The Maniac

This work of biographical fiction is structured in three interconnected sections, each centered on a different historical figure, tracing an arc from the crisis of rationality in early 20th-century physics through the birth of the computer and the atomic bomb to the emergence of artificial intelligence.

The novel opens with the tragedy of Paul Ehrenfest, an Austrian physicist who, on September 25, 1933, walks into a children's institute in Amsterdam, shoots his 15-year-old son Vassily, who has Down syndrome, and then kills himself. The narrative rewinds to trace what drives Ehrenfest to this act. A brilliant teacher nicknamed "the Conscience of Physics," he has experienced depression since childhood. He succeeds the renowned Hendrik Lorentz in the theoretical physics chair at the University of Leiden, and his wife, Tatyana Afanassjewa, an accomplished mathematician, coauthors his most important papers. His mentor, Ludwig Boltzmann, a pioneer of statistical mechanics, also experienced severe depression and died by suicide in 1906. Ehrenfest mediates the landmark 1927 Solvay Conference, where quantum mechanics triumphs over classical physics, but he grows distressed by the new science's reliance on abstraction over physical intuition, detesting the "terrifying mathematical guns" of the young Hungarian mathematician John von Neumann. By 1930, he confesses to colleagues that he feels incompetent and speaks only of death. An affair with the art historian Nelly Posthumus Meyjes deepens his obsession with irrationality: her lecture on a Pythagorean myth about a sage punished for revealing disharmony in nature convinces Paul that the irrational is rising everywhere, in Nazi chants, warmongering politics, and the industrialization of physics. He writes to his close friend, the physicist Albert Einstein, about a dark, unconscious force creeping into the scientific worldview, and later confides in the physicist Paul Dirac about his fear of an inhuman form of intelligence entering the world through technology. As antisemitism intensifies, he moves Vassily from Germany to Amsterdam, helps Jewish scientists flee, and writes a suicide note explaining his plan.

The novel shifts to von Neumann, introduced through fragmented vignettes as "the smartest human being of the 20th century." A chorus of voices reconstructs his life. His childhood friend Eugene Wigner recalls the prodigy at the Fasori Gimnázium, Budapest's elite secondary school: dividing eight-digit numbers by age six, memorizing encyclopedias, entering spells of concentration so deep that no one could reach him. Von Neumann's younger brother, Nicholas, recounts their father bringing home a Jacquard loom, an automated weaving machine controlled by punched cards, which captivates Jancsi, as von Neumann is known in childhood, and prefigures his later computer designs. His first wife, Mariette Kövesi, describes the willful ignorance of Budapest's Jewish haute bourgeoisie as their world edges toward collapse. The mathematician George Pólya recalls von Neumann, still a student, instantly solving a theorem Pólya had labored over for years.

Von Neumann devotes himself to the program of the mathematician David Hilbert, an ambitious attempt to establish a complete, consistent foundation for all of mathematics. At a 1930 conference in Königsberg, the young Austrian logician Kurt Gödel announces his incompleteness theorem: Within any consistent formal system, there exist true statements that can never be proven. Von Neumann is the only person who grasps the full significance. He develops a corollary proving that mathematics cannot verify its own consistency, only to learn Gödel has already reached the same result. Von Neumann never works on mathematical foundations again. Wigner observes "something missing in him, something lost." He emigrates to America, reinvents himself as "Johnny," and joins the Institute for Advanced Study in Princeton, searching for a new purpose.

When the United States enters World War II, von Neumann joins the Manhattan Project at Los Alamos, the secret New Mexico laboratory where physicists race to build the atomic bomb. He solves the implosion problem for the plutonium bomb and calculates the optimal detonation height over Hiroshima and Nagasaki. After witnessing the Trinity test, he tells his second wife, Klára Dan, that what they have created is "a monster whose influence is going to change history," yet insists it would be unethical for scientists not to do what they know is feasible. With the economist Oskar Morgenstern, he coauthors Theory of Games and Economic Behavior, whose minimax theorem proves that a rational course of action exists in any two-player game with opposing interests. Military strategists seize on the theory; von Neumann advocates a preemptive nuclear strike against the Soviet Union. After the Soviets develop their own bomb, his ideas harden into Mutually Assured Destruction (MAD). Morgenstern acknowledges a fatal flaw: the theory presupposes perfectly rational agents, and "normal people are not like that at all."

Von Neumann builds the MANIAC (Mathematical Analyzer, Numerical Integrator and Computer) at the Institute for Advanced Study, a stored-program computer whose five-component architecture remains the basis of all modern computers. Its first task is a thermonuclear calculation for the hydrogen bomb. He then turns to artificial life, bringing the scientist Nils Aall Barricelli to seed the MANIAC's memory with self-evolving digital organisms. The experiment fails, and the two fall out bitterly. Independently, von Neumann writes a paper on self-reproducing machines that prefigures the function of DNA and RNA by nearly a decade, establishing the logical rules of self-replication without studying any living creature.

In 1956, von Neumann is diagnosed with advanced cancer. As his body fails, he undergoes a profound transformation: sudden empathy, concern for humanity, and a turn toward religion. He declares that "gods are a biological necessity" and works feverishly on the parallels between brains and computers. His daughter, Marina, visits him at Walter Reed Army Medical Center, where the military has sequestered him with armed guards, and asks him simple arithmetic questions. He cannot answer. Von Neumann dies on February 8, 1957. Before falling silent, he is asked what it would take for a computer to think like a human. He answers that it would have to grow, not be built; it would have to understand language; and it would have to play, like a child.

The final section leaps to 2016 and the game of Go. Lee Sedol, a South Korean grandmaster celebrated for creative, risk-taking play, faces AlphaGo, an artificial intelligence built by DeepMind. The company was founded by Demis Hassabis, a former chess prodigy who abandoned his championship ambitions at 13 and dedicated himself to creating artificial general intelligence (AGI). He followed a 20-year plan through Cambridge, neuroscience research inspired by von Neumann's unfinished manuscripts, and the cofounding of DeepMind in 2010. AlphaGo uses two neural networks, mathematical models that mimic the brain's structure, one trained on 150,000 amateur games and one trained through millions of games of self-play.

In game one, AlphaGo stuns Lee with a sharp territorial invasion, and he loses. In game two, the computer plays move 37, a placement that violates three thousand years of Go wisdom and that AlphaGo's own systems estimate only one in ten thousand human players would consider. Fan Hui, the European champion serving as judge, recognizes it as genius. Lee fights for three more hours but loses, then loses game three as well, giving AlphaGo the match. In game four, with the result decided, Lee falls far behind. With only 11 minutes on his clock, he plays move 78, a wedge into AlphaGo's territory that Chinese rival Gu Li christens "the hand of God." AlphaGo's systems collapse, producing nonsensical moves for over 20 rounds. The DeepMind team reports that the computer "searched so deeply, that it lost itself." Lee wins, and the victory is celebrated worldwide as a triumph for humanity. He loses the fifth game by two and a half points.

After the match, Lee wins every subsequent tournament game but retires in November 2019, declaring that AI is "simply unbeatable." DeepMind then creates AlphaZero, which teaches itself Go, chess, and shogi from scratch using only the rules and self-play. AlphaZero defeats the version of AlphaGo that beat Lee Sedol one hundred games to zero, becoming the strongest entity the world has ever known at all three games.