Switchcraft

Chapter 1: The Element of Control

There is a moment in the history of every technology when its raw capability stops being the story. The question ceases to be What can it do? and becomes Who says so? For the electric grid, that moment arrived when the first utility executive realized he could shut off a neighborhood’s power not because the turbines had failed, but because a customer had fallen behind on payment. For the telephone system, it came with the invention of the switchboard, a human being who could listen in, reroute, or simply decline to connect. And for artificial intelligence, that moment is now. We have spent the last decade marveling at what these systems can do—write a sonnet, diagnose a rash, argue with a teenager about homework—but we have spent almost no time asking a far more consequential question: Who holds the switch?

The switch is the most underappreciated artifact of the digital age. It is not a chip or a server or a model. It is not even, strictly speaking, a piece of code. A switch is a point of control embedded in a system where a choice can be made: on or off, allow or deny, increase or decrease, proceed or halt. It is the boundary between possibility and permission. We tend to think of technology as a landscape of capabilities, an expanding frontier of what can be done. But capability is only half the story. Every capability is nested inside a network of switches