Nate Silver has a lot to say about poker players, risk tolerance, Silicon Valley, and the future of artificial intelligence. His knowledge of statistics and modern information technology is broad. His writing is frequently compelling, and the quotations he’s obtained from prominent interview subjects are entertaining and informative. But his latest book, On the Edge: The Art of Risking Everything (Penguin, 2024) is a fascinating, ill-constructed mess.

The first half of the book, mostly about the world of casinos and professional poker players, is a fun kind of pop sociology. “Professional gambler” seems like a glamorous and fun vocation that many readers likely will have daydreamed about at some point in their lives. The second half of the book dives deep into the possibility that rogue computing applications will create an apocalypse that could destroy all human civilization. Pondering that prospect is potentially fascinating, but more of a downer.

Silver argues that there’s a cohesive whole to be made from those two halves and warns readers against skipping ahead to the apocalypse. Whether that’s true will have to lie with the judgement of each reader (caveat emptor). Despite being a bit all-over-the-place, the book addresses some of the most important economic and philosophical questions being debated today about artificial intelligence and the corporate versus government roles in managing long-term risk.

We might want a market-driven economy, for example, because that’s the path that has historically given us the most growth and the highest standard of living. But what if governments taking a hands-off approach to technology ends up allowing the emergence of an AI application that destroys the world? We wouldn’t allow Microsoft and Google to manufacture their own nuclear warheads, after all.

But the conflict here isn’t the usual one between profit-maximizing corporate types and ostensibly public-minded government regulators. Silver argues that the people most aggressively pursuing AI development aren’t just garden-variety businesspeople. They’re a very specific and particular subsets of capitalists — they’re part of “the River,” his term for the world of highly rational men and women with large appetites for risk and a cool, dispassionate view of potentially negative consequences. The gamblers and crypto developers and hedge fund investors and academic philosophers that are part of the River are more likely to favor the creation of more powerful AI applications for their own intellectual reasons, and are more likely to consider elevated risk levels tolerable.

Silver sets the River against what he calls the Village. Denizens of the Village are the left-liberal politicos and educated professionals most likely to roll their eyes at the mention of Bitcoin — people in the mainstream media, social science departments of elite universities, and the leadership of the Democratic party. Silver says that the Riverians of Silicon Valley tend to see those in the Village as “neurotic scolds.” Think of Hillary Clinton and her 1996 book, It Takes a Village. The way Silver describes the Village also makes it roughly analogous to what people like Curtis Yarvin (aka Mencius Moldbug) call the Cathedral or — to the extent that many Village institutions are dominated by women — the Longhouse.

Of course, not everyone is in either the River or the Village. Unlike some sociological or political typologies that attempt to sort the entire population into a new category set, Silver’s two groups are both elite classifications that only really include a handful of the most influential and high-status people in society. But both groups are unusually influential in their own way. And the most prominent members of each clan have their own rabid fanbases, whether it’s Peter Thiel and team River or, say, Elizabeth Warren and team Village.

These affinities obviously also map on to the current red vs. blue political divide in the US, though Silver is quick to point out that Donald Trump is not part of the River. While Trump is certainly competitive and risk-taking (and despised by the opposing team), according to Silver “he’s shown little of the capacity for abstract, analytical reasoning that distinguishes people in the River.”

Silver’s ability to give uninitiated readers a quick tutorial on AI development while also deeply considering its long-term societal implications is a rare skill, and both newbies and tech aficionados will surely find much interesting material to engage with. That very ability, though, will make some readers wonder why he spends so much time in the first half of the book doing personality profiles on quirky rascals from the world of professional gambling rather than getting to the point of the allegedly life and death questions of AI development. He spends, for example, many pages examining — in extreme detail — a scandal over alleged cheating at a livestreamed poker tournament in 2022. While that controversial surprise win was no doubt intriguing for obsessive poker fans like Silver, it lacks an obvious connection to the world of rationalist philosophy and computer science that he ultimately wants us to be concerned with.

And, as the subtitle suggests, the real weight of the book lies in its musings on computational power and potential AI annihilation rather than the latest casino action. Companies like OpenAI, according to their critics, really are risking “everything” by moving as quickly as possible toward more powerful computing models and applications. The world of gambling, which Silver claims is necessary to understand the stakes of the AI revolution, isn’t about that at all — making money at poker is about making highly constrained moves in a way that prolongs one’s ability to keep playing. No one is betting on — as Silver mentions briefly as a thought experiment — literal games of Russian roulette.

The disconnect between the “poker is fun” half of the book and the “AI is potentially terrifying” half is, unfortunately just one of the book’s organizational issues. Silver starts out, for example, with a well-defined dichotomy in the River vs. the Village, one that will likely be easy to grasp for the many readers who know him most from 538, the popular politics website he founded. But he also splits the people he writes about into additional sub-groups, and sows confusion.

For instance, Silver uses the cognitive dichotomy described by the midcentury philosopher Isiah Berlin of the hedgehog and the fox (i.e., “a fox knows many things, but a hedgehog knows one big thing”). Within his discussion of foxes and hedgehogs he further divides people into “mavericks” and “mediators.” So potentially everyone mentioned in the book, from casino mogul Steve Wynn to AI researcher Eliezer Yudkowsky, is some combination of a 2x2x2 matrix composed of those three opposing sets of characteristics. That makes things a little muddy.

In addition, Silver sometimes attempts to personalize the process of reading the book by describing it with various not entirely consistent metaphors. Early on, we’re told he’s taking us on a journey, with him cheerily advising us to “buckle up” and “bring a few bucks to gamble with” because “there are poker games in the back of the bus.” But then, in the second half of the book, it becomes a performance we’re watching. Silver tells us that “Chapter 6, Illusion, is the first of three chapters that can be thought of as a book within a book, structured as a play in five acts.”

Even the numbering of the chapters themselves, a progress that in most books proceeds without obvious novelty, is here highly visible. The book itself is divided into Part 1 and Part 2. Part 1 consists of Chapter 1 through Chapter 4. After Chapter 4 comes, counterintuitively, Chapter 13, a reference to Silver’s Steven Covey-esque “Thirteen Habits of Highly Effective Risk-Takers.” Part 2 consists of Chapter 5 through Chapter 8 (three out of four of which contain the book within a book that is, simultaneously, also a five-act play). After Chapter 8 comes Chapter ∞ [infinity symbol], which readers could certainly be excused for assuming would be the last. But Silver’s impish delight in non-standard numbering is still not sated, and the final chapter (the second of the two-part conclusion) is Chapter 1776, a reference to the founding principles of the United States. Is that all clear?

The fact that On the Edge wasn’t swiftly divided by an experienced editor into two different projects is no doubt because Silver has built a significant public platform for himself, and presumably now possesses the publishing industry clout — after the success of 2012’s The Signal and the Noise — to resist even the most obviously necessary editorial interventions. He even mentions, at one point, that he has become so famous that it has been “a challenge to stay grounded in my values.” Perhaps it is also become a challenge to accept criticism of one’s writing projects.

The quirky organization of the book (and some tics in the audiobook version, such as Silver doing celebrity voice impersonations when reading quotations of famous interview subjects), were no doubt fun for Silver personally. They quicky grow tiresome and distracting to the reader and listener.

Anyone interested in either professional gambling or AI technology will no doubt be excited to dive into those respective halves of the book. Apart from Nate Silver superfans, the real test will be how big the Venn diagram overlap between those two individually interesting topics will be for the reading public at large.