A Map of the Next Migration of Scarcity

June 17, 1914. The House of Commons, London.

Winston Churchill, First Lord of the Admiralty, rose to propose that the government acquire a majority stake in a Persian oil company. The opponents’ case sounded entirely respectable: for the government to enter the oil business was unheard of. Churchill’s case was simpler. The Royal Navy had already decided to abandon coal for oil, yet Britain produced not a drop of oil at home; the fleet’s fuel supply could not be left hanging on someone else’s oil fields. According to the record, the proposal passed by an overwhelming majority.

Eleven days later, a shot rang out in Sarajevo. Within weeks, Europe was at war—and an oil-burning fleet sailed into the petroleum century carrying a newly signed stake in an oil field.

The Migration of Scarcity recorded the outline of this answer: seeing where scarcity will migrate earns a passing grade; buying the land before it arrives earns full marks. This chapter unfolds that test into a map for the current cycle: five stops toward which scarcity is migrating as intelligence becomes abundant. At each stop, it marks the arrival signal, the positioning move, and the trap on the platform.

First, a statement of discipline in drawing analogies: the map bears a date. It was written in 2026, and the order and weight of the five stops will drift. Each stop therefore includes a “departure signal.” Scarcity has no terminal, only a next stop. (The description of current conditions at each stop in this chapter must be checked paragraph by paragraph for timeliness before publication.)

Stop One: Data—The Mine Is Exhausted; the Crop Is Becoming Valuable

What is becoming abundant is “what the internet has already said”: the marginal value of public text as training fuel is falling steadily. Scarcity is migrating toward three kinds of things no one has recorded: process data (the intermediate steps an expert takes when making a judgment, rather than the conclusion), correction data (errors that users correct after a system is deployed—the most honest teacher in the world), and physical data (the world inside sensors, not the world inside documents).

The historical rhyme is already on record in The Migration of Scarcity: NBC originally produced content to sell more radios. Read the giants’ free offerings today in the same way. Model companies build free applications not for the applications’ revenue, but for the closed loop of feedback from real-world scenarios. They are already buying land at this stop.

Data is not oil; it is a crop: the stock that has been scraped will run out, while what grows inside workflows regenerates every day. There is therefore only one positioning move: own a workflow that continuously produces process and correction data, rather than merely owning a database. Arrival signal: model companies begin acquiring application and service businesses with “no technical content”—they are not buying revenue, but farmland. Trap: treating a stock database as a moat—the data version of the legacy-asset illusion.

Stop Two: Distribution—The Physics of the Entry Point Is Changing

The rhyming examples are ready-made: telecom operators that finished building the information superhighway became dumb pipes; newspapers that owned printing plants died when classified advertising was unbundled. This round goes one step further: the physics of the entry point itself is changing. “People find answers” is becoming “answers find people”; search boxes are giving way to conversation and delegated action. One step further still, when purchasing itself is handed to an agent, the concept of the “shelf” changes owners: your product is no longer displayed for a person to see, but for an algorithm to read.

Watch the queues for the arrival signal (criterion two in The Migration of Scarcity). Businesses once queued to bid on search keywords. In the future, whoever queues to compete for the position of being “called first by an agent” will have identified the new entry point. Positioning move: direct user relationships—lists, communities, and subscriptions—are the only assets that pass through no new entry point. Audit every other form of distribution using the three steps from The Morning the Cable Was Cut: dependence on an algorithm’s favor is a personal favor; protection by terms is a contract; only the ability to migrate the whole operation is redundancy. The trap is a question: Are your users yours, or does a platform rent them to you by the month? When prosperity is rented, the landlord holds the renewal right.

Stop Three: Trust—When Content Is Free, Verification Commands a Price

Synthetic media is pushing “content” toward infinite abundance. By the law, value migrates to its complement: what cannot be synthesized. Monarchs of the Radio counted three such things that are currently visible: live presence, long-term consistency, and accountable identity. What must be added at this stop is their industrial form.

There are two rhymes. The first is the birth of UL in The People Swept Away by Electricity: safety is never an attribute that comes bundled with a new technology, but a public good purchased through accidents—first fires and public outrage, then a certification industry. The second is the era of food adulteration at the end of the nineteenth century. When fraud became cheap enough to be everywhere, the trust machine called a “brand” became valuable for the first time. Packaging, trademarks, and a promise backed by a willingness to pay compensation were all, in essence, reputation pledged in advance. The business model of trust has never been virtue. It is collateral.

The positioning move at this stop is therefore to turn “someone is responsible” into a product: auditing, certification, insurance, guarantees, and verifiable identity and provenance. The UL of the AI era does not yet have a name—before 1894, neither did UL. Criterion: who in your product occupies the position where “saying the wrong thing carries a cost”? If the answer is “no one,” what you sell is becoming free. Trap: outsourcing trust to a platform or model provider. Trust is one of the few assets that, once outsourced, has simply been given away.

Arrival signal: the first public incident in which “a deepfake causes a major loss to a named person or organization” (Predictable Repetition places a bet on this). Departure signal: the day verification becomes infrastructure, freely available to everyone, the premium on trust migrates to the next stop.

Stop Four: Energy and the Physical Layer—A Rare Migration Back Upstream

Across the previous waves of revolution, value generally migrated toward the application layer. In this round, it has made a rare partial return upstream—because abundant intelligence is purchased with electricity, land, and buildings. Railway Mania already separated foundations from horse feed: GPUs are consumables that depreciate over three to five years; grid connections, substation capacity, land, cooling, and construction capability are the rails.

The rhyme is the three-layer foundation in The Morning the Cable Was Cut: Britain’s cable supremacy did not grow on the cables themselves, but on gutta-percha, cable-laying ships, and landing points. This round’s “landing point” is the point of interconnection with the grid. The arrival signal has already appeared: the real queue is not measured in chip orders, but in years of waiting for grid access.

Positioning move: long-term contracts, permits, and location. The balance sheet at this stop is the most honest, and the heaviest. There are two traps. First, hoarding consumables as though they were foundations—stockpiling chips at high prices is buying horse feed at high prices. Second, the capital call: take-or-pay long-term contracts will turn around and collect from you in a downturn (criterion two in Railway Mania). If you buy land at this stop, buy real land: things that will still be valuable on liquidation day after the collapse. The standard is in Insull’s Pyramid.

Stop Five: Taste—When Answers Are Free, Questions Become the Bottleneck

The terminal stop, and the hardest one to explain. The Migration of Scarcity closed with a sentence: when answers are nearly free, questions become the bottleneck. This stop supplies the mechanism and the rhyme.

Every wave of abundance has crowned a profession of “selectors”: abundant print crowned the editor; abundant film crowned the director; abundant information crowned curation and search. Selectors share one trait: they do not produce content; they produce a judgment about “what is worthwhile.” The selector crowned by abundant intelligence does not yet have a professional title, but the work is already clear: define problems, set standards, and exercise veto power over the system’s answers. The business version of the Petrov test in The Timetable War is here: taste is the ability to say “no” to an answer that is fluent, confident, and probably correct. A decision-maker who always accepts the system’s output is not a decision-maker, but an ornament.

Taste is difficult to outsource because it is bound to two other things: preference (what do you actually want?) and responsibility (who answers for a bad choice?). The positioning move is to institutionalize taste: write down standards, establish review, and, most importantly, maintain a refusal list. The final form of strategy is not a vision list, but a refusal list. Criterion: where in your company is “what we will not do” written down? Who guards it? How far is that person from the user?

This stop has no departure signal. Taste is the conserved quantity in the migration of scarcity—the terminal of every revolution.

The Whole-Map Test: How to Buy a Platform Ticket

1. Redraw the slope chart every quarter. List every link in your industry’s value chain and mark the ones whose cost curves are falling exponentially. Every part of your profit sitting on those links must begin moving.
2. Answer one sentence at each of the five stops. Data, distribution, trust, energy, and taste—for each, answer in one sentence: am I an owner, a tenant, or a passerby? Five “passersby” are not frightening. Five “tenants who think they are owners” are.
3. Bet on no more than two stops. Migration is a relay, not a single move. Historically, the capital efficiency of buying land at every stop has never beaten concentrating capital at the one stop you understand—Churchill bought oil; he did not buy the whole world.
4. Label every asset with its stop. Reclassify the balance sheet according to the five stops. Assets that cannot be assigned to a stop are probably still sitting on the previous station’s platform.

Eleven days after Churchill signed that stake, the shot was fired in Sarajevo. History did something rare: it graded an answer on the spot.

Your answer will not be graded by a war. Only by the ledger, and by depreciation, which is more patient than war.