We grab lunch together pretty regularly at my workplace. I'm a software engineer at one of the world’s biggest tech companies, and while not mandatory, it's the company norm to eat with your teammates weekly—if not daily—at a cafeteria in our building.
Conversation tends to be as routine and predictable as the lunching itself. "What'd you do over the weekend?" usually spans Monday and probably Tuesday. "Plans for this weekend?" is a Thursday to Friday favorite.
But sometimes, on Wednesdays—too far from either end of the week—we run out of things to say (after exhausting the obvious topic of work itself). What then? Software engineers are not known for our conversational prowess. We fumble around, spinning the conversational rolodex in a sort of lunch-topic roulette, and usually end up somewhere safe and uncontroversial, like movies or travel.
But sometimes we don't.
One Wednesday, lunch-topic roulette landed on police brutality. At my company, where only 1% of engineers are black, the sole black engineer at the table recounted a time when he narrowly avoided wrongful arrest. He then compared it to one of the many recent cases of excessive police force against people of color.
The table balked at his accusation.
"How can you be sure this is racism?" one white male colleague protested. "You can't generalize anything from an isolated incident."
How can you be sure this is racism?
Another time, lunch-topic roulette landed on sexism in tech. I explained that while rare for me personally, sexism does happen at my company, where only 18% of engineers are women. In fact, recently, an engineer had infuriatingly mansplained to me programming concepts that you’d learn in an “Intro to Computer Science” class.
My male colleagues frowned.
"I don't know if that's sexism. Maybe he's like that to everyone."
Trying to bolster my point, I shared another incident from years ago that was even more egregious. While visiting my team, on which I was the only woman, one male engineer from another office pretended I didn't exist. He spoke solely to the men, sharing his fondness for scouting for "cute chicks" working late—presumably single women he could corner after other colleagues had gone home.
This, too, was met with skepticism.
"Is that sexism though?"
A table of kind-hearted software engineers provides one of the safest conversational environments. You can reveal your nerdiest hobbies to a rapt audience. You can make the lamest puns to polite chuckles, if not genuine belly laughs.
Why, then, are software engineers often terrible allies to people of color, women, and other marginalized groups?
I suspect it lies in the way we're trained to think about problems.
Logic interferes with compassion
Obviously, #NotAllEngineers are terrible allies. I'm talking about a broader issue of our conditioning as engineers to be skeptical of subjectivity, and how that affects our ability to understand discrimination.
Software engineers are accustomed to solving software problems. A computer will do exactly what code tells it to do, so if there's a bug, it's because the computer is being instructed incorrectly. To fix a bug, engineers usually try to pinpoint its source by coming up with a series of diagnostic questions to help figure out the issue. For example, if a bug prevents YouTube from playing sound, an engineer might try to determine: Is the audio being downloaded? Is it being processed correctly? Is it being sent to the speakers?
Engineers can get concrete answers in the realm of computing. We can definitively answer a question by simply checking whether or not the data exists. If we’re not sure why an application is behaving a certain way, engineers prefer staying accurately uncertain ("I'm not sure if the audio data downloaded"), rather than proceeding based on unproven assumptions. The rationale? You solve problems faster by being 100% sure about what you consider to be true and untrue.
This line of reasoning, which strongly favors uncertainty over assumption, is common to all scientific thought—not just engineering. For example, hypothesis testing is a technique used heavily in science to make conclusions about the world. For this, you come up with a "p-value," which is basically a number that tells you how likely the phenomenon you observed was just a random coincidence. If you can’t produce a reasonably low p-value (i.e. it’s not likely a coincidence), you conclude nothing, and the result is "insufficient evidence."
I suspect this is what prompts scientists and engineers to demand proof upon hearing stories of discrimination. How can anyone make conclusions without basing them in undeniable truths, or at least truths bounded by a p-value?
Fair point, but there’s one problem: It's impossible to prove any instance of discrimination.
IT'S IMPOSSIBLE TO PROVE ANY INSTANCE OF DISCRIMINATION.
No matter how sexist or racist a person seems to be, there's no way to rigorously prove it. We don’t have a blueprint for the human brain’s inner workings (at least not yet). We can't get a computer printout of the exact line of thinking that prompted someone to say something ignorant and hurtful. We have no way of truly knowing a person's motivations behind their actions.
As a result, "Are you sure it was sexism?" is a pointless question because the answer is always the same: No, I’m not sure. I can’t be sure. It’s impossible to be sure.
Engineers solve technical problems at work every day, so it's tempting for us to solve all of life's problems using the same approach. But it's ridiculous to apply engineering techniques to social problems, where it's impossible to get the evidence we would normally demand to solve them. We need a different approach.
Presenting an alternative model
Suppose you’re an engineer working on Gmail. One day, you're checking the application’s usage statistics, and see something alarming: It appears a tiny percentage of users are getting other people's emails. In other words, Jessie sends an email to Alexis, but it goes to Grandpa instead.
You try to "reproduce" this bug, but your own email works fine. And nobody else on your team is encountering this problem. Perhaps the stats are wrong?
Then the reports start to trickle in. Emails addressed to sisters going to bosses; emails sent to friends going to strangers. Within 30 minutes, there are 150 new reports of this problem.
This bug would be declared a "P0,” or an all-hands-on-deck emergency.
Notice how this is treated differently than a usual bug: We’re acting on speculation, rather than fact. It's possible that every one of those 150 bug reports is invalid. Maybe Jessie and others are misremembering whom they addressed their emails to. Perhaps the stats really are misrepresenting the situation, and it’s just a big misunderstanding. That's all technically possible, but a far more plausible explanation is that something's actually broken.
To my fellow engineers: The systematic oppression of women and minorities is a P0. The innumerable studies and statistics are the data. The overwhelming anecdotal evidence are the bug reports. We can only conclude, then, that something is terribly broken.
We can only conclude, then, that something is terribly broken.
It doesn't matter that you can't reproduce the bug yourself (“But I've never encountered any sexism in tech”).
It doesn't matter that the absolute number of occurrences are low ("But it's really unlikely that you'll be wrongfully arrested”).
It doesn't matter that you haven't figured out precisely what’s happening ("How can you be sure he was being sexist?").
This is an emergency that needs to be solved as soon as possible, not later on ("We just need to wait for the racists to die off”).
Stories of discrimination represent stories about a much larger P0 emergency—and there’s overwhelming evidence that this P0 is real. As the listener, you don’t have the authority to close someone's personal experience as "Invalid - Not Really Sexism." Unless you're a human-resources rep addressing a particular accusation of workplace discrimination, you have no business challenging anyone's experience.
That said, hearing a first-hand account of institutionalized oppression may be uncomfortable. You may not know how to respond. Luckily, this has an easy fix. Try: "Thank you for sharing." Or, "That's awful. Is there any way I can help?" Or, "I'm embarrassed to admit I don't know much about this, and I want to learn more. Can you suggest any educational resources?"
Acknowledge the P0, respond with empathy, then enjoy the rest of your lunch.
