Leverage on Leverage: 1929’s Real Engine
Sorkin’s anatomy of 1929 shows forced selling, not fear, broke the market—and why AI-era opacity can produce the same cascade.
Andrew Ross Sorkin spent eight years on this book, “1929: Inside the Greatest Crash in Wall Street History--and How It Shattered a Nation”. He is a New York Times financial journalist; his previous book, Too Big to Fail, is already treated as a benchmark for crisis reporting. This time, he leans on a large body of archival material and reconstructs the 1920s boom, the 1929 crash, and the Great Depression through more than seventy key Wall Street figures. The English edition, published by Viking in 2025, quickly entered the New York Times nonfiction bestseller list and was named a notable book of the year.
My main takeaway is simple: this is not a book about how many points the Dow lost in a day. It is a book about how systemic crises actually form over time. Credit expansion, leveraged speculation, hesitant regulators, policy mistakes, wealth illusions, crowd psychology — the book shows how these strands wrap around each other until they crush confidence. The crash is not an isolated accident; it is the release of years of accumulated risk. The more comfortable the boom feels, the more explosives tend to sit under the floorboards.
How confidence actually disappears
For Sorkin, confidence is the bloodstream of the economy. In 1929, the bloodstream started failing from small, almost trivial fractures.
Stock quotes came via ticker machines: prices transmitted by telegraph, printed on paper tape. In October 1929, trading volume surged beyond what the machines could process; the tape fell hours behind real time. Investors were buying and selling off stale prices — like watching the scoreboard for the third inning while the game is already in the eighth. If you do not know the current price but sense that something is wrong, the rational self‑preservation move is to sell first and ask questions later.
Delayed information fed straight into margin calls. Middle‑class investors were routinely buying stocks with 10 or 20 percent down and borrowing the rest. Rising markets made a 10 percent margin feel like free money. Once prices turned, brokers demanded more collateral. Those who could not post cash were liquidated. Forced selling pushed prices lower, triggering more margin calls, and the market fell into a self‑feeding liquidation spiral. Sorkin’s key point is blunt: the crash was not mainly about willing sellers; it was about forced sellers. Confidence “happens slowly, then collapses suddenly” because the institutional plumbing gives you no graceful exit.
If you map that onto 2026, the picture becomes uncomfortable. Information moves in milliseconds, but the line between signal and noise has blurred. When AI agents sit inside critical systems, a large‑scale coordination failure or correlated hallucination would not just delay information; it would call the reliability of the entire information layer into question. Markets that have been trained to expect millisecond certainty have almost no tolerance for ambiguity or opacity.
The “New Era” hallucination
The 1920s were soaked in technological optimism. Cars, electric power, and radio spread quickly. Politicians and commentators started saying the “New Era” might have broken the old business cycle. Sorkin is unsentimental about this: long stretches of prosperity breed a particular illusion, in which optimism stops being a stance and becomes a faith. At that point, people lose the ability to price risk or distinguish quality.
Radio Corporation of America (RCA) became the emblem of this mood. It sat at the frontier of broadcast technology, and its stock became the purest proxy for the new medium. Investors ignored dividends — RCA did not pay any — and bought the story of “limitless potential” instead. What people thought they owned was not current cash flow but the total imagined future of radio as a technology. As rising prices pulled in more capital and reinforced the narrative, the basic discipline of risk assessment eroded.
Reading that in 2026 is unsettling for obvious reasons. AI is scaling fast; large models demonstrate accelerating capabilities; agentic systems are moving from demo to production. A familiar belief is taking hold: AI will rewrite the rules of productivity and unlock every bottleneck. The other side of that story is distribution. Historically, each major technological wave first channels gains toward capital, platforms, and a narrow group of technology owners; workers and institutions take time to adapt. When the speed of technological change outruns the speed of social adjustment, you can get a strange combination: markets celebrating a new growth story while broad segments of society feel exposed and anxious. That tension is itself a source of financial volatility. It is the modern form of the “New Era” hallucination.
Dragging tomorrow’s money into today
For Sorkin, the common thread across major financial crises is debt. Debt is an “extremely optimistic force”: it lets you pull future income and opportunities into the present and spend them as if they were already secure. The problem is that no one knows exactly where the safe line is. Once people realize that the line has been crossed, expectations about the future shrink sharply, and fear becomes rational.
The late‑1920s boom rested on two main structures of leverage. One was margin finance, already mentioned. The other was more intricate and more dangerous: investment trusts. Companies such as Alleghany Corporation issued debt and preferred stock to raise funds, used that money to buy other companies’ shares, and then launched new trusts that held the first trust’s stock, creating vertical chains of ownership. The surface story was diversification. The reality was “leverage on top of leverage,” with highly overlapping underlying assets. When markets turned, the entire stack collapsed almost simultaneously.
Inside these trusts, the incentives were worse. Shares were often allocated at undisclosed discounts to an inner circle. In the Alleghany deal, public investors bought at 35 dollars while partners of J.P. Morgan & Co. and their allies — bankers, politicians, figures like Charles Lindbergh — received shares at around 20 dollars. Insiders could sell into rising prices and dump risk onto later entrants. Goldman Sachs Trading Corporation became the most notorious casualty: its shares traded above 200 dollars at the peak and then sank to low single digits after the crash. The price path destroyed public confidence in the entire structure of investment trusts.
Now look at current debt levels. U.S. gross federal debt crossed 39 trillion dollars in March 2026 and is widely expected to hit 40 trillion before the 2026 elections if current trends continue. Interest costs alone are becoming a visible policy constraint. After a long period of low rates, governments, firms, and households have grown used to using debt as the default growth tool. A world of structurally higher rates forces a repricing of how much debt the system can carry without breaking.
Buffett’s 2026 shareholder meeting captured the mood cleanly. He described the market as a “church with a casino attached” and argued that one‑day options are not investment or even speculation, but pure gambling. Retail enthusiasm, zero‑dated options, and an AI‑heavy narrative have started to reinforce each other. All of this is just another way of dragging future wealth forward and burning it now, with a more polished interface.
Markets as a mirror of human nature
Jesse Livermore, the most famous short seller of 1929, said the stock market is a mirror of underlying economic conditions; causality runs from the economy to the market, not the other way round. Sorkin uses the line to make a broader point: the market reflects not only output and profits but also desire, fear, ego, and herd behavior.
In the 1920s, that human layer was weaponised through “pools.” Pools were tight groups of insiders who secretly accumulated a stock, then traded it among themselves to manufacture volume and a rising price — “painting the tape.” Rising prices and busy tape action pulled in the public. Once enough outside money had entered, the pool operators sold into the demand and walked away, leaving latecomers to face the collapse. The story is not about numbers; it is about calculated exploitation of trust.
In the current AI wave, the market is still that mirror, but the reflections are more complex. Algorithmic feeds and social‑media campaigns can function as a loose, modernised version of the pool, giving certain assets manufactured visibility and narrative momentum. Regulators already see this in meme‑stock episodes and influencer‑driven frenzies. When AI agents begin to operate more independently in high‑frequency trading, routing, or resource allocation, another layer appears. If their decision rules converge, a small shock can produce a coordinated sell‑off. In that world, human fear and greed are not replaced; they are encoded and accelerated. The feedback loop is faster, harder to observe, and harder to interrupt than the tape‑driven spiral of 1929.
After the rules break
Sorkin’s narrative never leaves politics out. 1929 tracks the struggle between financial barons and political figures over who writes the rules of the game. Charles Mitchell of National City Bank used securities affiliates to sell stocks aggressively to the public, blending commercial banking with securities distribution. His camp insisted that Wall Street and Washington occupied separate spheres. Senator Carter Glass saw it differently. He attacked what he called “Mitchellism” — the fusion of deposit banking with speculative promotion — as an invisible tax on American prosperity.
The end of that fight was not subtle. Private markets could not stabilise themselves. The state moved in forcefully. The Glass–Steagall Act separated commercial and investment banking and laid the foundations of the U.S. financial regulatory architecture for most of the next century. When financial elites push their advantage past a certain point, the response tends to be blunt: rules return with a vengeance.
In 2026, similar contests will be running on several levels. AI firms in Silicon Valley signal that agentic systems do not fit easily inside old legal categories; the iteration speed makes traditional oversight look permanently late. Governments, meanwhile, are slowly arming themselves with new regulatory tools and could eventually choose structural remedies that look much closer to forced separation than to gentle nudging.
Zoom out further, and the backdrop is not stable. The United States designed much of the post‑war liberal trade order; it is now one of the main forces testing or eroding some of those rules. Tariffs, industrial policy, sanctions, and explicit geoeconomic rivalry have risen across multiple theatres. The pillars that supported decades of global growth — low tariffs, expanding trade, deep cross‑border supply chains, and thick multilateral forums — are under pressure. For firms and investors, that means more policy risk and shorter planning horizons. For markets, it means that the “rules of the game” feel less predictable.
Financial markets ultimately sit on trust. Investors commit capital over long horizons only if they believe contracts will be enforced, property rights respected, rules changed gradually rather than capriciously, and cross‑border flows not cut off overnight. When that long‑term trust erodes, valuation models change. Risk premia shift. Multiples compress, sometimes quietly. That, more than any single anecdote, is the echo of 1929 into 2026.
Closing
Sorkin frames the book with a line attributed to Einstein: no one really learns from someone else’s experience. He returns to a simple idea: history is a race between education and disaster. The point of 1929 is not to count how many billions evaporated, but to explain why versions of the same disaster keep returning — how each generation finds a new way to believe “this time is different,” and walks into an old psychological trap.
1929 is not a book about a completed past. It is a set of questions aimed at governments, central banks, regulators, financial institutions, and individual investors. Can you stay alert during a boom; can you see risk in the middle of euphoria; can you rebuild constraints when the institutional order itself is loosening? As the AI‑agent wave rolls through the global economy, the live question is how much of today’s prosperity rests on durable productivity shifts and how much on the same old mechanism of pulling tomorrow’s wealth into today’s prices.
You do not read 1929 to be frightened of the next crash. You read it to at least improve your odds of stepping around it, or, failing that, to lower the damage it can do to economies, markets, and societies. Recognising recurring scripts does not guarantee a different ending, but it is hard to change the ending if you refuse to see the script.