If you're so smart as a leader, you'll keep it to yourself
A team in the AI field recently made a hire: a machine learning expert who is a recent computer science graduate who interviewed brilliantly and has a track record of impressive research projects and demonstrations. The candidate seemed to the team a great fit to join a group of mostly hardware experts, expanding the nascent software group. It’s not easy hiring developers with data science and machine learning algorithm skills, as the practical application of new research is still in its infancy with skills centered in academia or at powerhouses like Google and Microsoft.
After several weeks, management was frustrated with the new hire. The developer seemed slow to deliver on their assigned workload, their code wasn’t well organized, and their work was difficult to integrate into the team’s evolving software foundation. The management team began to institute tight, narrow schedules for the new developer and made it clear they were being watched closely, with their output measured and reviewed on a daily basis. This created a hostile atmosphere in which the employee felt unsupported and nervous, hardly able to deliver peak performance.
In the meantime, another developer was thriving, delivering high-quality work on a steady schedule, and building processes for the wider software team to follow. This developer understood the difficulty the new hire was experiencing and felt able to teach the development process by which all team members could increase their effectiveness.
The management team did not enlist this developer to assist with the “problem” employee.
This raises several questions: Does a technology team leader have to be the smartest person in the room? Do leadership qualities contain elements of intellectual domination or intimidation? Are there “softer” attributes that make up a great leader?
Many of today’s influential thinkers push the idea that yes, the leader should in fact be the smartest person in the room and should be the primary influencer of ideas and attitude within a team. The team is challenged to meet the leader’s intellectual capabilities, striving to find ideas that survive the leader’s challenge, to compete among themselves for the leader’s notice, and to position themselves for promotion through the emulation and improvement on the leader’s qualities.
There’s a Darwinian feel to the internally-competitive environment such team structure imposes. Similar to the culture at Microsoft back in the early 1990s under Bill Gates, or under Steve Jobs at Apple, this represents a cult of leadership. The success of those two businesses tend to serve as models for those who don’t recognize the historical uniqueness and think it can apply to mere mortal managers.
Technical recruitment, specifically within the area of machine learning and data scientists, presents a serious challenge. These two skill sets (you might rightly think of them as one) are in exceedingly high demand. You can imagine that the largest sources of new recruits are from academia or recent graduates of computer science programs with an emphasis on machine learning. High demand in an area of low supply means that a large number of recruits are going to be new to the professional software development industry.
What sort of teams are going to be the most successful in working with these new recruits?
The team in our example made a hire based on demonstrated skills, but the managers assume that with domain skills comes, necessarily, team dynamics skills. They view things through purely an intellectual lens without recognizing that it’s incumbent upon them to provide training in team software development practices to new hires who have the rare skill of machine learning but little practical team experience.
They compound their errors in also failing to recognize another employee who demonstrates natural software management skills. By pairing that employee with the struggling one, they could be on their way to huge organizational success: they simultaneously train a new employee in team dynamics through the development of a new manager. Instead, their software team is stymied by the manager’s myopic view of dominant leadership.
The lack of machine learning and data science experts blends into the notion of leadership and team building. It’s a fascinating confluence of two problems in the industry. If organizations are built on the notion that the leader has to be the smartest person in the room, does that mean they have to be an expert data scientist? This is unrealistic given the current state of the industry.
If a team is built under leadership that emphasizes internal competition over cooperation, with a leader who imposes will through intellectual and social domination rather than through collaboration, the result will be a dysfunctional group that cannot deliver on strategy, no matter how good it might be. While a Darwinian model of management may be temporarily effective, it’s long term destructive.
An alternative model for effective leadership is one where the leader serves the role of conductor, teacher, or mentor. These leaders bring forth the best ideas without regard to ownership. The best managers are those so secure with their sense of self that sharing their skills and knowledge is second nature to them. They encourage team members to rise in stature by the quality of contributions as well as their encouraging support of their teammates.
Good leadership starts with empathy and is bolstered by self-confidence. Teams grow, both in their effectiveness and in their ability to evolve and expand new leadership roles, by working cooperatively and directing their fierce competitive attitudes not inward toward teammates but outward to the competition. A leader is more effective when the legitimacy of their position is established organically through unspoken recognition than through demonstrations of dominance, no matter how subtly that dominance is expressed. Leaders are leaders not because they’re the smartest people in the room, but because they bring out the best in their team with the resulting whole being greater than the sum of its parts.