When Armstrong first set foot upon the Moon, he said “This is a
small step for a man but a giant leap for mankind.” Minutes later, responding to President Nixon, he
said1 “It is a great
honor…to be here representing…men of peaceable
nations, men with an interest and a curiosity, and men with a vision for the future.”
Taken out context of the 1960s, the repeated use of man and men certainly doesn’t read very much like space exploration
included women. More importantly, however, it also didn’t sound that way
to many of the ~650 million
listeners2 at the time. Worse, even current writings about the Apollo space program tend to parrot
the same language using phrases like when man first
walked on the Moon or what a great achievement it was for mankind .
To help skeptics empathize, imagine if instead of a crew of all white,
Christian men as the first Apollo mission was (maybe all Apollo
missions), the first mission to Mars is crewed by all Black, Muslim
women and the first to walk on Mars says something like “This is
a small step for a woman but a giant leap for woman-kind…Allahu
Akbar.”
Joanna Lee,3 a contributor to the
Inclusive Naming Initiative4 and
author of inclusive language guidelines on
inclusiveprism.org5, calls the use of
such language in software projects an inclusivity bug because it
creates barriers to full participation by all stakeholders. Inclusivity
bugs arise out of the use of language that discourages whole groups of
people, often from underrepresented and/or historically oppressed
populations, from contributing as their full authentic selves. This can
happen regardless of how common-place any manner of speaking or turn of
phrase is or may have been for its time. In fact, an alarming variety of
inclusivity bugs stem from language we all commonly use.
Readers might dismiss Armstrong’s specific words as merely a figure of speech or a sign of the times. That's just the way we spoke then. Armstrong didn’t intend to exclude women. Or, readers might think that the problem is the English language because it does not easily avail itself to gender neutrality. But, that would be wrong. While English is notoriously lacking in gender-neutral pronouns, it is otherwise one of the most accommodating languages for gender neutrality.
As HPC/CSE software developers maybe we’re not doing anything nearly as
momentous as being the first Earthlings to walk on another planet. Do we
really need to worry so much about inclusive language? If we care about
our project’s reach and maximizing our ability to attract and retain
collaborators, developers, users or sponsors, then yes we do. That is
because the language we use has the power to welcome others in as well
as push them away.
Software projects involve a lot of communication. As developers, we
write code, comments, documentation, research papers, presentations, job
postings, promotional materials, discussion posts, emails, chats, tweets
and more. It is now well known that language used in job
postings6 has the effect of biasing
the respondent pool. The more we adopt inclusive language, the more we
reduce biases and create welcoming and safe spaces for all participants
to be their authentic selves, to be productive and to thrive.
Any readers who have written proposals likely already appreciate the
role language plays when communicating our work. When writing proposals,
we often struggle with and debate at length how individual word choices
will align our proposal with the sponsor’s call. In other words, we all
already appreciate the importance of and can be quite mindful of the
impact of language on funding. Being mindful of the impact of language
on inclusion is no different.
Inclusive naming is closely related to inclusive language but
focuses primarily on the naming of things. That is the names we
choose for abstract objects of our software systems. There is probably no other human activity that involves the naming of new things as much as software development. There is power in naming and with great power, comes great responsibility. The names we use in our code and its associated artifacts (PR reviews, documentation, presentations, social media posts, etc.) are often shaped by our culture.
A special challenge
with inclusive naming is that people tend to get pretty attached
to names. Was Edmund Halley the first to observe and record the
comet7 that bears his name?8 Chinese
astronomers observed and recorded9 it
almost 2,000 years earlier! Was Pythagoras the first
person10 ever to discover and write
down the rule we know as the Pythagorean theorem? Plenty of scholarly
research11 says others developed it
earlier.
For many terms and phrases in common use, we are simply unwittingly
parroting others. That can be problematic when one is not mindful of the
meaning, history or impact of a term. For example, are readers fully aware of why we call
defects in software bugs?12
How might women feel reading documentation that constantly refers to the
user as he, his or him?13 How might Black people
feel reading that existing users will be grandfathered
in14 when the software license
terms are changed? How might neurodiverse people feel about being tasked
to add sanity checks15 to a
software package’s test suite?
By the way, do readers recognize any bias in the preceding questions?
It's subtle but worth mentioning. The way these questions are worded
assume that only women would care about documentation that is laden with
male pronouns or that only Black people would care about language
derived out of racially oppressive practices or that only neurodiverse
people would care about language associated with atypical patterns of
thought or behavior. The truth is, we all (should ) care. And, to
use language that assumes or suggests otherwise is, well, not inclusive.
No way to phrase something will be acceptable to literally every individual. And, there is no reason we should be aiming for
that either. Catering to any one individual’s tastes and
sensibilities is not what inclusive language is all about. The goal of
inclusive language is to reduce the use of terms and phrases that
discourage whole groups of people from participating as their full
authentic selves. It is not about being politically correct or being the
language police. It is not about avoiding
offending16 people either. It's about
being willing to acknowledge that certain terms and phrases (and,
honestly, even names, icons and logos), however common in current
culture, can be unnecessarily exclusive and being willing to consider
and adopt alternatives that are less so.
If you’ve read this far, you may be asking yourself, how do I get
started? We suggest reading some of the resources available from other
major organizations such as the federal government’s Plain
Language17, Google’s Inclusive Style18 or
Microsoft’s Bias Free
Communication.19 Be aware that some resources and efforts are focused more upon identifying specific words and phrases to be fixed whereas others are focused more on principles of inclusive language. In particular, if
you are looking for tooling to help alert you to inclusive
language issues, some resources provide tooling as a web service. That
said, community standards and DevOps-hardened tooling for inclusive
language similar to spelling or grammar checkers is still in its infancy
stages.
Ever since adopting the practice myself, not a week goes by
that I don’t have the experience of questioning a term or phrase I am
about to use. I often spend a few minutes searching the web to learn
more about it. This includes terms and phrases I have used many times
before without really thinking about it such as describing an off-topic
enhancement request as having gone off the
reservation,20 or replying to it
with no can do,21 or
introducing a colleague as a Python Guru22 or a Java Ninja,22 or complaining my
bug reports are falling on deaf
ears,24 or that the level of effort
is too many man-hours.25
Many on-line resources may be useful to readers who would be similarly inclined. But, beware. In seeking greater understanding
of the history of terms and phrases you thought you knew, you find a lot
of misinformation26 out there to
sift through. In addition, when you are seeking to learn more
about any possible negative impacts of a given word or phrase, it's best
to seek opinions from experts within the groups most likely
impacted. In other words, if a given phrase has potential negative
impacts against Blacks, then language experts within the Black community
will likely have the the most authoritative guidance. If a given phrase
has potential negative impacts against the deaf, then language experts
within the deaf community will likely have the most informed guidance.
Finally, in seeking to fix inclusivity bugs, it's important to take care
that we don’t introduce another kind of problem…excluding the use
of perfectly acceptable language for no other reason than out of fear of
looking bad or being labeled non-inclusive. A good example is the word master27 alone, wholly apart
from slave . Hopefully, we all can agree that master/slave28 language is
not acceptable.
But, there is much less agreement, notably even within the Black
community, about the word master alone and with no relation to slave . Many uses of master have no
historical roots in oppressive or genocidal systems. These include such
terms as mastermind, postmaster, master key, master recording, achieving mastery of a skill and even
common tech terms like webmaster and scrum master. This
issue garnered much attention when GitHub
announced29 it would change its default branch name (which most users simply adopt without
question) from master to main. In that move it is worth
pointing out that the default language of a widely used resource
such as git is a significantly greater proliferation potential
and represents a qualitatively different situation than the choices
individual projects make. Furthermore, in GitHub's move no project was prevented from
using master as a git branch name if they chose.
Given the current social justice climate in which we all operate, some
readers will feel that inclusive language efforts are nowhere near
enough to meet the moment and are really just a
distraction28 from bigger issues.
It's hard to argue with that. Nonetheless, others will feel like such
efforts go way too far. While we can acknowledge both perspectives
exist, it is worth considering that when we’re accustomed to parroting
the status quo, pausing to be more inclusive may feel like
oppression.30
Mark C. Miller is a computer scientist supporting the Weapon Simulation and Computing (WSC) program at Lawrence Livermore National Laboratory for more than 30 years. He has recently developed a passion for addressing issues at the nexus of computing and inclusion. Among other things, Mark contributes to
VisIt,
Silo,
HDF5 and
IDEAS-ECP. Contact him at
miller86@llnl.gov.
Footnotes
8The comet is named for Halley not because he saw or recorded it first but because he is believed to be the first person to have estimated its orbit and based on that connected recorded observations from 1531, 1607 and 1682 as being the same celestial object from which he correctly predicted its return in 1758. That said, apart from the very significant achievement of the comet's orbital estimation, it isn't clear if astronomers much earlier than Halley had recognized this repeat visitor as the same celestial object.