Don’t Be a Bystander

My friend Cate wrote a few tweets about bystanders. This is a little story about a time when I worked with some bystanders and why little things matter.

When I started one job, I found that our existing software had a diagnostic tool that took some input parameters, you clicked submit and it showed you what our service would return for those inputs. Like a good tool, it came pre-populated with inputs and the resulting outputs that demonstrated how it works. But because humans can be thoughtless, someone in the past thought it would be cool to choose inputs that resulted in a picture of a scantily clad woman. It wasn’t a good choice to demonstrate how the tool works because due to data decay, only that picture showed. Nothing else interesting would show up and most of the time people used the tool for other outputs. So. Just a picture that looked like it came from the cover of a bad porn DVD. No, I did not work for a porn company.

Now this tool was used by our team. It was also used by many other people at the company when they were trying to figure out if the problem was our software or systems we integrated with. Whenever anyone at the company opened up this tool, they were presented with a picture that reduced women to their sexuality. At some point, I quietly changed the default for this tool so it was much more neutral and so that it was actually useful: all the useful outputs actually existed, not just the image. No big deal, right? Well, consider the impact.

Many people around the company had to use this tool to debug issues. Some of these are women. I’m kind of jaded and roll my eyes at a lot of bullshit, but a young woman still trying to figure out her place in the world made to look at that every day and decide this place – the company, tech – isn’t for her. After all, even the tools she needs to do her job reminds her that a woman’s place is as a sex symbol, not as a professional programmer or product manager. Many people are junior and don’t know what is appropriate in the work place. How many people just entering the industry saw that tool and would now consider it an okay thing to do, creating similar environments elsewhere? Some folks come from cultures that are fairly conservative about nudity (some of course don’t). How many of those people were made uncomfortable with this tool but didn’t feel like they could speak up because they’d be seen as prudes or otherwise not fitting?

For how many people was that a “little” thing that contributed to them feeling uncomfortable at work and ultimately leaving? They might not even consciously know the exact things that made them feel uncomfortable and not fit in. My team was probably responsible for a little bit of that.

A few years later I was talking to a former teammate (we’d stayed friends). We had both left the original team by then. He asked me directly about sexism, etc. in tech. I was much more vocal about it at that point. To be honest, when someone actually makes an effort to ask about the topic, they automatically get a lot of credit with me. I reminded him of this tool and why it was a problem. It just wasn’t something he’d thought about. I think it was that way when he joined the team. He was a bystander then but so was everyone else on the team. I don’t say this to shame him (if he reads this, he’ll recognize this story). I bring up this story and his reaction because it’s a good example of how easy it is to be a bystander. To be a generally good person who’d never intend to hurt anyone and still be a bystander.

To not be a bystander requires saying or doing something at the moment when the harm may not seem that great and the risk to yourself, socially or professionally, seems immediate. “Just a joke” or “just one time” is easy to say if you’re worried your team will laugh at you. But over time that stuff does matter and becomes harder to shift. If the team has tolerated little sexist jokes for months or years, then when you do speak up you’re going to sound like a hypocrite. That only encourages you keep your mouth shut.

How much harm could have been avoided if someone had looked at that tool when it was created and suggested different default inputs be chosen? They wouldn’t have had to make a big deal about it. No need to say “this is sexist bullshit”. They could have just said “maybe we should use example data that is more neutral.” Not being a bystander doesn’t have to be about making a statement. I doubt the developer would have fought a different choice. But someone had to just push back, a little bit. And no one did.

Good Intentions Fail at Scale

You’re an internet company. You have a hot product and a busy website viewed by thousands a minute. Unfortunately it’s having outages regularly. The website is slow. It goes down at least once a week. The backend systems are even more a mess. We’re going to fix this you say! We’ll send all the teams to a training provided by some vendor. They have a mediocre flash-based training tool that barely runs on modern browsers. It’s not ideal but that’s what we have. We’ll give the managers a bit more training maybe. Some random folks from various teams will get some mentoring from a few of our more senior engineers.

That’s absurd! No company does that. At your company, you have incident post-mortems. You insist teams have metrics and dashboards that surface uptime, latency and other relevant metrics to senior leadership. Individuals are empowered to look at technical challenges, relevant metrics and suggest ideas to their managers to improve them. The idea that some stupid infrequent training would solve scaling and availability challenges is absurd.

But consider: some infrequent mediocre training is what nearly all companies do to handle diversity related challenges including hiring, attrition, harassment, inequity and more. At most companies, the metrics are not visible: you work at a “good” tech company if you know basic numbers like percentage of black people or women in technical roles (good luck finding metrics for “black women in technical roles” much less “black women in senior leadership of technical orgs”). Managers and teams are not held accoutable for attrition as they would be if their team’s software was constantly failing. There are no post mortems to explain why such-and-such org can’t seem to retain women for more than a year or so. If there are more programs than just training (e.g. mentorship programs), they aren’t available to everyone, many aren’t even aware of them and they usually aren’t measured for effectiveness.

Why is that? The people running these companies are not stupid. The middle management aren’t stupid. People in general aren’t stupid.

What people do have are “good intentions”. In a small company, you can mostly rely on good intentions to maintain culture. If you start a new tech company now, you can chose to make an effort to be inclusive from the start. Your communicative overhead is relatively low, the number of folks that have to adopt and believe in your plans for company culture are small and good intentions will get you pretty far in building something better.

But if you’re hundreds or thousands of people strong? Transmission of culture is hard. Only part of it will happen. Even if you start out with the intent to build an inclusive culture that can actually retain a workforce similar to humankind, it won’t easily happen as you grow. When you’re small, your mechanism for culture transmission can include “one on one meetings with a founder”. When you’ve grown, when you “scale” your company, what is your mechanism?

Good intentions can’t be your mechanism. You’re bringing in too many people with extensive work histories and their own ideas about how to do things. The junior people are just trying to get a handle on working at all, never mind learning all the culture you want them to learn: they are going to learn a mangled version of it. You’re going to have pockets of the company that behave very differently than the founders’ original vision. That’s for all of your culture. If you’re in a typical tech company your company wasn’t founded with inclusivity as a value. If you don’t count on good intentions to transmit values around ownership for production problems, why would you count on good intentions to “fix” your lack of diversity?

So that’s why this story is absurd. No reasonable leader would expect even a good one day training to fix a team’s technical challenges. Even with training, they’d have metrics to track, there would be ongoing coaching up and down the management chain and it would be a regular subject of organization meetings from senior leadership offsites to team’s daily standups. And ultimately failure would have consequences.

But to build a better, more inclusive, culture we see “good” tech companies where leadership hides metrics, barely invest in programs or training, don’t hold people accountable even for egregious failures and shrug and think “good intentions” work. Is it any wonder the “numbers” have barely moved?