that we're experiencing," at times from the world at large, Hogan said, though work within the INI has largely moved past those fundamental issues, she said.
Tech diversity requires multifaceted effort
Leaders of the INI recognized that tech diversity also requires cultural shifts outside the scope of their open source initiative. The group published a guidance document on how to launch corporate tech diversity initiatives on its website, but every company will respond to such suggestions differently.
"The way you handle this in a small [open source software] OSS project and a large OSS project are different conversations," said Stephen Augustus, co-founder of INI and head of open source at Cisco. "You have to make sure [changes] happen in a way that aligns with how a community or company does work."
The Linux Foundation and organizations like Lafayette's Black Technology Mentorship Program are also working to address the skills gap in IT, which disproportionately affects women and communities of color. This has also been exacerbated by the pandemic, which all but destroyed work-life balance among many women in the workforce, leaving less time to do unpaid work such as contributing to open source projects.
Nithya Ruff
"Upstreaming something or making time to contribute is often a privilege," said Nithya Ruff, head of open source at Comcast. "And especially with COVID-19, and increased family [commitments], very few people really find the time to contribute [and] not too many women contribute versus men because of [differences in] privilege of time."
Some companies, including Comcast, are trying to foster tech diversity by setting aside time for employees to work on open source initiatives and encouraging contributions to open source during work. Comcast hosts hackathons and has women in technology and Black engineers groups that encourage employees in underrepresented populations to contribute.
Other companies also set aside time regularly for employees to work on side projects.
"I am the technical product manager for an engineering team and I make it my business to make sure that on a regular cadence, I stop 100% of what we're doing," said Margarita Caraballo, a chapter director of Techqueria, a nonprofit that supports Latinx people in tech, and a product manager at Mailchimp. "We've called it recess, quiet hours, whatever you want to call it ... two or three days during the time that they're going to be at work anyways, [to] do whatever they want, whatever recharges them or they're passionate about."
This helps people from marginalized populations do some of the same community work more privileged people are able to do during their free time, such as mentor others, Caraballo said.
Contributing to documentation is also a way to broaden participation in tech and open source initiatives within enterprises, which also doesn't require everyone to learn a programming language. Though documentation is typically a neglected area of open source projects, the right documents can make projects more accessible, help novice developers know where to start, and broaden the base of contributors with translation into different languages.
"Mostly people think open source as code, but there are other aspects of open source like documentation and web design, [where people can] contribute in different ways," said Khem Raj, distinguished engineer at Comcast. "Those aspects of projects are underrepresented at that point, and as they become more mainstream, they'll be challenges for the business."
Ultimately, it's still far too easy for companies to talk about tech diversity, but not do the more difficult work required to make real change, Lafayette said. Representation from diverse groups in such efforts is also important, which can be a Catch-22 for companies that lack diversity to begin with.
"I've done work with companies that have gone to great effort on diversity initiatives only to realize they can't actually execute what they've built because they didn't really understand the problem," he said. "These efforts fail because companies are trying to solve this like a mathematical equation, and it's a multivariable situation."
Beth Pariseau, senior news writer at TechTarget, is an award-winning veteran of IT journalism. She can be reached at [email protected] or on Twitter @PariseauTT.
Christopher Lafayette
Priyanka Sharma
Nithya Ruff
