Read from top to bottom: unstated foundations, then historical claims, then the analogies that carry those claims into the present, then the claim the conclusions depend on most, then the conclusions. Click any node. Gold shows what it rests on; cyan shows what depends on it. Red dashed nodes are premises the essay needs but never states.
The essay's main move is to substitute one word for another. Flip the switch: each pillar of seventeenth-century mercantilism is rewritten with intelligence in place of wealth, and the logic stays exactly the same.
This is the fourth pillar, and the place where the essay's analogy has the most force. When farming needed fewer hands, the people it released went on to do everything else. Drag through time. Then switch to the essay's speculative cognitive version.
Gold: England & Wales, adult males (Cambridge Group, Shaw-Taylor 2024). Grey: United States labour force (≈75% in 1800 per Historical Statistics of the US; 1860–90 per Gilder Lehrman; ≈40% in 1900 and ≈1–2% today, rounded). Cyan dashes: the essay's speculative cognitive version, not data.
"I can barely keep a houseplant alive." The point: being unable to produce food stopped mattering economically once food became abundant. The essay predicts the same decoupling for intelligence. Watch the token move.
Before steam, Britons chose to work more, because tea, sugar, tobacco and calico gave them something new to want. The essay predicts the same for thinking: when attention pays off more, people will choose to spend more of it.
One square per day. Drag to add the days people chose to work.
The 250→300 figure is the essay's, following Appleby; the term is Jan de Vries's. Voth's estimates do find a much longer working year in England between 1760 and 1830. He also finds that consumption per head, once adjusted for the leisure given up, barely changed. See Act 08.
Economists call this the price elasticity of demand (ε). When the price of intelligence falls, ε decides whether we buy only a little more of it (ε<1, like food) or so much more that total spending rises (ε>1, like Jevons's coal). The mercantilist and the Malthusian assume the first. The essay bets on the second.
Constant-elasticity demand: quantity = price−ε, spending = price1−ε, both on log scales. This is a teaching model, not a forecast. Faint curves show ε = 0.3 and ε = 1.6 for comparison.
Farming released its workers because demand for food is inelastic (Engel's law: as people get richer, the share of income they spend on food falls). Food became cheap, people did not eat proportionately more, and the labour left the fields. If demand for intelligence is instead elastic, as the essay hopes, then by the farm analogy the cognitive sector should grow, not shrink. The two halves of the essay therefore pull against each other unless one more premise holds: the growth in demand is met mostly by AI, while humans move from producing intelligence to directing it. That premise is plausible, but it carries real weight and the essay leaves it unstated. The keystone holds only with it.
The essay's boldest extension: if peasants went from being factors of production to being sources of demand, could AI systems make the same climb? The dashed rungs are hypothetical.
The essay's own hedge: "Some of those fears are justified… powerful autonomous systems require constraints." Its claim is narrower than it first sounds: risk does not by itself justify permanent servitude.
The essay's typology on one plane. Across: is intelligence hoarded or diffused? Up and down: is it used for conquest, treasure and status or for production and creation? Hover over each figure. The positions are my reading of the essay's argument, not measurements.
Nine objections, ranked by how much damage they could do (my assessment). Click a card to see the essay's best reply.
An argument like this one becomes testable once you say what you would expect to observe. Signals to watch: