Description du poste
Disclosure: adapted from my essay at piszczek.pl with the help of an AI model and reviewed by me before posting. This is the first essay of a series; the second is here.
Epistemic status: a framing I find useful and intend to test, not a result. The underlying law (scarcity relocates) is old and not mine; the definition of slop and the two-axis split are the parts I'd like criticized.
The mechanism. Technology doesn't destroy value, it relocates scarcity, and value pools at the next binding constraint. Aluminium was served to Napoleon III's most honoured guests until electrolysis turned it into sandwich wrap. Nordhaus estimated that the real price of artificial light fell by about four orders of magnitude in two centuries. Each time the commodity survives and its pricing power dies. I'm not claiming a conservation law. Value lives in differences the way work lives in gradients, and a technology flattens exactly one gradient.
What's being flattened now. Execution: competent prose, working code, plausible analysis. The mistake I see on both sides of the debate is treating execution as the top of the ladder. Its cost used to hide the judgment of whether a thing should exist, the verification of whether it's true, and the accountability of whoever signed it.
Two axes that get conflated.
• Capability, ordinary to exceptional. Machines compress the middle, and the far tail's relative worth rises (Rosen's superstar economics, applied again).• Exposure, anonymous to accountable. This axis has nothing to do with how good the work is. It measures who stands behind it.The definition. Most definitions of slop are aesthetic (generic, soulless), and those expire as models improve. An economic definition doesn't: slop is content nobody is exposed on. Zahavi's handicap principle says a signal is honest only when it's costly for the sender. A claim that costs nothing to make carries no information about the claimant, however well it reads.
Implications I'd bet on.
• AI-text detector