Driving the subway train from one busy station to the next, he made it through the day with difficulty. The heavy air, the noise, the constant motion - all of it conspired to make him feel even more fragile and anxious than normal. The cap stood silent vigil, detecting fluctuations in his brain waves and sending the signal wirelessly to a central processing server. Once there, it was preprocessed in several stages, converting it to a standardized form of input for the neural network. This network, consisting of many interconnected layers of varying type and connection scheme, accepted the input data and, step by step, modified the values according to both predefined functions and weights learned during the training process. After many stages, the resulting output was a group of values representing probabilities of particular psychological states - sleepy, frustrated, excited, and the like. The probability that his brain activity represented frustration and anxiety were particularly high several times that day - both over .9, or 90%, for several minutes or longer at a time. These measurements of his psychological state were recorded, and a brief report sent to his manager. At the end of his shift, the manager asked him if everything was alright, and offered to change his hours around if it would help. Chu wondered, for a fleeting moment, how the man had known, but he seemed very understanding about it, and appeared reassured that he had just been worried about his parents, so he didn’t think any further of the brief conversation.
Up to this point, the analysis had all taken place on the subway operator’s employee monitoring system. But because it exceeded threshold values for certain metrics, and because of his sensitive role as a train operator, his psychometric profile triggered a hand-off to the wider, government-operated surveillance network. This system centralized many different types of data - cell phone location, metadata, and call content, email, employment status, criminal history, social credit status, and many others. Chu’s data history was examined by various algorithms for signs of disloyalty or troublemaking, and his status was elevated to a higher level of data collection, so that a fuller social, psychological, and behavioral profile could be constructed.