Words Matter

A few years ago, I was asked to give a talk about the future of technology to some senior executives. The presentation was a beautiful, polished set of charts, all ready to go. I began practicing the pitch. Printed in big, bold letters on a page was: “Man + Machine.” I read from the accompanying speaker notes: “We’re seeking to reduce the barrier between man and machine.”

It immediately stopped me in my tracks. This was a narrative that had been delivered by many other senior leaders to wide acclaim. And yet I felt if I conveyed the message in precisely this way, it would be disingenuous, and I would be alienating half my audience. What was I saying? As a woman technologist, was the barrier with machines not going to be reduced for me? I went into edit mode and quickly switched out every instance of “man” and replaced it with “human.” It didn’t change the meaning…and I no longer felt excluded from the future of technology.

I have no doubt that those who crafted this presentation intended “man” to stand for any generic person (just as most choose to assume that the line “all men are created equal” in the U.S. Declaration of Independence pertains to all people — not just men, not just white men). I also appreciate the appealing alliteration of “man” and “machine.” Yet this is an example of how the words we use can either estrange or include. In a world where we desperately need to join together and inspire others — words matter.

Recently, there has been a renewed focus around inclusive language, with some important steps made in the past year in the wake of the George Floyd protests, which have shone light on long-standing racial inequities and injustices.

For many years, Native American organizations in the United States have campaigned to eliminate the widespread use of derogatory terms and harmful stereotypes of Native people in popular culture. In particular, the term “Indian” and related Native language and imagery have been used extensively for U.S. school and professional sports mascots. According to the National Congress of American Indians (NCAI), “Sports mascots are systems of disrespect that degrade, mock, and harm Native people, in particular Native youth.” The Change the Mascot campaign started more than fifty years ago, and there has been some progress along the way. In 1972, Stanford University changed their mascot from the Indians to the Cardinal; in 1994, St. John’s University changed from the Redmen to the Red Storm; in 2012, the University of North Dakota dropped their Fighting Sioux nickname.

However, today there are still 1,899 schools across the nation that have Native-themed mascots. This year, the high school where my husband teaches just changed from the “Indians” to the “Wolves” after contentious debate. Those in favor of keeping the name cited tradition, history, and legacy, and the honor of the Indian tribe from the region after which the town is named. Those advocating for change argued the name was inaccurate, disrespectful, and could offend students and members of the community. Ultimately, the superintendent decided changing the name was the right thing to do.

The 2020 season was the first in which the National Football League team formerly known as the Washington Redskins became the Washington Football Team. The prior name was long regarded by many as a racial slur, but no action was taken until pressure from Nike and FedEx, who threatened to withhold sponsorship. In December 2020, Major League Baseball’s Cleveland franchise announced plans to retire their “Indians” name after more than a century. But other professional sports team like hockey’s Chicago Blackhawks, baseball’s Atlanta Braves, and football’s Kansas City Chiefs have yet to make changes.

In cybersecurity, the industry has for years used the terms “whitelist” and “blacklist” to designate who has access to certain systems or networks. Someone on the whitelist has permission to run applications, download data, and modify code; someone on the blacklist does not. This is such common practice that few consider the deep-seated racism embedded in the use of these terms. Yet they carry with them a presumption that white is good and black is bad, perpetuating stereotypes. Due to greater awareness, this is beginning to change. In April 2020, the U.K.’s National Cyber Security Centre announced they would instead use the terms “allow list” and “deny list.” In June 2020, Cisco Talos announced plans to use “allow list” and “block list,” and the Ansible open-source software community announced a transition to “allowlist” and “denylist.” Others have followed suit to curb inflammatory jargon and encourage or mandate alternative terminology.

In computing, the terms “master” and “slave” have long been used to describe the scenario in which a dominant control system exerts power over one or more subordinate devices. Most engineering and software textbooks in use at universities today use “master/slave” terminology. But these terms have sparked controversy, as many challenged them as insensitive and offensive. In 2014, the open-source framework Drupal replaced the terms with “primary/replica.” Django opted for “leader/follower.” In 2018, the Python programming language went with “parent/worker” or “parent/helper.” Instead of “slave,” Kubernetes uses “replica” or “worker” or “minion.” Credit to Boston University graduate student Santiago Gomez who in 2020 convinced publisher Pearson to begin changing this language in hundreds of their print and digital technology books. In 2020, Red Hat took a leadership role with a comprehensive plan for eradicating problematic language from its practices and vocabulary: “If open source is truly meant to be inclusive and a place where anyone can participate, it must be welcoming to all.” This is progress, but there’s much more work to do to rid the technology industry and society at large of repellent terminology.

Sometimes terms become part of our lexicon, almost a default, and we don’t even realize they carry connotations or may deliver unintended messages. If we post a job description that says, “An ideal applicant should know Python and Java, and he should have at least five years of experience” — it may turn off women. When we ask how many “man hours” of work this job will take — it suggests only men are doing the work. If we describe a technology use case and personas all have traditional American male names — it makes others feel excluded. Alternatives demonstrate respect, inclusivity, and awareness of the power of words.