How predictable is technological progress?

Everyone knows that Moore’s law says that computers get cheaper at an exponential rate.  What is not as well known is that many other technologies that have nothing to do with computers obey a similar law. Costs for DNA sequencing, some forms of renewable energy, chemical processes and consumer goods have also dropped at an exponential rate, even if the rates vary and are typically slower than for computers. Doyne Farmer and Francois Lafond from Oxford University's Mathematical Institute and Institute for New Economic Thinking at the Oxford Martin School have developed a mathematical model for explaining how the uncertainties in applying Moore’s law grow in time and show both how to forecast future costs and how to predict the accuracy of these forecasts across technologies. They analyse historical data on the costs of many different technologies and test their methods by pretending to be in the past and predicting the present. This makes it clear how to combine and compare the results of forecasting different technologies, using results from forecasting technology A to get confidence in the ability to forecast technology B.   

Their analysis makes it clear that technologies are very different in the rate at which they improve. For example consider technologies for generating electricity. Once adjusted for inflation, the price of fossil fuel technologies such as coal, oil and natural gas is close to what it was a century ago. In contrast, solar photovoltaic modules have dropped in price at a rate of about 10% per year and are a factor of several thousand cheaper than they were when they were introduced in the mid 1950s. The analysis of many technologies indicates that while these trends are not ironclad, they do tend to be persistent. Forecasts show that solar photovoltaic modules are very likely to continue to get cheaper, and that if other factors such as distribution costs come down too, their electricity is likely to be cheaper than conventional sources within two decades. That said, there is a small risk that solar photovoltaic modules will be as or more expensive than they are now. The fact that it is very likely that solar energy will get cheaper is one of the few pieces of good news concerning climate change.

Doyne and Francois's paper can be found here.