Power-Over and Power-With

In the AI Story

Follett articulated the distinction in her 1920s lectures to industrialists who had organized their factories around the Taylorist assumption that efficiency required the concentration of knowledge in management and the reduction of workers to executors. Against this framework, she argued that power is not a fixed quantity to be distributed but a capacity that grows with genuine participation. The argument bewildered her audiences then and continues to bewilder the quarterly-earnings-focused strategists of the present, because it violates the zero-sum assumption that structures most thinking about organizational power.

The empirical test of the distinction is straightforward: observe what happens to organizational capability over time under each regime. Power-over organizations become faster but brittler — decisions at the top propagate through the system with unprecedented speed while the distributed intelligence that would have caught errors, surfaced objections, and identified alternatives atrophies through disuse. Power-with organizations become more adaptive because the capacity to read situations accurately grows with each engagement of situated knowledge from below.

The AI amplifier intensifies the stakes of the choice without changing its structure. When large language models are deployed within a power-over framework, they execute the senior leader's vision with unprecedented scope while never challenging it. When deployed within a power-with framework, they amplify every member's capacity to contribute to collective reading of the situation. The tools are identical. The organizational context determines what the tools amplify.

The Trivandrum case described in The Orange Pill — twenty engineers each operating with the leverage of a full team — is the paradigmatic instance of power-with in the AI age. The boardroom arithmetic that would have eliminated fifteen of the twenty treated capability as a fixed pie to be sliced more efficiently. Follett's framework reveals what the arithmetic concealed: that the interactions between team members were generating intelligence no aggregation of individuals, however augmented, could replicate.

Origin

The distinction emerged from Follett's community-organizing work in Boston, where she observed that neighborhoods which coordinated through shared deliberation generated capabilities that dominated neighborhoods could not produce. The insight crossed from civic to industrial life when Follett began lecturing to business audiences in the 1920s, and it survived her death in 1933 to be rediscovered by management theorists in the late twentieth century as organizational complexity exposed the limits of command-and-control design.

Key Ideas

Power is not a fixed quantity. It is a capacity that grows through genuine co-creation and atrophies through concentration.

Command-and-compliance suppresses situated knowledge. The worker who executes orders without engaging her judgment develops only the capacity for compliance.

Co-active intelligence is emergent. The team operating through power-with produces insights attributable to no individual member.

The AI tool is neutral to the choice. The organizational framework within which AI is deployed determines whether it amplifies intelligence or concentrates dysfunction.

The boardroom arithmetic measures the wrong function. Team intelligence is not the sum of individual productivities but a product of the interactions between members.

Debates & Critiques

Critics of Follett argue that power-with is a luxury available only to organizations with slack resources, and that under competitive pressure the power-over deployment of AI is rational even if regrettable. Defenders respond that the argument inverts cause and effect — that competitive pressure is precisely the condition under which the adaptive intelligence power-with generates becomes most valuable, and that the organizations which concentrate capability for short-term efficiency are consuming the resource they will need to survive.