The Experience Curve

Why Wrights Law > Moore's Law

Risk & Progress| A hub for essays that explore risk, human progress, and your potential. My mission is to educate, inspire, and invest in concepts that promote a better future. Subscriptions are free, paid subscribers gain access to the full archive, including the Pathways of Progress and Realize essay series.

---

What enables the “evaporation of everything?” What drives the incredible declines in the costs of goods that we have witnessed since the Industrial Revolution? As we have discussed, the accumulation of knowledge underpins all progress, but how exactly does knowledge accumulate? It turns out, there is a certain synergy that exists between innovation and production. It’s not enough for ideas to remain on pen and paper or their digital equipment. We learn best by doing: progress requires production. This phenomenon, known as the experience curve effect, explains how production improves quality and affordability.

Wrights Law > Moore's Law

In 1965, Gordon Moore forecasted that the cost of transistors on a microchip would halve about once every 18 months. Moore’s Law, as it became known, is a household name. Moore’s Law held up quite well, but in some sense, this may have been a quirk of history. It can be argued that Moore was focused on the wrong variable, confusing why and how the cost of transistors fell over time. Moore focused on cost as a function of time, but time alone does not make costs fall. Instead, the experience accumulated during production, the experience curve effect, caused the rapid advancements in computing. Indeed, slowing progress in the sector, “the end of Moore’s Law,” indicates that the law may have been misformulated from the beginning.

To better understand the synergy between innovation and production, and how the cost of technology, like a transistor, falls over time, we should look to a law developed 30 years before Moore’s: Wright’s Law. Wright's Law is named after Theodore Paul Wright, an engineer at Curtiss-Wright in the United States. In 1936, Wright noticed that with each doubling of aircraft production, the cost required for manufacturing a new unit decreased by approximately 15%. This observation has since been validated across many different sectors and technologies. Wright’s Law demonstrates that with increased production, companies gain experience, refine their craft, and become more efficient, which leads to reduced costs.

Unlike Moore’s Law which was more an observation of the chip industry over a short period, Wright’s Law is applicable across many industries. In a study comparing Moore and Wright, the Santa Fe Risk Institute concluded that using Wright’s Law, we can “forecast” future cost declines of any technology. They compared the forecast error rates of Moore's and Wright’s laws respectively across 62 technologies, finding that Wright’s consistently outperformed Moore's. Indeed, Gordon Moore’s forecasts, as incredible as they were, could have been more accurate by applying Wright’s Law to the chip industry instead.

Using Wright’s Law, we can forecast the price of a particular technology so long as the “learning rate,” (the percentage cost reduction that emerges with each doubling of production) is known. This is a bit more tricky than it sounds because we also need to be able to forecast the demand for that particular technology. Specifically, we need to know the price elasticity of demand, which is the response of the market demand to falling costs. With some products, demand can surge upward once a technology falls through a price threshold. A good example here is lithium batteries.

Lithium Ion batteries were first commercialized in the early 1990s. Adhering to the experience curve effects, their cost per kWh declined by about 10 percent per year until around 2005 when those declines slowed. A few years later, however, lithium battery prices became low enough that they could power electric vehicles. As those vehicles began selling in mass after 2013, they drove the demand and production of lithium batteries to new heights, thus resuming the learning curve effects and the rapid decline in prices.

When looking at cost declines in lithium batteries as a function of time, as Moore would, progress appears discontinuous, slowing down from about 2005 to 2013. However, when looking at cost declines as a function of production and experience, that is, accumulated knowledge, progress was obviously continuous. This is why Wright's Law is a better predictor of progress than Moore's. Knowledge and experience, not time, is the salient variable.

Wright's Law, although broadly applicable, is not universal. It applies well to processes involving repetition, such as those found in the manufacturing industry. Moreover, the “learning rate” varies from one technology to the next. Nonetheless, learning curve effects explain much of the cost reductions we see in everyday goods across all domains of life. The result is material abundance, or as

Gale Pooley

and

Marian L Tupy

call it, superabundance. It also explains why some industries see rapid productivity improvements, while others lag. See my essay on “cost disease” for a discussion of this phenomenon.

While Wright’s Law is concerned primarily with the cost of a particular technology, it should also be noted that the quality of the technology also tends to improve with production. To understand why, we can turn to the parable of the pottery class. The parable goes like this: A teacher divides his pottery class into two groups, grading Group 1 on quality and Group 2 on quantity. Group 2, will obviously produce more pottery peices than Group 1, but it will also make better quality pieces than Group 1. This is because Group 2 would have more iterations upon which to improve its craft. Production, that is, accumulated experience, drives a virtuous cycle of improvement in both cost and quality. Indeed, returning to the example of lithium batteries, the above graph illustrates a 97 percent cost reduction. It does not indicate, however, that the average energy density of lithium batteries also improved 3.4-fold in the same period.

Of course, some question the causality of this effect. Could it be that production and experience increase only because costs have fallen first? Researchers Francois Lafond, Diana Greenwald, and Doyne Farmer attempted to answer this in their paper, “Can Stimulating Demand Drive Costs Down?” To tease out causality, they examined the demand for military technology in the Second World War, where there would be no concerns about reverse causality. Their research concluded that as the government’s demand for weapons grew, experience from production led prices for weaponry to decline steeply. In other words, the demand drove prices down, not the other way around.

If this all sounds a bit counterintuitive, it is. Normally, we expect prices to rise when demand rises. At least, we certainly do not expect them to fall! But as we have seen in our study of human progress, and as I warned at the outset, nothing about an investigation of human progress conforms to our innate expectations. The factors that lead to progress are counterintuitive and thus progress itself is remarkably fragile.

Implications

The implications for Wright’s Law are interesting for it raises the specter that government may be able to accelerate progress in certain domains by subsidizing demand for a technology. In a sense, this could be a workable justification for industrial policy. Indeed, this appears to be the strategy used in China, where the government is subsidizing the demand for renewable energy technologies, including lithium batteries and solar. This, in turn, is leading to rapid improvements in China’s technological and industrial prowess, allowing China to catch up, and in some cases, leap past, international competition. We will revisit this and other ways to sustain and accelerate the advancement of technology in later essays.

---

You also may like….

Billionaire Planet

Sapien Sorcerers