The Trillion-Fold Fact

In 2001, William Nordhaus published a paper that did something the economics of technology had never quite done before: it went back to the original machines and measured. Not estimated, not approximated, not inferred from secondary sources—measured the actual cost of performing a standardized computation on each successive generation of computing technology, from the mechanical calculators of the 1850s to the integrated circuits of the 1990s, adjusted for inflation, expressed in common units. The result was a fact so extreme that it required a new vocabulary to comprehend. The cost of computation had declined by approximately one trillion over a hundred and fifty years—not from a single technology's improvement, but through five entirely different technological paradigms, each driven by different physics, invented by different people in different decades for different purposes, each resuming the same relentless slope the moment the previous paradigm hit its limits. Mechanical calculators gave