Two leaders look elegant. Three are often more durable.
A leadership duo at the top feels reassuring. It suggests balance, complementary strengths, and a fair distribution of power. In many companies, the idea sounds almost self-evident: two leaders must be better than one. You divide expertise, share the load, and reduce the isolation of a single boss.
The issue is that what looks balanced on an org chart does not always remain stable over time.
A study published in Academy of Management Discoveries on December 5, 2025, by Leslie DeChurch, Alina Lungeanu, Megan Chan, and Noshir Contractor examined a question that is rarely explored with this level of originality: how stable shared leadership actually is over time. The researchers combined two worlds that seem impossible to compare: 13 NASA analog crews and historiometric data from Ancient Rome between 32 BCE and 491 CE. Their central finding is striking: two-person shared leadership arrangements are the least stable. By contrast, a single leader or a collective of three or more leaders is more likely to endure (Academy of Management Discoveries, Northwestern University SONIC).
In other words, the question is not simply whether leadership should be shared. The real question is the geometry of that sharing.
With only two, everything can become a duel
In a pair, the structure carries a hidden weakness: it tends to polarize almost everything.
Who decides when the two disagree?
Who embodies the vision when signals become contradictory?
Who represents real authority in the eyes of the team?
Who steps back without feeling diminished?
With only two people, shared leadership can quickly drift into mirroring, rivalry, or constant compensation. Even when both are highly capable. Even when they respect one another. Even when their expertise is truly complementary.
Because in moments of tension, there is no third point of stabilization.
That is what makes the finding so useful. A duo is not necessarily ineffective. It can even work very well in a short phase, in a crisis, or in a tightly bounded mission. But in terms of stability, it seems to carry a specific structural fragility. The authors explicitly refer to “dyadic instability,” a form of instability unique to two-leader arrangements (Academy of Management Discoveries, Megan Chan).
Why three changes the dynamic
At three, something fundamental changes.
The system moves beyond a permanent face-off. It introduces the possibility of regulation. A third leader can slow things down, reframe, arbitrate, restate, redistribute the conversation, and absorb an emerging tension. The structure becomes less binary. It breathes more easily.
That intuition is not new in sociology. Georg Simmel argued long ago that a triad can be more stable than a dyad because a third person can help mediate conflict when the first two clash (OpenWA Sociology).
What is new here is that the same pattern appears in two extreme environments: crews preparing for Mars-like analog missions, and power structures in Imperial Rome. When the same structural tendency appears in such different settings, it becomes hard to dismiss it as coincidence.
From Rome to Mars, structure matters more than most companies admit
The most practical value of this research is not academic. It is managerial.
In many organizations, power is still designed as if structure were just packaging. Boxes move, co-leadership roles are created, tandems are named, responsibilities are layered, and then everyone hopes personalities will make it work.
Quite often, the reverse is true.
Structure creates part of the behavior it claims only to organize.
That directly echoes what I develop in my book, chapter 11: organizations and structures are never neutral; they shape friction, arbitration, learning speed, and the ability to experiment. P001-304-9782100876556_ep05 – f… P001-304-9782100876556_ep05 – f…
When a company creates a power duo without clarifying decision rights, escalation rules, time horizons, acceptable disagreement, and how the team should read authority, it is not creating fluidity. It may be creating slow tension.
And that tension is expensive:
misalignment,
mixed messages,
slower decisions,
political fatigue,
and eventually talent erosion.
Shared leadership is not the problem. Poor design is.
Precision matters here.
This research does not say that shared leadership is a mistake. It says that not all forms of shared leadership are equally robust.
That is a crucial difference.
In fact, the broader literature has long highlighted the potential benefits of shared leadership: stronger information flow, trust, coordination, collective intelligence, and adaptability in complex environments (Oxford Research Encyclopedia of Business and Management).
But this newer work adds an uncomfortable reminder: a model may look attractive in theory and still be fragile in its real operating geometry.
So the question is not “shared or not shared.”
The question is: shared by how many, under which rules, in what architecture, for how long, and with which stabilizing mechanisms?
What leaders should ask before creating a top duo
Before installing a co-leadership pair, some questions matter more than a governance workshop:
Do the two leaders have truly distinct territory, or blurred boundaries?
Who makes the call when speed matters?
How does the team interpret real authority?
What happens if both leaders want to embody the same symbolic role?
Is the duo a long-term design choice, or just a temporary political compromise?
And above all: was the pair designed to last, or simply to calm the present?
Many duos exist because they postpone a harder decision. Yet an org chart designed to avoid short-term discomfort often creates greater discomfort a few months later.
A broader lesson about governance
What this study shows goes well beyond leadership.
It reminds us of something many companies forget: performance is not only about individual talent. It is also about human architecture.
We love discussing the right profiles, the right skills, the right leaders.
We talk much less about the right number.
And even less about the right configuration.
Yet inside an organization, the shape of power behaves like an invisible technology. It accelerates or slows down. It creates flow or rigidity. It protects innovation or suffocates it.
A duo can be brilliant.
A well-designed trio can be more resilient.
And a single leader can sometimes be clearer than poorly designed co-leadership.
The useful lesson is not ideological. It is structural.
Before searching for the right people, it is sometimes wiser to first design the right geometry of leadership.
References
(Academy of Management Discoveries) = https://journals.aom.org/doi/10.5465/amd.2025.0028
(Northwestern University SONIC) = https://sonic.northwestern.edu/home/projects-2/past-projects/mdt/
(Acta Astronautica / ScienceDirect) = https://www.sciencedirect.com/science/article/abs/pii/S0094576525005314
(Oxford Research Encyclopedia) = https://oxfordre.com/business/display/10.1093/acrefore/9780190224851.001.0001/acrefore-9780190224851-e-442
(OpenWA Sociology) = https://openwa.pressbooks.pub/sociologytcc/chapter/6-2-group-dynamics-and-behavior/
(Megan Chan) = https://megan-a-chan.com/shared-leadership-from-rome-to-mars/
