He was almost consumed by his passion for logic, and during his entire life, that strange gift of his let him see things with remarkable clarity, granting him a vision so blinding that to others, whose focus is smeared by emotional considerations and prejudices, his point of view seemed completely incomprehensible.
Without a doubt, this book was one of my favorite reads of 2023/2024 (I started it late December and finished in early January). It swept me away with a force nearly unlike any other book. Somehow, it manages to tackle subjects like Johann von Neumann (the star of the show), AlphaGo, and even the strange story of Paul Ehrenfest (with whom I was previously unfamiliar) with great aplomb, bearing the force of an addictive contemporary thriller. Although the novel’s focus is von Neumann, it is of course greater than any one person—the structure itself is formatted through recollections of his acquaintances; these are fictional, of course, but Labatut is so skilled that I honestly had to remind myself that they weren’t real. In a word, the entire novel is concisely packed together with narrative, plot, ‘characters’, and dramatic tension—it comes together beautifully, and I struggle to even find where the seams are. The ideas behind the novel are a huge part of what makes it a success. Computational thinking is the key here, more or less; from von Neumann’s ‘MANIAC’ to AlphaGo in 2015, the subtly hidden but omnipresent character is none other than the almighty computer in all of its various forms. I really appreciated how Labatut manages to boil down complex discussions of advanced-level mathematics and computer science into things even I, a poor humanities pleb, can understand. It seemed like magic, honestly. I have always been interested in the more philosophical strands of math, and this book does appeal to that side, which helps; but even so, it’s sheer madness how he was able to explain things almost as easily as von Neumann himself. For example, Gödel’s incompleteness theorems were something I was vaguely aware of, but it wasn’t until seeing them explained here that I actually felt like I understood. The nature of von Neumann’s single-minded focus towards his obsession—first in mathematics and then theoretical physics, and then onwards to the burgeoning field of information systems and computer science—is extremely captivating. Labatut must be tapping into some kind of writing genius for his ability to channel von Neumann’s colleagues, friends, and rivals—it was hard to remember that I was indeed reading a fictional book at times. As a figure, von Neumann is entirely alien—his mind works in such a drastically different way from most of us, and yet, it is that very quality that makes it extraordinary to even take a peek into his mind; and somehow, Labatut manages to achieve this.The section on the development of the atomic bomb was also quite interesting, since I’ve got the Oppenheimer film under my belt; going through that again in Labatut’s delightful prose, and especially in his precisely captured voices of figures like Feynman and the rest, was pure delight. Of course, the perspective here is more on the theoretical and computational aspects, but it was no less enjoyable; and of course, the figures ruminating on the nature of the bomb and the ethical consequences of all that theory… well, as a theme, it fits seamlessly into the book.Still, the throughline of the novel is about the power of computing more generally, as well as the limits of computing—and whether it can approximate something like natural minds or environments. We see hints of this even in early computing, and Labatut focuses on the fine line between some of von Neumann’s domineering ‘collaboration’ with his colleagues, at times refining on ideas and taking over projects that he believes he can do better. (He did and probably could, but still, it’s that characteristic ego of geniuses.) For example, his collaboration on the work of Nils Aall Barricelli, whose influence was not duly recognized. By the time we get to AlphaGo, the various pieces of the book click together in place, and you begin to see the overall web Labatut has been weaving all this time. The discussion of AlphaGo may seem at first a non-sequitur—a topic as creative and abstract as go, however, is a natural companion to the abstract theories of computation. Whereas von Neumann’s machines are artificial and structured, go is chaotic and messy, something that was only tamed by AlphaGo in 2015. Having tried my hand at go the last few years (though I’m still very much a beginner), I loved reading about Lee Sedol and the other go champions’ approaches to the game as well as to their interactions with the computer program. The final part of the book begins its discussions not with AlphaGo but with one of the founders of DeepMind, Demis Hassabis, a wunderkind whose prodigious talents make him a natural figure in the world of ‘artificial intelligence’. Sometimes you look at what other people have accomplished and begin to doubt what you’ve even accomplished in your life; Hassabis’s skillset is certainly beyond the standard deviation, though. His goal of working on deep learning and neural networks has undoubtedly had a positive effect on humanity—for example, AlphaFold predicts protein structures, which tackles one of the most pressing problems in modern biology. Examining the humans behind the programs is also a kind of unveiling, at the least, of the motivations behind such projects, if not for their lofty ambitions.One of the most mind-bending sections comes towards the end, where AlphaGo’s successful reign is eclipsed only by its ‘successor’, AlphaGo Zero (and there are many successive iterations), which manages to learn the same capabilities as the original but without human involvement. Very much a mind-blown sort of moment. We are certainly in the early stages of machine learning, but this kind of example shows that so much more is likely possible. Perhaps not enough to replace human thinking—the human brain is incredibly complex in its entirety, and I doubt machines can replicate that anytime soon—but with the capacity to come up with creative maneuvers in go, and play extremely rare moves that almost seem to taunt their opponent, the programs are coming closer than perhaps we will even be able to understand. I’m not sure where I first read it, but I believe the nature of some machine learning isn’t even clear to researchers themselves; we just see its effects.Anyway, I’ll stop rambling on and on but this is an incredibly brilliant book that made me reflect a lot on the nature of computation, theoretical models of reality, and the nature of being human in an age of machines… to name but a small slice of the many complex themes invoked in the book. This is a fictional ‘novel’, yet its immediate concerns are not far from reality—nor are its characters unreal. In a liminal sort of way, this book offers a lot to a sufficiently curious reader at the thresholds of what is possible and what may be in the future. I am looking forward to reading Labatut’s previous book, and whatever he publishes next. I may be overzealous in this, but I would recommend this book to anyone with a pulse.
