Predictable Repetition: Seven Public Bets on the AI Era
Seven public predictions about infrastructure, platforms, organizations, trust, and value migration in the AI era, grounded in historical patterns.
In 1889, in New York, the shareholders of the “Electric Sugar Refining Company” finally pushed open the door of that workshop.
For years, the company had claimed to hold the secret process for refining cane sugar with electricity, raising nearly a million dollars in financing; the only condition was that no one was ever allowed into that workshop — trade secret. According to reports at the time, once the door opened, there was no process inside, no sugar, only a pile of fake machines built for show. It lies among the wreckage listed in The People Swept Away by Electricity.
This chapter does something different from every other chapter in this book: instead of dissecting what has already happened, it pushes the framework forward — writing down, with a signature and a date, the plays that this round of revolution is about to re-stage, publicly and on the record.
Ground rules first. Predicting parameters is divination: which company wins, which year it collapses, what the numbers land on — this book touches none of it. What can be predicted was never the event, but the way a structure responds to pressure — five revolutions have given five samples, and the structure’s response has never varied. These bets will not be deleted or edited if they turn out wrong: the error itself becomes a boundary marker for this method. A framework that dares not put itself on the record is not worth betting your fortune on.
The seven bets follow.
Bet One: A Textbook-Grade “Mystery Workshop”
Structural basis: the market size of a fraud equals the public’s awe minus the public’s understanding (The People Swept Away by Electricity). In this round, the gap between the two sits at a historic high.
The bet: this round will expose at least one landmark fraud case, starring a marquee company that has raised financing in the billions; its “mystery workshop” will no longer be a factory floor no one may enter, but a demo script or a benchmark leaderboard — what is demonstrated is human labor, what is delivered is billed as intelligence; or, the model is bred purely to top the benchmark, with no bearing on delivery. The scattered small cases of the past few years are only the trailer; the textbook-grade main case has yet to appear.
Falsification line: if by 2031 this round has still produced no fraud case at billion-dollar financing scale, this bet is void — that would mean humanity, for the first time, failed to breed a great fraud at a historic high in the awe-understanding gap. Delightful, commendable, and hard to believe.
Bet Two: The Collapse Burns on a Financial Structure Unique to This Round, and No Lights Go Out
Structural basis: the chain in Insull’s Pyramid — the capital hunger of an infrastructure boom inevitably summons financial structures that sell “certainty”; history never repeats the way the fire burns, only the fact that it catches. Each round burns its own new recipe: the holding-company pyramids of 1929, vendor financing in 2000.
The bet: if this round produces a landmark collapse, the point of ignition will not be “AI turned out useless,” but a structure unique to this round — circular deals, compute-collateralized loans, securitized leases, and their kin; on the day of the collapse, not a single light will go out: the models will keep answering, the data centers will keep humming, and what dies is the holding structure. The pattern of real assets, fake structure will be reprinted verbatim. The final-act signal — “the shovel-sellers start lending” — is already on the scene.
Afterward, the legislators will arrive, precisely burying this round’s playbook — the first clause of the next PUHCA can already be guessed with eighty-percent confidence: it will concern the recognition of circular revenue and disclosure of related-party compute transactions. And, as always, it will not be able to control the round after next.
Falsification line: if the collapse arrives in the form of “demand vanishing” (AI proven useless), I am wrong; or if this round is absorbed smoothly with no collapse at all, I am also wrong — and glad to be, since that would be the first time in five hundred years.
Bet Three: No Chips on the Second Owner’s List
Structural basis: Railway Mania — the economic return on infrastructure often belongs to the second owner who buys it at thirty cents on the dollar; but this round’s “foundation” is mixed with a large share of consumables.
The bet: if a collapse occurs, a second-owner moment will follow within two to four years: what changes hands at a discount will be the buildings, the power interconnects, the substation capacity, and the land — not the chips; the chips will be cleared out at scrap-consumable prices, like horse feed in 1849. The next wave of applications will grow on discounted compute and data centers the way YouTube grew on dark fiber. For the reader, only one operating tip: on that day, the most valuable qualification is a clean balance sheet — make sure you’re still eligible to bid.
Bet Four: A “Named” Accident, and the Birth of a New Industry
Structural basis: The People Swept Away by Electricity—the full cycle from Feeks’s death to UL: accident—public outrage—regulation—certification industry; safety has always been a public good that must be bought. The Timetable War has already shown the trailer: a single automated update chain simultaneously crashed roughly 8.5 million devices worldwide, and an agent deleted a production database during a code freeze.
The bet: this round will produce a “named” accident — a specific victim, a specific date, an automated chain with no malicious intent — and it will become this round’s death of Feeks, compressing every prior, incremental safety debate into a single legislative jump. The accident will most likely take the form of an agent chain triggering a chain reaction in the physical world. This will give birth to this era’s UL: auditing, certification, and insurance for AI systems will turn from a compliance cost into an independent industry; the Petrov test will be written into procurement contracts.
Falsification line: if the shaping of safety regulation is completed entirely through incremental legislation, with no single landmark accident as trigger — I would be glad to be wrong on this one.
Bet Five: A Red Flag Act, and a Belated Emancipation Day
Structural basis: the three mechanisms of the Red Flag Act — an asymmetric ledger, the capture of safety rhetoric, and a law that outlives the technology it regulates — none of them has ever gone out of service.
The bet: at least one major jurisdiction’s currently standing “form-regulating” AI legislation will, within a decade, be recognized as this round’s Red Flag Act, marked by statistically measurable industrial emigration — startups relocating their registration, R&D, or launches elsewhere; that jurisdiction will then face its own “Emancipation Day”-style overhaul — and by the day the law is amended, the industry will already have taken root in its own “Mannheim.” This book maintains its discipline of not taking sides within the framework and will not name which piece of legislation; the list will be filled in by the statistics ten years from now.
Bet Six: A Patent-Pool Moment (Conditional Bet)
Structural basis: The Patent War—the blocking equilibrium of a systems technology has only two exits: industry self-rescue (Albany, 1856), or external pressure that forces the gate open (the courts in 1911, the war in 1917).
The bet (conditional): if AI competition escalates into war-level state confrontation, 1917 will repeat itself — governments will force critical capabilities into a pool: cross-licensing, mandatory open-sourcing, or requisition, turning the castle into a public square overnight. Weak form (no war required): the copyright wars over data will end in a judicial interpretation that “keeps the industry alive,” backed by a collective licensing pool — a hybrid of 1911 and 1856. Looking back from that point, the legal fees both sides of today’s lawsuits are pouring in will turn out to be the same tuition as the decade-long Wright-Curtiss litigation.
Bet Seven: The Peasants’ War — The Reorganizers Become the Old Order
Structural basis: the second act of Gutenberg’s Bankruptcy and Luther’s Viral Rise — in 1525, the very same printing press and the very same eight-page pamphlet playbook turned its muzzle around with nothing more than a change of author; the reorganizer, having just settled into power, became the old order for the next wave of dissemination, and Luther’s response was to call for suppression.
The bet: the reorganizer enthroned in this round will re-stage Luther in 1525. There are two forms it could take, and hitting either counts as a match: first, some biggest winner enthroned by openness (at the model layer or the application layer) will, once seated, publicly lobby to restrict the very openness that enthroned it — the classic opening move of demanding a red flag for one’s rivals while reserving a funeral for oneself; second, generative tools will be turned around on a mass scale, and among the first trust levees to be swept away will be the reorganizer’s own platform — content generated by its own model flooding its own distribution and review systems.
Standby Observation: Operation Ivy Bells
During the Cold War, American submarines fitted eavesdropping pods onto Soviet undersea cables near Soviet shores and ran them undisturbed for a decade — the Soviet military believed its own coastal cables were absolutely secure, and vast amounts of communication went out unencrypted, in the clear. Trust in a channel has nothing to do with its actual security, and everything to do with its invisibility (The Morning the Cable Was Cut has already established the positive case: assume the channel is being read).
The observation: this round will see an “Operation Ivy Bells-style disclosure”—some widely trusted cloud, API, or data channel will be shown to have been systematically read for years (what is read need not be content; metadata alone is enough). At that point, countless companies will discover that they have spent all these years transmitting in the clear. This one carries no falsification line—its very nature dictates that no disclosure does not mean it never happened.
Signature
All seven bets share a single mother theme: the parameters may be wrong, the dates may be wrong, but the point of ignition will not be wrong by much — because the way the fuel is stacked has not changed in five hundred years.
This chapter is signed: July 2026.
Please dog-ear this page. Whoever turns back to it a decade from now, please check every item off one by one — every bet that misses will have planted a boundary marker for this method; every bet that lands, please answer a harder question:
You were there. What did you do?