We’ll talk about Ethics in Data Science, but first we need a firm grounding in ethics in general. When we announced this talk on meetup.com it had the title “10 Common Ethical Errors in Data Science.” When I sat down to prepare the talk I realized that was a horrible title.
If I were to give you ten common ethical errors in data science, I would risk committing an ethical error. I would be doing you a disservice. So, the very first thing we need to do is change the title.
Instead, I’m going to share with you something much more important. I will show you how I make ethical decisions. Our culture and profession of data science is young and evolving. The following is how I try to stay ethical while enjoying the ride.
Ethics: A Data Scientist’s Perspective
A quick disclaimer: I make a point not to reference anyone or any of my clients in this talk. These views are mine alone. These are my personal opinions on the topic of ethics.
I try not to give you empty advice; something I haven’t tried before. This is based on what seems to work well for me. What works for me may not work for you. Your mileage may vary. I hope you try some things out, keep what you like and discard the rest. I don’t mind if you use any of this for yourself. If you do share something or find it valuable I’d appreciate hearing about it.
Before I tell you more, I need to establish some base assumptions. These are assumptions for this talk only. I realize these are controversial ideas, so to avoid the debate let’s agree on some terminology. In this talk, I define ethics as “knowing right from wrong.” It’s simple and easy to remember.
In our conversation, I try to use the term ethics instead of morals. I appreciate that places like Wikipedia explain that there’s a difference between the two. For this talk, if I use the word “morals” please think of it as a synonym of “ethics.”
If you don’t know this already, you’ll realize by the end how murky the line is between what is legal and what is ethical.
Base Assumption : Ethics are Individual
The final point is that external or outer ethics informs a person’s internal ethics. What do I mean by that? There is you the individual, and when I say internal or you see “self” in this slide, that’s what I mean. A person doesn’t develop a sense of ethics or principles or values in a vacuum. We use the examples of the outside world to build them.
By tribe I mean the groups you relate to. Your family, your friends, your coworkers, your neighbors. Whatever group of people you interact with. You know the names of most of these people.
The larger circle represents the ethical values of the greater community with which you’re affiliated. This is the circle labeled “society.” This could be your state, your city, your geography, or the entire planet. The country you live in. The industry where you work. You don’t know most of these people, but you identify with them. You feel you belong to these groups in some way.
I could have drawn many more circles on this chart. Each larger circle would represent a bigger less familiar body of people and ideas. I could have expanded the circles to include humanity or the universe. We can think of those as lumped in with society. If you get the idea, we can represent it with these three circles.
The self-circle is not completely isolated. A circle is influenced by the other circles. Self is more influenced by tribe than by society. Keep these relationships across the borders in mind.
Cost of Fixing a Mismatch
This next diagram has some problems. Don’t think of it as a unified theory of everything. With a little effort, I’m sure you could come up with a more robust version (and if you do please share it with me).
In life, you are likely to come across a mismatch between your personal ethical values, and those of one of your tribes. This tribe could be your neighbors, classmates in school, or co-workers. The most common example is that you find out that one or more tribe members did something that violated your personal sense of ethics. Worse is when the rest of the tribe seems OK with it.
How to Consider a Mismatch
There are at least two dimensions to consider when thinking about this mismatch.
- What it means to you in the long run. This is the total potential cost to you as the individual. Do the actions make you uncomfortable or feel guilty? Would they put your livelihood at risk? Do they violate criminal statutes and could result in you facing jail time? That’s what I mean by total potential cost of what a mismatch means for you. It may cost a little guilt, or a lot of jail time.
- The amount it will suck to change right now. These rows represent the short term effort required to resolve the mismatch and at least lower the total potential cost. In my mind, there are really easy options that cost relatively little. If the misaligned (or even maligned) tribe is your group of friends, make new friends! This costs almost nothing.
The cost of letting the mismatch fester and doing nothing could cost a lot. You could desensitize your personal sense of ethics and become less ethical yourself. If it’s illegal as well as unethical saying nothing could result in you being viewed as an accessory or accomplice.
The Cost of your Consideration: An Example
You could try to fix the external and depending on your approach this could also result in a huge cost to you as an individual. There’s an interesting article Malcolm Gladwell penned in the New Yorker comparing Daniel Ellsberg with Edward Snowden. Gladwell argued that both saw something wrong and tried to fix it. Who they were and how they approached the problem is the interesting part.
For the purpose of this blog post, it doesn’t matter whether you agree or disagree with what they did, or how they did it. What matters is that they believed there was a mismatch between what they thought was right and what a larger circle thought was alright. Addressing the mismatch was not easy for either to do. They tried to fix the mismatch at great personal risk and cost.
Where to Begin
If you encounter a mismatch the place to start is not the bottom left corner that lists low cost, low effort options. To fix a mismatch is not about working on the easiest, lowest cost option. Ethics is not about picking the low-hanging fruit. The right starting point is in the middle. Take time to reinforce, re-clarify, recalibrate your internal sense of ethics.
Always Start From Within
That leads me to the key takeaway I want you to have by the end of this post and its series. Beware of any misalignment between an inner circle and an outer circle. A misalignment can mean trouble. That trouble could manifest as a little guilt on one end of the extreme, on the other it may result in a lot of jail time. Even though the outer circles are harder to change, the most important thing is to start from within. Establish your internal sense of ethics and build it outward. Always start from within.
Here is one more consideration when trying to decide on your own code of ethics. Many years ago I was volunteering for Hale and Dorr, now known as WilmerHale Legal Services Center. During orientation one of the more seasoned professors/advisors shared the hard-bound copy of the code of conduct and that following them was not optional. He said building a reputation of someone with integrity takes a lifetime. Losing that reputation only takes a second.
That didn’t sit right with me. I struggled with the idea of integrity and reputation being so ephemeral for a long time. It wasn’t until years later that I was able to reconcile this. The resolution came from the following excerpt given in a speech by the late UCLA basketball Coach John Wooden:
“Your reputation is what you’re perceived to be; your character is what you really are. I think that character is much more important than what you are perceived to be. You’d hope they’d both be good, but they won’t necessarily be the same.”
Reputation is important, but character even more so. This is important whether you are looking to do the right thing as a good data scientist. It’s important when deciding to do the right thing as a good citizen. And important if you’re looking to do the right thing as a good human being.
There are a lot of ways to pontificate on ethics in a tribe or society. Regardless of this please remember to start from within. Start with character, and the rest will follow.
This is part of a series. It is based on a talk titled “Ethics: A Data Scientist’s Perspective.” I gave this talk to the Ethics in Data Science meetup group in June 2017 at Gap Inc. HQ in San Francisco. You can view the original slide deck on SlideShare here. This series is a version of that talk, and this is part one.