Markov Chain Norris May 2026

He put on his coat. The hospital smelled of antiseptic and overcooked carrots. Ward 14 was a long, fluorescent-lit room with six beds. Chloe was in the third. She was thirty-four, but she looked sixty. Her hair was gone. Her skin was the color of damp paper. Tubes ran from her arms like tributaries of a sad river.

The past came flooding back, not as a sequence of independent steps, but as a single, unbearable weight. And he realized his great mistake: a Markov chain is a beautiful abstraction, but a human being is not. A human being carries every previous state, not in the mathematics, but in the marrow. markov chain norris

This was the Norris method: a man as a memoryless process. It was clean. It was mathematical. It was, he believed, the only rational way to live. He put on his coat

He stayed for three weeks. He slept in the plastic chair. He read her old stories—not stochastic processes, but fairy tales, the ones she’d loved as a child. He held her hand when the pain was bad. He told her about the lighthouse postcard, that he’d kept it all these years, that he’d lied about filing it away. Chloe was in the third

For the first time in his life, Professor Norris allowed his chain to have a memory. And he found, to his quiet astonishment, that it did not break. It merely became human.

He was a man who believed in the elegance of forgetting. Not memory loss, but conditional independence : the future should depend only on the present, not the past. It was the central tenet of his life’s work—the Markov chain. And for twenty-three years, he had applied it to everything: the movement of gas molecules, the rise and fall of stock prices, the shuffling of a deck of cards. Even to himself.

He spent the next hour grading papers on transition matrices. Then he made tea. Then he stood at the window again. The rain hadn’t stopped. The future state— go or stay —depended only on the present. And the present contained a daughter in a hospital bed, and a father who had spent a decade pretending that the past had no probability mass.