Meetings -- by David J. Deming, Katrine V. Løken, Alexander Willén, Yaling Xu

TEXT START: Why do we have so many meetings?

The Dissection

The paper quantifies a coordination overhead that firms have normalized: meetings consume an average of 12 percent of work hours and 14 percent of firm wage bills, based on more than 9,000 Norwegian workers linked to employer–employee data. Its real function is to measure the price of human coordination while treating the wage-labor firm as the permanent container. It identifies a costly symptom, not the approaching death of the system that produces it.

The Core Fallacy

The implicit fallacy is confusing current necessity with structural necessity. Meetings are necessary today because planning, problem solving, information sharing, and project coordination remain distributed across human workers and bureaucratic hierarchies. Under the Discontinuity Thesis, that condition is temporary. Once AI achieves durable cost and performance superiority in cognitive coordination, meetings become an avoidable labor expense rather than an indispensable workplace feature.

The paper’s cost calculation therefore understates the strategic threat. The issue is not merely that meetings waste 14 percent of wage bills. It is that the coordination layer itself becomes automatable, severing another major justification for mass cognitive employment.

Hidden Assumptions

• Human workers will remain the primary agents of coordination.
• Firms will continue to organize production through synchronous human discussion.
• Meeting-heavy roles retain economic necessity because their coordination tasks cannot be decomposed and automated.
• Productivity gains from reducing meetings will be distributed through employment rather than captured by owners of AI capital.
• Preserving consumption and firm output is equivalent to preserving productive participation.

None of these assumptions survives P1–P3 intact.

Social Function

Classification: partial truth and transition management.

The evidence is real: meetings are common and expensive. But the framing domesticates the implication into an efficiency problem—reduce unnecessary meetings, improve planning, reclaim work hours. That is a managerial lullaby. It suggests the institution can be repaired when the deeper process is replacement: AI will not merely make meetings shorter; it will absorb much of the coordination work that made the participants economically necessary.

Meetings will persist temporarily as lag defenses: legal accountability, political signaling, trust rituals, status contests, and human verification around machine-generated decisions. Those are hospice functions, not proof of durable demand for human coordination labor.

The Verdict

This is a useful measurement of the old order’s coordination tax, but it is not a survival argument for meetings or for the workers who attend them. The paper records the expense of human coordination just before AI turns that expense into a target. The meeting is not the core institution being preserved; it is another visible piece of cognitive labor waiting to be stripped from the wage circuit. As automation advances, surviving participants will be Sovereigns, indispensable Servitors, or transition intermediaries. The rest will be classified as overhead—and overhead gets removed.