Do the right thing

Do the right thing. In addition to being the name of one of Spike Lee’s greatest films, it’s a motto that a lot of companies use. It sounds simple enough, and the phrase “do the right thing” frequently comes up in business contexts when people talk about strategies that impact users (e.g., “Don’t use people’s data inappropriately!” “Don’t charge hidden fees!” etc). But why should you care? How does learning the “right thing” help make you a better data scientist, engineer, or person in general?

Let me try to answer those questions and few key others through my experience below:

1. Doing the right thing matters
2. What is the “right thing”?
3. A hypothetical scenario of doing the right thing
4. Three key approaches to doing the right thing
5. Revisiting that hypothetical scenario

(1) Doing the “right thing” matters

For anyone who is not a director, doing the right thing isn’t about an organization’s’ corporate strategy so much as it is about how you conduct your day-to-day work and handle difficult situations. And yet this experience is not something you typically gained in school or took a course online about.

Instead, data scientists, engineers, and quantitative folks like myself practice working on problems where you can determine the “best” solution mathematically. This means we often lack experience identifying and doing the right thing in ambiguous scenarios. We make decisions based on data. Which model is more accurate? Which piece of code is more efficient? Which product leads to more clicks? Decisions are easy if they’re quantifiable.

But when someone sends an email with inconsistent asks and a slightly hostile tone, there’s no formula for figuring out how to respond. Determining what the “right” thing to do (and then actually doing it) can feel challenging and awkward. But don’t worry, we’ve all been there, so let’s talk through some of my experiences and mistakes.

(2) What is the “right thing”?

So what does the “right thing” even mean? The golden rule is a great starting point: “Do unto others as you would want done unto you.” I highly endorse this mentality when engaging with anyone in life, but it doesn’t prepare you for when someone does something that irks you. In these tough situations — where your natural reaction is often an emotional one — I use a corollary to the golden rule: Exhibit through your actions how you want others to act. If you want an even simpler way to approach it: what is the nicest and calmest way to handle a situation?

In a professional context, a challenging situation might be one where a colleague is dismissive of your ideas or a supervisor is overly demanding in terms of deadlines. It’s also the situations where a colleague changes their mind frequently, a client sends a heated email, or project ownership becomes unclear. It’s in these situations that doing the right thing becomes most critical. It’s in these instances that you have the power to both lead by example and set the tone of your company’s culture, even as an entry-level data scientist or engineer. Your actions have ripple effects no matter how junior or senior you are. People like to mimic and mirror those around them, so be conscious of how you set the bar for those around you. Your navigation of these situations determines whether your team maintains momentum or loses its way.

(3) A hypothetical scenario

Imagine a circumstance where a colleague sends out a project proposal via email. The proposal has duplicative efforts to your team’s current work and the technical solution your colleague is proposing is not viable. There are two immediate reactions you might have for resolving this sort of conflict:

• Tell them their approach is ill-defined and that they should just fold their work into yours
• Ignore them and let them fail on their own

If you’re a data scientist or engineer, you could empirically prove your solution is better so the first reaction seems obvious. The second approach of not even inserting yourself into the matter is even lower effort. Neither of these approaches feel “wrong” since they require little mental effort on your end and pose minimal risk to your own project.

However, just because an action is easy and low risk doesn’t mean it’s the right thing to do. If doing “the right thing” means exhibiting through your actions how you want others to act, I would encourage you to put yourself in your colleague’s shoes. Think about how you could be convinced that your idea is not an optimal one. What example do you want to set to lead others? Do you want to set a tone of helpfulness or dismissiveness? What behaviors of yours do you want mimicked and mirrored?

By helping your colleague, you save their time and help them grow from a technical perspective. Doing so not only helps them but also has benefits the company financially and culturally.

So what is the right thing to do? Effective and successful navigation of contentious scenarios like this one often come down to three approaches:

1. Don’t overreact
2. Don’t assume malicious intent
3. Giving constructive and empathetic feedback

(4a) Don’t overreact

I will be honest — I am a passionate person. I laugh loudly, am visibly enthused when things go well, but have to work extra hard when difficulties arise or things don’t go as planned. This means I have to be conscious of my emotions and how they are steering my responses to situations.

In my last role, every time an outage occurred or metric dropped one of the lead PMs would frantically gather people together and start talking about how “all eyes were on us” and that “we immediately need to fix it”. I found this incredibly stressful and unproductive. In these circumstances I didn’t possess the right mentality to problem solve or give the PM the proper feedback.

The best advice my wife ever gave me is to learn when to (temporarily) walk away. I had to train myself to walk away and not respond in the moment. So when instances like the one above occurred, I took my time. Once composed I would write an organized and even-tempered description of the actions I thought were necessary to get to the root of the problem. I combed every sentence I wrote to ensure blame was never a sentiment I expressed. It was a ton of mental work, but it kept me from escalating the level of agitation I already felt. I was trying to do the right thing.

Overreacting distracts you and your colleagues from your collective goal. When there is a heated email exchange or meeting, don’t feel you need to respond in the instant. Take a couple of hours to distract yourself with other activities so that you come back to the table with a fresh mind. Stepping away to reset has been critical for me in resolving both personal and professional conflicts.

Because in these situations, you’re ultimately on the same side. You all have a mutual interest in the company’s success, so take a moment to remember that. When in doubt, explicitly write down your team’s goal and remind everyone of that shared motivation.

(4b) Don’t assume malicious intent

The second approach is not immediately assuming there is malicious intent behind a colleague’s actions. As soon as you do, the relationship is irreparable. Every action they take will be cast in that light. Let me explain this probabilistically (yes, this is about to get very nerdy).

You likely interpret a colleague’s underlying intent as malicious because of the actions they take. In short, you think P(Malicious | Actions) is high. But let’s apply Bayes’ Theorem to look at your underlying assumptions:

The only number I care about in this equation is the one in the top right: P(Malicious). What do you think that number is? For any given person that you work with, what are people’s natural likelihood to be malicious? Maybe you’re different than me, maybe I am naive, but I have to believe that number is very very low. If you’re of the belief that a lot of your colleagues are inherently malicious, then I would strongly encourage you to (a) reconsider the company you’re in and more broadly (b) reflect on whether you might be projecting negative intent.

You often don’t intend to be malicious or you may even justify some low levels of maliciousness by believing the gains of your actions offset the damage (e.g. aggressively asking someone to deliver within a short turnaround). Once you deem someone inherently malicious, you won’t be able to work with them. Before you do, give them feedback and the opportunity to convert on that feedback.(

Really? You’re supposed to take the time and effort to help improve the behavior of someone who is coming across like an ass? Yes. It’s the right the thing to do. I know this step can feel super intimidating and difficult, but it’s absolutely necessary for resolving (and preventing future) conflicts. Let’s talk about the best way to do it.

(4c) Give constructive and empathetic feedback

Creating a culture of feedback

Before you ever get to the point of giving someone tough feedback, you have develop a rapport with them. They have to see your criticisms as coming from good intentions. Whenever you start a project, schedule 1:1s with the people you will be working with, even if it’s just to talk about non-work stuff. That initial trust you build will be essential later on.

A key way of maintaining healthy bidirectional feedback is to adhere to a good praise to criticism ratio. In short, for every 1 instance of constructive criticism find 5 instances of positive reinforcement. This ratio sounds unattainable, but it’s a lot easier than you think. Every time you review someone’s slides or documentation, leave a few comments about something you like, even if it’s as minute as thanking them for including references. If you’re doing a code review, drop +1’s wherever you see good commenting or formatting. Even in the weaker pieces of work, you should still be able to give praise for the problem they are trying to attack and they eagerness to do so.

You must also be open to critical feedback about your work. This harkens back to the idea of demonstrating behavior you want to see. Be proactive in asking for tough feedback, particularly if you think things have not been smooth, and their critiques will likely be honest and valuable.

Giving tough feedback

So how do you actually have a conversation about a tough situation? Start by framing the problem as how it’s perceived or how it makes you feel, as opposed to stating it as fact. Describing a scenario as your impression can feel really weird to data-driven people like myself. We’re used to describing things as either objectively true or false. A certain algorithm doesn’t appear or feel more efficient; it is either is or isn’t. We can calculate and verify.

But this isn’t true of people. How you read a scenario is not the absolute truth and you need to express that in your feedback. Assure them that you are not assuming malicious intent and that judgement has no part in your understanding of the situation. Examples:

• “You earlier said X, but your logic now points to Y, are you sure? Your recommendation feels inconsistent, can you help me understand?” is far more effective than just saying “I don’t understand, you’re being inconsistent.”
• “I know you don’t intend it, but it comes across like you aren’t valuing the teams’ ideas or opinions. Can we take more time to talk through or just document others’ ideas that we aren’t pursuing?” explicitly attempts to remove blame from the equation.
• “It feels like we are not moving very quickly and may miss our quarterly deadline. Do you agree? What do you think we can do to accelerate progress or prevent further setbacks?”

The other aspect of these examples is framing the improvement as something collective as opposed to just their responsibility. Making the effort to deliver your feedback in this manner may feel overly cumbersome or like you’re watering it down, but I assure you it will increase the likelihood that the person (1) actually hears you and (2) actually acts on it.

(5) Our hypothetical scenario revisited

Going back to our original situation, with the tools you have now: what is the right thing to do? Take a moment. Stop and jot down your thoughts. Again, if you were to were attempting to take the nicest and most tempered approach possible, what would that look like?

There isn’t a single approach, but I would advise that you…

1. Do NOT respond immediately. Compose your thoughts, regroup, and assume that they do not have malicious objective.
2. Think about where your goals align and what the ideal outcome would be between your two projects. Orient towards achieving that outcome. When starting your initial conversation (via email or in-person), start by acknowledging these common parts.
3. Start from a place of concern and inquiry. Ask them if they are familiar with your initiative and whether they see any opportunities for collaboration? Do they have any concerns about duplication?
4. When you agree that there is some redundancy, frame finding a solution as a joint effort… “how can we combine certain parts to ensure both of us don’t end up wasting time?”
5. Provide technical feedback in a concerned/worried tone “If you use that technique, I’m concerned you won’t account for … , what do you think?”
6. If you are unable to resolve or if they respond with hostility, do not match their tone. Keep the mindset of being as calm as possible at all times. The moment you lose your temperament, you also lose the air of objectivity and rationality. If things reach this level, loop in your manager to help address the situation.

Yes, I just asked you to take up to six steps instead of just potentially ignoring the email and moving on. You don’t always have the bandwidth to resolve every conflict like in this way, but the more you practice, the easier it becomes.

At the end of the day I am advocating for more work on your end because it’s what’s best both for your company, your coworkers, and you. We all have posited ill-guided research proposals, sent emails with improper tone, or stepped on someone’s ideas before. The best thing you can do is take your time, be empathetic, and try to do the right thing.