The recent New York Times article investigates successful group dynamics, and reports on some of the conclusions that Google, following some academics, has drawn about what makes a team successful. I think these insights are valuable for academic teams, and I’d like everyone involved in our research to understand them. However, I think that some of the individual goals due to the structure of the academic setting do require a nuanced view of these insights.
Psychological Safety
The main question is, “what defines a successful team?” with the implicit followup question, “how do we create a successful team?” The NYT article reports on research in Google that converged around the importance of psychological safety. I’m far from being an expert in the field of group dynamics, but it seems from a cursory survey that the primary publication for psychological safety is “Psychological Safety and Learning Behavior in Work Teams” (Administrative Science Quarterly 44 (2): 350–383) by Amy Edmondson at Harvard.
From that article:
Team psychological safety is defined as a shared belief that the team is safe for interpersonal risk taking …[A] sense of confidence that the team will not embarrass, reject, or punish someone for speaking up. This confidence stems from mutual respect and trust among team members.
Core to psychological safety is the idea that it is important for every group member to be comfortable taking risks. The obvious deduction is that repercussions for risks should be non-existent. “…[P]eople tend to act in ways that inhibit learning when they face the potential for threat or embarrassment.” According to the NYT article, teams that exhibit high psychological safety demonstrate two key aspects:
“equality in distribution of conversational turn-taking” - No single person, or sub-group of people should dominate the conversation. Everyone should feel comfortable contributing. Implicitly, this requires people to be willing to take risks when they speak as different people will approach a given topic with different levels of confidence.
“higher average social sensitivity” - Everyone should be in tune with what others are thinking and feeling while in a discussion. When can someone be pushed for more information? When are they uncomfortable with a topic and shouldn’t be pushed? When has something been explained sufficiently? Core to this is that everyone wants to feel like if they’re going out of their comfort zone, that others will know not to push them. I don’t see this much in the literature, but Google saw it as key. I wouldn’t be too surprised if this is because of the traditionally callous and unforgiving environment that traditionally surrounds much of the technical computer science world.
A limited read of the literature seems to suggest that psychological safety is increased when organizational decisions are made in an open and democratic manner (even if the final say stays with a single manager). Leading by example is, as always, a fundamental suggestion for the managers (advisers) of the group.
Academic Research Teams
Academic research teams are quite different from to corporate teams in many ways. I believe these differences require consideration in creating a team psychological safety. I’m going to focus on research teams like my own which is one of many different types of academic teams. Our team is defined by:
A focus on research within a single core software system. Most of the projects involve extending, in one way or another, that core. In contrast, other research teams focus on “one offs” which are single artifacts that investigate a research hypothesis. Put another way, our team focuses on long-term system building.
A pervasive, though secondary, emphasis on continual improvement of the core system to enable sustained innovation. This ensures system longevity, and increases impact over time.
We’re in a small department, so the availability of courses that can help train students is limited. This puts the onus of most training on the team. This, paired with the necessary depth of knowledge, requires a deliberate effort in knowledge transfer.
Factors that differentiate this type of an academic environment from a corporate one include
PhD students are ultimately responsible for their individual achievement. Their work culminates in a doctorate thesis which establishes them as an expert in the world in some specialized topic, and as an independent researcher. The natural structure of doing a PhD makes the team a second order constraint. This is a fact that most healthy academic cultures push against strongly. I strongly believe that the strongest research comes out of teams of researchers that are continually growing together.
PhD students often work on projects that are an offshoot of the core code-base. This is not equivalent to adding a feature, and instead causes a lot of code movement that might conflict with other work. Their fundamental goal is to progress their own research, not to ensure the quality of the main code-base. Research groups that are based on long-term projects (not one offs) have to push back against the natural incentives to ensure long-term impact.
The intellectual progress of the students is primary. They must take classes to achieve a breadth of knowledge that is intentionally outside of their area of specialty. Though this seems inefficient, it is essential due to the nature of research. You often do not know, when solving a problem, what techniques or knowledge will be required, so an understanding of the vast computational toolbox is important.
There is an ingrained, and strong hierarchy in academia. As one of my friends who has a lot of management experience in industry puts it, “if the professor knows what the right answer is most of the time, shouldn’t that be the final word?” This is a natural conclusion that comes directly from the fact that the professors are quite literally teaching students most of what they know about an area. However, it is a dangerous structure that must be resisted systematically. Research is not an endeavor where one person knows all of the answers. It takes creativity, which is often better approached as a collective goal. It takes depth of technical knowledge that must be spread out amongst the group. So academia has a built in structure that must actively be subverted.
These are interesting factors as they show how the obvious incentives for doctoral students both work against maintaining a team-focused environment. They also work against having a flat intellectual hierarchy. Certainly, they complicate maintaining a core piece of software that persists across many graduate student generations.
The key is that all students are involved in a broader research agenda that they buy into when entering into the group. I think that they very quickly understand that the more involved in the team as a whole they are, the more they benefit from the knowledge and perspective of other students. This is where it becomes quite important to create the most beneficial group dynamic possible, and were we can consider psychological safety as a reasonable goal.
Psychological Safety in Academia
First, I want to point out the obvious. Academic teams working in highly technical fields, especially if they are building real systems, don’t have the luxury of being inefficient. We’re already trying to “punch above our weight” by doing high-impact work that provides previously unknown insights to industry and academia with small groups of smart people who are often around only for the medium term (2-7 years). Training is a built-in and necessary part of the system. A PhD is often closer to being an apprentice than being a student, and undergraduate research is an accelerated, and specialized education. Almost by definition, when a PhD student grows their knowledge-base and becomes capable enough to do world-class research on their own, they should graduate and move on to lead a lab themselves. All of this means that the focus must be on small teams, learning quickly, and working together toward a larger goal. Each student contributes to the group goals, and profits from the group dynamics and the previous research that has been done.
My colleagues and I talk a lot about setting an “academic culture” for our research groups, and for our undergraduates. Many of the tenants that we strive for (explicitly or implicitly) have overlap with psychological safety. So what can we do to make sure that our research teams are efficient, while optimizing for psychological safety?
Open discourse - This is almost a direct transcription from psychological safety. Everyone needs to contribute to each discussion. When surrounded by more senior graduate students, it can be difficult for a student to feel comfortable adding to the discussion. This is an extreme version of enabling people to take risks when speaking in the group. They have to take the risk and speak up even while knowing that they are often at a knowledge disparity. The key here is for everyone to realize that when in a group, the goal is to increase the collective knowledge of the group. It is OK if others know more than you. That is almost by necessity. However, you still need to maintain a continual learning process, which means interacting, questioning, and contributing.
Checking our egos at the door - When a lot of smart people are in the room together, it is not uncommon for ego to play a role. This can decrease the chance of others (especially more junior students) speaking out. Key to an academic environment is that there is no monopoly on knowledge, and that it is everyone’s job to raise the average knowledge level of the entire group.
Intellectual curiosity - Key to doing research is a pervasive understanding that there is a lot that we don’t know, and a deep yearning to plunge into the darkness and learn as much as we can. This curiousity must extend to the fact that your fellow team-mates know things, and have experience and perspective that you do not. A key aspect of intellectual curiosity is listening to others, and trying to learn as much as you can from them. This often extends to the future potential of them contributing to your knowledge. In other words, it is important for to help others (especially junior members) to learn so that in the future they will increasingly contribute to the collective knowledge of the group.
Student-led initiatives - Part of having an equal stake in the group’s dialogue is that students have to be willing to have some control over what the discussion is about. Students must self-organize around issues that will benefit the group. Reading groups that focus on surveying current research in areas of collective interest are a perfect example. Such initiatives should be led and driven by students so that they have explicit control over what they get out of the experience. The professors are explicitly just another member in the room and are stripped of any hierarchical status.
I believe that the natural incentives of academia complicate the creation of an academic culture based around goals such as psychological safety. Consequently, it is important that each group member understand the power and benefit of having a strong team, and what is required to create and maintain one.