Reversing the Division of Cognition
Reversing the Division of Cognition In order to streamline our organizations, we have inadvertently made ourselves less intelligent.
Reversing the Division of Cognition
In order to streamline our organizations, we have inadvertently made ourselves less intelligent. We have abstracted reality into smaller and smaller pieces, and those pieces do not always reassemble cleanly. The organization got more efficient and less coherent at the same time. We have been calling that progress.
Pinhead Logic
Adam Smith watched men making pins. One man working alone could make twenty pins a day. Ten men, each doing one step, made forty-eight thousand. The math was real and the logic was sound — for a pin factory. The product was simple, the process was fixed, and the fragments reassembled perfectly because nobody was asking the pins to make a decision.
Frederick Winslow Taylor took the same logic and applied it to cognitive work. Separate the thinking from the doing. Have experts design the process, have workers execute it. The efficiency gains were real. The cost took longer to show up.
Neither Smith nor Taylor had a model for what happens when you apply specialization at scale to organizations whose primary product is not pins but decisions — organizations processing ambiguous signals about a reality growing more complex faster than any specialist can track.
Scale Changes Everything
At small scale the fragments find each other. The people who divided the problem share enough context to reassemble it. They sit near each other. The translation losses are small.
Add scale and the specialists stop sharing context. They develop their own language, their own metrics, their own definition of a good outcome. The finance team optimizes for margin. The engineering team optimizes for elegance. The sales team optimizes for volume. Each is locally rational. The aggregate is frequently incoherent.
Add more scale and you need managers to coordinate the specialists. Managers do not do the work — they translate between people who do. Each translation is lossy. By the time the signal from the person who observed the situation reaches the person deciding what to do about it, it has been through enough compressions and reframings that it barely resembles the original.
This is not a failure of individual managers. It is what the structure requires. The org chart is a signal processing system and like all signal processing systems it introduces noise at every stage. The deeper the hierarchy the more stages. The more stages the more noise.
The organization became less intelligent not because the people got worse but because the architecture degraded the signal.
The Hidden Cost
What gets lost is harder to name than efficiency. It is general knowledge — the capacity to hold the whole picture, to see across domains, to notice when the fragments are not adding up.
General knowledge is not expertise. The cardiologist has expertise. The person who notices that the patient’s financial stress, sleep disruption, and work situation are connected has general knowledge. One is easier to credential and easier to slot into an org chart. The other is what the org chart has no place for.
Organizations systematically undervalue general knowledge because it is hard to measure, hard to assign to a role, and uncomfortable to the specialists whose authority it quietly questions. The person who says we are solving the wrong problem is rarely the most popular person in the room. The structure selects against them.
The result is organizations that are locally expert and globally blind.
Reversing the Division
The reversal is not a return to the generalist organization. You cannot unspecialize. The knowledge base is too large, the domains too complex, the expertise too genuinely valuable.
The reversal is architectural. It is the deliberate reconstruction of the general knowledge layer that scale destroyed — not inside individual humans who cannot hold it all, but inside the system itself. Organizations asking not just what each function knows but what the whole picture says — and building the infrastructure to produce that answer rather than assuming it will emerge from the sum of the parts.
It will not emerge from the sum of the parts. It never did. The org chart hid the cost of its absence well enough that most organizations never calculated what they were paying.
A system that knows what it knows — and knows what it isn’t seeing.