Intelligence isn't the only quality you want in a data scientist. Here's how to ensure your team members develop their breadth of knowledge and become more effective.
Reality TV is one of my guilty pleasures. I had lost interest in Survivor (a CBS show), but this season caught my attention. One of the twists this time around is that the contestants were initially separated into three groups: beauty, brawn, and brains. And although Survivor tries to keep the competitions balanced with physical, agile, and mental aspects, the brains team is getting crushed; the team is even having problems with challenges in which they should be excelling, like puzzles. This is a perfect example of the risks you can encounter when dealing with really smart people.
Brilliant people have a tendency to sacrifice their breadth of knowledge for an extremely vertical depth of knowledge in one specific area. Sometimes this specificity can be extremely narrow, like someone who spends every waking moment thinking about deep learning neural networks.
As much as I admire this kind of passion, the utility for people like this in the workplace is very low. I've talked before about making sure your data science team maintains a balance of data science skills, and I've talked at length about making sure you have more actors on your team than just data scientists. Now what I'd like you to consider extends beyond this to a more general sensitivity: You must take regular measures to build your data science team's breadth of knowledge.
Too smart for their own good
It takes more than just intelligence to make things happen. I love the scene from The Breakfast Club when Brian says, "Did you know without trigonometry, there'd be no engineering?" And Bender responds, "Without lamps, there'd be no light."
Some PhDs and other very smart people distance themselves from critical skills of effectiveness for the sake of their craft. These critical skills include: communication, social behavior, perspective, emotional intelligence, and adaptation. In Survivor, the problems with the brains team extend beyond just competitions — they can't self-organize. The one person who's desperately attempting to lead this group is completely devoid of leadership skills, and the team suffers as a result.
The reason people develop this way has a lot to do with operant conditioning. Very early in life, they realized they were smarter than others in particular areas such as math and science, but they also realized they weren't so great in other areas such as sports, dancing, or socializing. The reinforcement from their academic accolades coupled with the punishment (ridicule, feelings of insecurity, etc.) from attempting anything else kept them behaviorally motivated down the path of excellence in their specialty. Over time, even something that seems very basic to you and me becomes a bit of a challenge.
That said, don't ever lose sight of the fact that they can solve serious problems you couldn't even attempt to contemplate, so be careful with your judgments. With the right team, you have a lot of power at your disposal, but you must know how to develop people in the right way. For most data science teams, that means going broad — very broad.
Techniques for breadth
Developing your team's breadth of knowledge is a change management exercise that shouldn't be underestimated. I'll present ideas to consider, though, you'll want to enlist the services of a good change management expert — at least initially.
The most important piece of advice I can offer is make sure your data scientists are allowed to grow horizontally in a safe environment. Sometimes, their starting point is very embarrassing (they won't admit it, but they know it), and since they're so smart in other areas, this can be a terrific blow to their ego. Your change management professional should know the best way to plan their development, but here are some ideas to get you started.
1. Run team-building exercises to help them with social skills. Put them through classic social exercises like group problem solving and conflict management. Start with just the data scientists, expand out to the rest of the team, and finally include other key stakeholders in the organization.
2. Encourage them to stay current on world events. Pick a daily news show that everyone should watch and discuss the news at your staff meetings.
3. Facilitate a session where you probe for special interests that team members may already have. For instance, I know a very talented data professional who loves to restore old Volkswagens, and another colleague almost became a tennis professional; if these two were on your data science team, they could learn a lot from each other about their alternate passions.
4. Take them on field trips that have nothing to do with data science. You might schedule a golf day or a wine tasting, or get tickets to the ballet. It's a good diversion from the office, and it will expand their creativity and perspective.
5. Lead by example. Develop your breadth of knowledge and make it a point to engage your team in conversations about history, geography, politics, and sometimes obscure topics like archeology or ontology. Challenge yourself to draw correlations between world events and something more relevant to their work. For instance, how is the recent posturing by Russia similar to the dynamics in some of our work streams? How does a jetliner just disappear from sight, and how is that similar to our information monitoring risks?
It's important to have the best and brightest on your data science team; however, it's equally important they develop a respectable breadth of knowledge. Not only will this increase the team's effectiveness and efficiency, but it will also expand their points of reference for innovative ideas and solutions.
Take time in your next staff meeting to talk about world events, and connect with a good change expert to put a plan in place that supplements your team's deep expertise with more of a liberal education. They may be smart, but you can't get anything done in the dark.