Nothing over the weekend moved because the world ran out of something. Next year's fertilizer bill is being written by a Chinese export licence and a Russian export ban. A lending protocol lost $8.5 million to a vote somebody bought. A British power station was worth attacking because it sat under the legal threshold for reporting it. In each case the price was set by whoever controls the permission, not the material. The honest exception is memory, where the shortage is physical.
American stocks gave back 1.4 percent on the week, and Friday's gain came too late, changing the level without changing the count. Crude ran up roughly 5 percent for a second straight week to settle near $87, and it is selling off this morning into the Iran sanctions announcement. The long bond printed its high on 18 August and handed it back by Friday, leaving that high unretested. Spot bitcoin funds took in $2.6 billion, the biggest week of the year, and the coin is back near $77,000 after dipping Sunday: prices moved and positions did not.
A better tool that arrives cheaper gets used by everyone without a decision, so it lifts the floor. One that arrives expensive has to be assigned to particular jobs, and that is a skill most firms lack. In July American businesses bought six percent of their Anthropic tokens from its most capable model; OpenAI's flagship, at half the price, took twenty-five percent. We expect the premium tier to stay under fifteen percent in Ramp's December index, because the binding constraint is price rather than unfamiliarity. Against that, Anthropic's share of US businesses paying for AI rose faster in July than in any month this year. Call it the Dispersion Switch: when an advance arrives dearer rather than cheaper, the gains go to whoever built the skill to place it.
In the early 1980s János Komlós conjectured it never has to exceed one universal constant, however many objects or attributes are involved. Nikhil Bansal and Haotian Jiang moved that ceiling last autumn by tracking a second quantity: nudge one attribute, and how far do all the others move? They then picked the step that disturbed the rest least. What decides whether you land near the bound is not effort per criterion, but whether you are scoring your moves on the wrong axis.
Six hundred and sixty-six teams with a fixture list is an industry choosing what to optimise for.
Nobody needs a mechanical watch and the order takes months, which makes it a clean read on who still has surplus.
Alexander kept losing his voice on stage in the 1890s and no doctor could say why, so he set up mirrors and watched himself speak.
“All those who wish to change something in themselves must learn … to inhibit their immediate reaction to any stimulus to gain a desired end.”
– F. M. Alexander, The Use of the Self (1932)
The instant he decided to recite, before a sound came out, his head pulled back and his throat shortened. The hoarseness was not happening to him. It was the first thing he did, and trying harder produced a more careful version of the same pull.
What worked was refusing to speak for a moment after deciding to. Just declining to begin. This is why so much self-improvement is effort without movement: the person who resolves to listen better interrupts more attentively. Between deciding and moving is a gap your habit has not claimed.
Today's practice: pick one thing you do twenty times a day, standing up or reaching for your phone. Put a two-second stop between deciding and moving. Correct nothing, then go and feel what your neck does.
The textbook proof that intuition is broken was a basketball study: a player who had just scored was no more likely to score again. In 2018 Joshua Miller and Adam Sanjurjo showed the counting itself was biased. Keep only the coin flips that followed a head and you get less than fifty percent, because keeping only those is a selection. Correct for it and the conclusion reverses.
A systematic error is not a mistake in a head. It is a mistake in a procedure, and it repeats in the same direction. A random mistake shrinks as you gather more data; a systematic one gets more confident, because every observation arrives through the same crooked pipe.
The tool. Before accepting any comparison, finish this out loud: what decided which things I am comparing was ______, which pushes the answer ______. Complete it and you can size the error.
Explore this model →Read the rules this week, not just the prices. That is where the news was hiding.