Benjamin Labatut's "The Maniac"

A singular book

(Note: all quotes in this essay are from the book, and thus are fictionalized versions of real people) 

“His answer to the nuclear dilemma was a perfect reflection of the best and worst in him: mercilessly logical, completely counterintuitive, and so utterly rational that it borders on the psychopathic.”

-Klara Dan (his second wife)

Every now and then, if you read a lot, you have the exquisite pleasure of reading a book that feels like it was written just for you. A few fond memories of mine, among many:

• At 16, discovering Brian Lumley’s Necroscope series, an ultraviolent mashup of vampirism, necromancy, psychic powers, espionage, Cold War paranoia, higher mathematics, interdimensional travel, Lovecraftian horror, and nuclear warfare.

• At 19, reading Richard Rhodes’ engrossing The Making of the Atomic Bomb, which is quite simply the best history I have ever read about anything. 

• At 25, reading Gene Wolfe’s staggeringly weird and utterly compelling ‘science fantasy’ (Wolfe’s term) series, The Book of the New Sun. I can't recall a fictional book I have read that has ever excited me more, sparked more ideas, or made me reflect more on my life.

• At 29, reading Thomas Traherne’s sublime Centuries of Meditation, the primary work of a man David Bentley Hart called “the sanest man who ever lived.”

And now, on the cusp of 40, I have found Benjamin Labatut's The Maniac, a fictionalized biography of the mathematician John Von Neumann, told in the form of recollections from his family, friends, and enemies (there is strong overlap between these categories). This is an inspired conceit for a complex and multifaceted person like Von Neumann, and I think more successful than a straightforward biography would have been. Sometimes, as in the best weird fiction, it is best to show the impact of a thing by focusing not on the thing itself but on the wake of its passage.

The book is exactly the spiritual sequel I had hoped for from Labatut to his When We Cease to Understand the World, a series of similarly lightly fictionalized biographies luminaries of 20th century physics and mathematics. Labatut writes these men’s’ lives like the biographies of medieval saints, where madness and illumination are closely allied. I nearly wrote an essay on that one, but then this came out and I came to think of the former as the prelude to The Maniac. 

If you aren't familiar with John Von Neumann, name a figure in the 20th century who met him other than Albert Einstein, and those people are on record saying - whether they loved him or hated him - that he was the very smartest person they ever met. He gifted the world a mathematical rendering of quantum mechanics that helped lay the foundations of the field. He invented game theory, systematizing it with Oskar Morgenstern. Along with Niels Bohr, he helped many of Europe's most brilliant scientists escape the grasp of the Nazis. He contributed to the Manhattan Project as a general purpose problem solver. He invented the architecture of all modern computers, the stored-program machine where the programs and the systems that execute the programs are separated, making it possible to program a computer without having to rebuild it. He used the first computer he designed, the MANIAC, to help affirmatively answer the question whether it was feasible to build a hydrogen bomb. He made the first attempts at computerized weather forecasting, and discovered (though he did not extensively develop) chaos theory. He invented the first algorithm for pseudo random number generators. He contributed to the theory of deterrence via mutually assured destruction, gifting his game theory to the bright boys at the RAND Corporation. He laid the mathematical foundations for the study of self-replicating creatures - biological, mechanical, or digital - before the structure of DNA was known and before RNA was discovered. He applied that idea to working out how humanity could colonize the galaxy at slower than light speeds via self-replicating machines, the eponymous Von Neumann probes. And he gave the first thorough treatment - while he was dying of an aggressive cancer, no less - to questions of how to build artificial intelligence systems through self-play and reinforcement learning. 

And if you want to know, at least according to Labatut, whom Von Neumann felt small next to, it was Kurt Gödel, whose Incompleteness Theorem dashed a young Von Neumann fervent hopes of completing Hilbert’s program and placing mathematics on an unassailable logical footing. He wasn't a man who had cause to feel small next to anyone else.

The Deep Structure of the Narrative

The structure of the narrative, going back and forth between a panoply of voices, provides the opportunity for exploring both how people reacted to him, but also to decide, as a reader, which viewer or group of viewers you think is closer to the truth. Personally, in the context of the book, I side with Eugene Wigner, a physicist and a close friend of Von Neumann. In one of the books earliest chapters, he remarks on what he figured out about his friend when he was still a boy:

Only he was truly awake

It takes the full book for this observation to play out. Various friends describe him as an overgrown child playing with toys, or as a moral imbecile, someone more interested in technology than its effects on people. And perhaps that is fair, unless you choose to view him as fully awake and aware of what he was doing. Here's the sequence I spotted in the book:

• He contributes to the Manhattan Project, and witnesses the Trinity test. He is deeply shaken by the thing Los Alamos had achieved (as was everyone). But he feels no regret,¹ and pours himself wholeheartedly into the advance of technology at all costs, deciding that it would be immoral not to do so. 

• He argues for a nuclear first-strike, or at least nuclear blackmail, of the Soviet Union before it can develop an atomic bomb, to secure America and the free world’s hegemony far into the future. He wants to win the “game” before its first turn. He doesn't find any takers in a war-exhausted United States trying to demobilize millions of men and restore its economy to a peacetime state. 

• So he accelerates technological development whole orders of magnitude by inventing the stored program computer, making Turing’s Machine a practical reality. He finds a backer for this project in the Department of Defense, which wants to use it to answer whether hydrogen bombs are feasible. 

• But this is fine by Von Neumann. Not only does he have no moral objections to thermonuclear weapons, but as part of the agreement he can use the downtime of the computer for anything he wants without restrictions. 

• And hydrogen bombs made deterrence truly effective. It's hard to see this now, as we blend together in our minds the city-destroying fission bombs used on Hiroshima and Nagasaki with the state-destroying hydrogen bombs that followed, but wars with fission bombs were, from the standpoint of the late 1940s and early 1950s, feasible. Even dozens of cities burned out by fission bombs could be reconstructed. Once nuclear weapons mounted on ICBMs reached the megaton range the fire and fallout from their detonations would kill tens if not hundreds of millions, many of them hundreds of kilometers downrange of the detonations. Hydrogen bombs made nuclear warfare not feasible by making mutually assured destruction truly mutual.

And you can call this monstrous, but the logic works. It's trusting an inhuman, monstrous logic to protect the world, but … it … works. That doesn't mean it can work forever, but deterrence via threatening to kill everyone on both sides of a conflict (and millions upon millions more in neutral countries) worked better than consulting feelings, or appealing to “the better angels of our nature.” That could come later. Deterrence buys time to think of other things, for tempers to cool, for arms control talks to happen. And for humanity to invent the next thing, possibly something to neutralize make nuclear weapons look like a Fourth of July roman candle in comparison.

If you still feel Von Neumann was an immoral monster of intellect, then you can rest easy because he suffered hard for it. If Von Neumann were just a fictional character in an Asimov or Heinlein story, I'd have called his end too on the nose for believable drama. But the historical Von Neumann died in great pain, physical and emotional, from an aggressive cancer that, in the end, ate into his brain and robbed him of his most cherished possession: his reason. So there is a kind of brutal justice in it, or a Prometheus punished by Zeus for bringing fire to mortals angle, if you want there to be. Or it is a terrible tragedy. The reader can decide. 

Labatut movingly portrays Von Neumann final days through the eyes of his wife, daughter, friends, and coworkers. Though, because the man was noted for his black humor, the author finds space for it, though the joke is not Von Neumann’s:

“I asked him, point-blank, how he used to contemplate, with total equanimity, the killing of hundreds of millions of people in a nuclear first strike against the Soviet Union, and yet he could not face his own mortality with any sense of calm or dignity. ‘That is entirely different,’ he replied.”

-Marina Von Neumann (his daughter) 

Reinforcement Learning

At the midpoint of the novel, Richard Feynman gets to say a few words. He recalls that the physicists and engineers at Los Alamos spent an extraordinary amount of time playing games. This was because their lives were filled with hurry-up-and-wait activity: waiting for a new plutonium sample to arrive, waiting for the engineers to fabricate a test stand, waiting for the human computers to finish an involved hydrodynamic calculation. And so they played games: poker, chess, and Go. Go in particular fascinated Von Neumann, because it started with a blank board. Even forgetting the larger size of the board, the fixed starting point of chess meant fewer possible board configurations. It seemed like something computers could never solve. 

I'm not spoiling anything if I tell you that the epilogue of the book is a detailed narration of Lee Sedol’s battle of wits with DeepMind’s AlphaGo. After AlphaGo bests Sedol, the researchers at DeepMind try an experiment: AlphaGo was trained off human gameplay, with explicit Go rules hard coded into it, and then learned through self-play. AlphaZero, the name DeepMind gave to the next model, started with only the ability to put down pieces and perceive the board. It discovered win and loss conditions playing random games against itself. And it became better than any human Go master even faster than AlphaGo did. Von Neumann intuition about how to build intelligent problem solving systems was correct.

We are the image of the thing, not the thing

One of the thing I take from Labatut's books is the folly of expecting the world to conform to our presuppositions.

A bright person notices an irregularity, a pattern, in some natural phenomena. They investigate it, turn it over in their head. They read some new works by prominent men of science. Then they may be driven by the logic of their discovery, their own inner demons, the pressures of society, whatever, but they propose something scandalous, wild, seemingly irrational - but a theory that fits the facts better than anything else. That not only fits the facts, but predicts new, unseen ones that then are confirmed.

And calm, staid expectations of the world are upended. 

This folly is particularly acute, in our real world and in Labatut's fiction, in the Solvay Conference which brought together the luminaries of physics, and where the major debates about the status of quantum mechanics were held. The earlier generation of physicists, apostles of Einstein, were appalled by this intensely mathematical, heretically statistical science, which declared that the world of the near-infinitely small did not conform to anything humans could visualize or imagine. The quantum theory won the day, buoyed by the success in predicting experimental results and the natural way that ideas change, with the passing away of the Old Guard. 

In mathematics, the issue was even deeper. Set theory, seemingly the perfect instrument for the systematizing of all of mathematics, had revealed troublesome and irresolvable paradoxes (irresolvable without seemingly arbitrary axioms like the Axiom of Choice, that is). David Hilbert sought to set the world to right by launching a program to lay mathematics once and for all on a stable, logical foundation. Profound mathematicians like Bertrand Russell and Alfred North Whitehead devoted Herculean labors to doing so. And then an obscure Austrian grad student up ended all of that with an unassailable theorem. 

There's something sad and pathetic in these people, many of them atheists who thought they had grown beyond God, or deists who thought God was a subject of purely rational contemplation, to expect the world to have a human face, the way White Evangelical Christians depict a Caucasian Jesus in their art. As if to look upon the world was the look into a mirror of the human mind, rather than our minds being but a reflection, a model, of the Original, something we learn and grow in relation to. It's just this side of solipsism, to expect the world to be just the way your brain decides at time T it must be, rather than go through the learning, the changing, the transformation of adapting yourself to the world, at times T+n

Transformation

“The thing about my husband that people don’t understand is that he truly saw life as a game, he regarded all human endeavors, no matter how deadly or serious, in that spirit. He once told me that, just as wild animals play when they are young in preparation for lethal circumstances arising later in their lives, mathematics may be, to a large extent, nothing but a strange and wonderful collection of games, an enterprise whose real purpose, beyond any one stated outright, is to slowly work changes in the individual and collective human psyche, as a way to prepare us for a future that nobody can imagine. “

-Klara Dan

To bring things full circle, and perhaps touch a bit too much on contemporary issues, I must confess to having, like Von Neumann, high hopes for artificial intelligence. Some thinkers hope for fantastic medical breakthroughs, others for solutions to climate change, others for weapons to permanently deter and contain our enemies. Others fear killer robots, or entrenching inequality, or the world (somehow) losing meaning if there are creations smarter than their creators at loose in the world. Often the same hopes and fears coincide in the same people.

I hope for something both grander and more elemental: profound transformation. When I contemplate the contemporary scene, where autocrats hunger after insignificant territories to sate their ambition and what they call their national “destiny,” risking the whole world in the process, where people praise evil in the streets and call for the completion of historic crimes, where whether a semiconductor manufacturer has hired enough Native American women contractors is a critical factor in whether they can break ground on a project in a nation’s vital interest, when two senile old men contest to be reelected king, I want a change. I don’t expect human nature to vanish, but I hope for new and strange forces in the world, things to drive our transformation from what we are into what we might become. Things that force us to grapple with fundamental questions rather than carrying on in a state of drift and increasing disorder.

I want a change to bring in a new age of terrors and wonders. To sweep the old vain, stupid, corrupt world away.² A world where we look back at what we longed for and feared, and laugh at how crude and unimaginative we were. Because, to borrow Nick Bostrom’s rocket in flight metaphor, we are at a midpoint in our flight between a stable, pre-industrial past, and a new world of open frontiers and possibilities, in orbit where we are, to quote the space systems engineers, “halfway to anywhere”.

That’s what I hope and want, not what I expect. I’ll have more to say about what I expect in a near future essay, about external barriers to achieving Artificial General Intelligence.

That may be a strong, wild ending for an essay, but a strong and wild book such as this deserves it.

1. Von Neumann even had a sick burn on Oppenheimer. Being told about Oppenheimer’s guilt and public regret about his work on the Manhattan Project, he remarked “some people claim the guilt to own the sin.”

2. And replace it with a world as stupid, vain, and corrupt, but in different and exciting ways, with different ways to be and exist, and more ways to be and exist beyond the stupid, vain, and corrupt.