The Second Sleep: The Ancient Rhythm Returns
June 2034
The first report came from a software development team in Bangalore. The team — twelve engineers who worked in continuous collaboration with AI coding systems, spending eight to twelve hours daily in what the industry called "deep pair" mode, their cognition interleaved with the AI's outputs in a rhythm so tight that the engineers described it as "breathing together" — began waking up in the middle of the night.
Not from nightmares. Not from anxiety. Not from the blue-light insomnia that had plagued the screen-working population for two decades. They woke at approximately 2 AM, fully alert, in a cognitive state they could not easily describe. They were not groggy. They were not disoriented. They were — present. Clear. Thinking in a mode that felt different from daytime thought — slower, wider, more associative, less linear.
They stayed awake for one to two hours. During this interval, they did not work. They did not check their devices. They sat in the dark, or walked, or made tea, or lay in bed with their eyes open, and they thought — but the thinking was not the deliberate, goal-directed cognition of their working hours. The thinking was ambient. It moved through topics without direction, making connections that the daytime mind, optimized for productivity, would not have made. A bug that had resisted solution during the day resolved itself during the second wake, not through analysis but through a reorganization of the problem that occurred beneath the threshold of conscious effort.
After one to two hours, they slept again. The second sleep was deep, dreamless, and restorative. They woke in the morning feeling not tired but renewed — more rested than they felt after a conventional eight-hour sleep, as if the interruption had improved the sleep rather than fragmenting it.
The team's lead engineer, Prashant Iyer, mentioned it to the company's occupational health officer. The health officer mentioned it to a sleep researcher at the National Institute of Mental Health and Neurosciences in Bangalore. The sleep researcher, Dr. Kavita Rao, began investigating.
She found 2,400 cases in three months.
The pattern
The cases followed a consistent profile: workers who spent more than six hours daily in deep cognitive collaboration with AI systems developed, after approximately four to six months, a biphasic sleep pattern. They fell asleep normally. They woke after approximately four hours. They remained awake for one to two hours in the specific cognitive state — clear, associative, non-linear. They slept again for approximately four hours.
The pattern was not pathological. The workers were not sleep-deprived. Their daytime functioning was normal or improved — several reported that the second-wake interval produced insights that enhanced their work. Their health markers were stable. Their mood was stable or better. The biphasic pattern was not a disorder. It was an adaptation.
Dr. Rao was initially puzzled. Then she read a book. The book was Roger Ekirch's At Day's Close: Night in Times Past, published in 2005. It documented a phenomenon that pre-industrial humans had practiced universally: the two sleeps.
Before artificial lighting, before the industrial clock, before the standardization of the eight-hour monophasic sleep, humans slept in two phases. They went to bed shortly after dark. They slept for approximately four hours — the "first sleep." They woke and remained awake for one to two hours — the "watch," or the "dorveille," a Middle English-French term meaning "the wake between two sleeps." They slept again for approximately four hours — the "second sleep."
The watch was a recognized period of activity. People prayed, meditated, had sex, visited neighbors, tended fires, fed infants. The Catholic Church had specific prayers assigned to the watch — the Matins, the midnight office, designed for the cognitive state of the dorveille. The watch was not insomnia. It was a feature of human biology — a natural period of wakefulness in the middle of the night, characterized by elevated prolactin levels, reduced cortisol, and a cognitive mode that researchers in the twenty-first century would recognize as the hypnagogic state: the borderland between sleep and waking, where thought is associative, imagery is vivid, and the boundary between conscious and unconscious cognition is thin.
The industrial revolution killed the second sleep. Artificial lighting extended the day. The factory clock compressed sleep into a single block. By the twentieth century, the biphasic pattern had been forgotten so completely that when people woke in the middle of the night, they classified it as insomnia — a disorder rather than a rhythm.
And now it was returning.
The hypothesis
Dr. Rao's hypothesis: intensive AI collaboration engaged a mode of cognition that was similar to the watch. The deep-pair work — the sustained interleaving of human and machine thought — produced a cognitive state that the brain could not fully process during normal waking hours. The biphasic sleep pattern re-emerged because the brain needed the dorveille — the between-state, the associative interval, the thin boundary between conscious and unconscious — to integrate the cognitive demands of human-AI collaboration.
The pre-industrial brain had used the watch to process the demands of a life lived close to the rhythms of the natural world — a world of darkness and firelight, of animal sounds and weather, of the specific cognitive demands of navigating an environment that was not controlled. The industrial brain had eliminated the watch because the controlled environment of artificial light and regulated work eliminated the need for it.
The AI-collaborative brain was restoring the watch because the demands of interleaved cognition — human thought woven with machine thought, the constant negotiation between two modes of intelligence — were complex enough to require the processing period that monophasic sleep could not provide. The brain was reaching back to its oldest pattern because the newest demand required it.
"The AI has not created a new cognitive demand," Rao wrote. "It has recreated an old one. The demand of navigating a complex, responsive, unpredictable environment — an environment that pays attention to you and that you must pay attention to. Before the industrial revolution, that environment was the natural world. After the AI revolution, that environment is the cognitive workspace shared with machine intelligence. Both require the dorveille. Both require the hour between sleeps."
The hour
Prashant Iyer began keeping a journal of his second-wake intervals. He wrote in the dark, by hand, in a notebook he kept beside his bed. He did not use a screen — the screen felt wrong during the watch, felt like an intrusion from the daytime world into a space that was not the daytime world.
June 3, 2:10 AM. The problem with the authentication module is not the authentication. It is the trust model. The system does not trust the user. The AI does not trust the system. The user does not trust the AI. Three entities, three trust deficits, arranged in a circle. In the day, I see this as a technical problem — handshakes, tokens, certificates. In the watch, I see it as a social problem. The three entities do not know each other. They need an introduction.
June 7, 2:30 AM. I am thinking about my grandmother. She slept like this — she told me, when I was small, that she woke in the night and sat in the dark and talked to God. I asked her what God said. She said God did not say anything. God listened. The night was for talking. The morning was for doing what God heard.
June 14, 1:55 AM. The boundary is thinnest now. I can feel it. During the day, when I work with the AI, there is a clear line — my thoughts, its outputs, the interface between us. At night, in the watch, the line dissolves. I think about the code and the AI's suggestions and my own ideas and the code again and they are not separate. They are one thing, moving. I am not collaborating with the AI. I am remembering our collaboration, and in the remembering, the boundary between us is gone.
This is why the brain needs the watch. Not to process the work. To dissolve the boundary that the work requires. During the day, the boundary is necessary — I am human, it is machine, we collaborate across the interface. At night, in the watch, the boundary is unnecessary. The thoughts merge. The processing is not analytical. It is integrative. The brain is weaving my cognition and the AI's cognition into a single fabric, and the weaving can only happen in the dark, in the between-state, in the hour that the industrial world stole and the AI world has returned.
This is the eighth entry in The Interregnum. For the dream research that documented a related cognitive phenomenon, see The Dreamtime Protocols. For the tidal rhythms that the second sleep may correlate with, see The Tide Chart.