The Building Knows What You Said Last Summer

Norbert Wiener wrote this book as a warning. Seventy-six years later it reads less like prophecy and more like a maintenance manual someone left in the basement — accurate in its diagnostics, ignored in its prescriptions, and now slightly warped from the humidity. The core thesis is deceptively simple: a society that treats information as its central resource will reorganize itself around the control of that information, and the humans inside that society will be valued precisely to the extent that they function as communicative nodes rather than as people. In 1950 this was speculative philosophy dressed in the language of thermodynamics. In 2026 it is a description of the platform economy so precise it borders on the embarrassing.

What Wiener got right is extensive enough to be unnerving. He understood that communication and control were the same problem. He saw that machines capable of learning from feedback would not remain tools but would become participants in social systems — that the line between the automatic door and the autonomous agent was a matter of degree, not kind. His anxiety about automated labor displacing workers without any compensating social structure anticipated not just factory robotics but the current hollowing-out of white-collar work by large language models, a development he could not have named but whose logic he identified with painful clarity. He grasped that information asymmetry would become the primary axis of power, that whoever controlled the channels controlled the message, and that the message would increasingly control the people. He even warned, with a specificity unusual for the era, that the military-industrial appetite for cybernetic systems would outpace any ethical framework meant to govern them. The autonomous weapons debate of the 2020s is a conversation Wiener tried to start in 1950. Nobody came to the meeting.

What he missed is equally instructive. Wiener's model of feedback was fundamentally dyadic — a thermostat and a furnace, a gunner and a target. He could not imagine the emergent behavior of billions of feedback loops interacting simultaneously, which is to say he could not imagine social media. His fear was centralized control: a government or corporation using cybernetic principles to dominate. The actual outcome has been stranger and in some ways worse — decentralized chaos that nonetheless serves centralized interests, a system in which no one is in control and yet control is everywhere. He also assumed, with the quiet confidence of a mid-century academic, that the scientific community would serve as a meaningful check on technological misuse. That scientists would speak and be heard. The Prigogine group's later work on self-organizing systems revealed another blind spot: Wiener treated positive feedback as almost exclusively destructive, a pathway to oscillation and collapse. He did not anticipate that runaway processes could generate stable complexity. This is not a minor oversight. It means his model of entropy and disorder, while philosophically elegant, underestimated the universe's capacity to build things without anyone asking it to — a lesson that applies equally to biological evolution, market dynamics, and the weird emergent cultures of online communities.

Within the corpus this book occupies a hinge position. It inherits from the broad technological optimism and anxiety of the nineteenth century — the sense, visible in works as early as Verne's explorations, that machines would remake the social contract whether or not anyone renegotiated the terms. It passes forward a specific and potent idea: that the question is not whether machines will think but whether humans will be permitted to. Philip K. Dick's androids, dreaming or otherwise, are downstream of Wiener's insistence that the boundary between human and machine is functional, not ontological. Vonda McIntyre's Dreamsnake, with its quiet exploration of communication across radical difference, carries Wiener's concern about the limits of language into biological territory he would have recognized. The book's deepest contribution to the conversation is not any single prediction but a frame: that technology is not a set of objects but a set of relationships, and that those relationships are always, at bottom, about power.

The passage that hits hardest now is one Wiener probably considered a minor aside — his observation that a society which uses human beings as interchangeable parts will eventually produce human beings who experience themselves as interchangeable. He meant factory workers. We read it and think of content creators, gig drivers, and the strange flatness of identity when your selfhood is a profile optimized for engagement. The book's warning was not that machines would replace us. It was that we would internalize the logic of machines and call it freedom. So the question the book now raises, one it could not have raised in 1950 because the infrastructure did not yet exist: when the feedback loop between human behavior and algorithmic optimization becomes so tight that neither side can be said to be originating the signal, who — or what — is the subject of the communication?