In a series of studies begun in the 2000s, psychologist Dacher Keltner and his colleagues discovered a pattern they termed the power paradox.
People typically gain power through qualities that make them useful to the group: attention to others, willingness to listen, the ability to understand another's position. But power itself systematically undermines these qualities.
In experiments, people were temporarily given a higher status—sometimes quite literally at random, by assigning them as group leaders for the duration of a task. This was enough for their behavior to change: such participants interrupted more often, were worse at reading the emotions of interlocutors, and took others' perspectives into account less frequently when making decisions. A separate line of research showed that people with a higher sense of power have a weaker ability to automatically mirror the experiences of others—the very foundation of ordinary interpersonal understanding.
The mechanism is related to attention. Understanding others requires effort, and effort is expended where there is dependency. When you depend on someone, you study them: what they mean, what mood they are in, how to talk to them. When they depend on you, that necessity disappears. Attention is redistributed, and the skill supported by it weakens.
This leads to an unpleasant reality for organizations. A leader who truly understood people well as a specialist may understand them noticeably worse after several years in the role—while subjectively convinced of the opposite. The decline is not felt from the inside: it looks like increased efficiency and the ability to make decisions quickly.
This is aggravated by the fact that upward feedback also degrades. The higher the position, the less willing people are to communicate uncomfortable truths. The leader receives an increasingly filtered picture precisely when their own ability to reconstruct what is missing declines.
What is done about this in practice. Channels are preserved where the leader is in a position of dependency—mentorship by a more experienced colleague, participation in projects not as the lead, regular contact with frontline work. Structured bottom-up feedback is introduced, where the risks of speaking up are minimal, rather than relying on an open-door policy. Hypotheses about people are verified directly rather than taken as obvious. And the length of time spent in a leadership role is treated as a risk factor for the quality of people-related decisions—not as tenure confirming competence.
Power is gained through the ability to understand. Retaining power does not require this skill—and therefore does not preserve it.