The interesting idea I read here is that systems that were originally intended to measure the real world, like a traffic map or social network, become so influential they have a major effect on the thing it measures. This could result in an interesting new steady state, like side roads filled to capacity with people following Waze directions, or fractued networks of filter bubbles. Or instead of steady state the system might be unstable, like massive volatility cause by algorithmic trading. The “analog” metaphor here is that the user isn’t the consumer of a carefully design system, but they’re one “electron” in the circuit that carries many users, and the overall behavior of the system depends on what they and millions of other users do as they interact with each other. In that way he suggests the outcome is not planned and is probably unpredictable.