wright's law
A formulation of the learning curve, named after WWII economist Theodore Wright, who found each doubling of cumulative airplane production reduced cost by a fixed 16%.
As defined in 64 | Ramez Naam on Renewable Energy and an Optimistic Future
Other definitions
Also called learning by doing; an empirical regularity describing how technology improves (gets cheaper or better) as cumulative production increases
293 | Doyne Farmer on Chaos, Crashes, and Economic Complexity
Where it comes up · 6
- AI 2027: month-by-month model of intelligence explosion — Scott Alexander & Daniel Kokotajlo1 claim
- Living with Exponential Change (with Azeem Azhar)1 claim
- AI 2027: month-by-month model of intelligence explosion — Scott Alexander & Daniel Kokotajlo1 claim
- Bill Gurley & Michael Mauboussin - Putting Theory into Practice - [Invest Like the Best, EP.370]1 claim
- 293 | Doyne Farmer on Chaos, Crashes, and Economic Complexity1 claim
- 64 | Ramez Naam on Renewable Energy and an Optimistic Future1 claim
Most important claims · 6 of 6Search all claims for “wright's law” →
- Solar, wind, and battery costs fall along learning curves: every doubling of cumulative manufacturing scale has, on average, brought solar's price down by about 30%—an instance of Wright's Law, originally observed in WWII airplane production where each doubling cut cost by 16%.64 | Ramez Naam on Renewable Energy and an Optimistic Future
- Under Wright's law, each doubling of cumulative output reduces cost per unit by about 20 percent, a relationship that fits lithium batteries and solar well and that outperformed Moore's Law in a Santa Fe Institute test of 60 technologies.Bill Gurley & Michael Mauboussin - Putting Theory into Practice - [Invest Like the Best, EP.370]
- Technological improvement following Wright's law (learning by doing) can be derived by assuming inventors throw darts at a dartboard at random and are just smart enough to recognize a better throw (as John Muth, the inventor of rational expectations, originally showed); extending this with a design-structure matrix capturing component interactions, Farmer's group derived that the more complicated and less modular a system is, the lower its improvement exponent and the slower it improves—a result that has been tested and appears largely true.293 | Doyne Farmer on Chaos, Crashes, and Economic Complexity
- Wright's Law states that the ability to improve efficiency in a manufacturing process is proportional to cumulative production volume; if superintelligences are producing 1 million robot units per month a year after their emergence, this production scale itself drives rapid efficiency improvements through learning-by-doing, enabling further manufacturing expansion.AI 2027: month-by-month model of intelligence explosion — Scott Alexander & Daniel Kokotajlo
- Superintelligences could scale up to producing a million humanoid robots a month within about a year of wanting them, because OpenAI-scale wealth could buy up car factories and convert them (the fastest historical analog being WWII bomber conversion, which took ~3 years despite being a comedy of errors), and superintelligence could do this roughly 3x faster by avoiding human bureaucratic mistakes.AI 2027: month-by-month model of intelligence explosion — Scott Alexander & Daniel Kokotajlo
- Wright's Law states that for every doubling of cumulative production, per-unit cost declines by a consistent percentage (about 15% for airframes) due to learning rates; it applies to many engineered, multi-component products including semiconductors, solar panels, and batteries, and is more predictive than Moore's Law for chip cost declines.Living with Exponential Change (with Azeem Azhar)