The Future of Predicting
Today, MIT announced the results of the first quantitative analysis of six hypotheses (Moore’s Law, Wright’s, Law, lagged Wright, Goddard, SKC, and Nordhaus) used to predict technological advancement, to decide which one was the most accurate. According to the findings presented in the paper, “Statistical Basis for Predicting Technological Progress” by Béla Nagy, J. Doyne Farmer, Quan M. Bui, and Jessika E. Trancik, the two most accurate predictors of how rapidly a technology will advance are Wright’s Law and Moore’s Law. These results will greatly impact industries, investments, and policy-making decisions, particularly as they relate to climate change and technology.
Their findings suggest that technology does indeed tend to improve exponentially over time, and also that, except for tech-specific parameters, the processes responsible for generating the cost of technologies over time are generic. In other words, you can change the parts and still get the same sum as the whole, when it comes to rate of advancement. This means researchers can now pool data from different technologies to make improved forecasts and error estimates on specific technologies, particularly in the case of studying new tech trends that have limited data available.
It wasn’t that long ago that Bill Gates said to a crowd at the inaugural Techonomy conference in Lake Tahoe, CA, “We’ve all been spoiled and deeply confused by the IT model. Exponential improvement—that is rare” (CNet, Aug. 9, 2010). He went on to explain that while there are isolated segments of technology that do produce that rate of improvement, such as hard disk storage, fiber capacity, gene-sequencing rates, biological databases, and improvements in modeling software, other areas, particularly around energy, don’t tend to produce that kind of change, using light bulbs and batteries as examples.
The research presented today, and conducted by scientists at MIT and the Sante Fe Institute, suggests that there is more regularity to the advancement of technology than we previously thought. To quote their findings: “The historical data shows a strong tendency, across different types of technologies, toward consistent exponential growth rates.” This doesn’t mean that Gates was wrong, as the authors aptly point out a la Niels Bohr, “Prediction is very difficult, especially of the future.” However, it does mean that companies, economists, and policy makers can make predictions and decisions more confidently using these formulas.
These findings will help new technologies designed to decrease our dependency on oil and shrink our carbon footprint to present predictions that speak to the business people who make or break their success. The plain truth is, whether or not a product reaches the masses has nothing to do with its impact on the environment. Its impact on society is measured by its cost of production.
In “Energy Efficiency Economics and Policy”, authors Kenneth Gillingham, Richard G. Newell, and Karen Palmer note that it’s a battle of energy efficiency vs. economic efficiency. Businesses tend to decide with their wallets. Maximizing energy efficiency does not usually mean maximizing economic efficiency, at least as we understand it in the short term. The new study sheds new light on predictions for technologies in this arena, which will help developers to make stronger arguments for the success of their designs.
The MIT study began with the creation of a database that compiled government reports, market-research publications, research reports and other published sources for which there was at least ten years worth of consistent data available, and which contained metrics for both the rate of production and for some measure of improvement, according to MIT News. They then used “hindcasting” to analyze the data, which means they assessed which of the formulas best fit the actual pace of technological advances in past decades. Their findings also suggest that the elder of the two laws, Wright’s Law, formulated in 1936, is a better predictor over time than Moore’s Law, formulated by Gordon Moore, co-founder of Intel, in 1965, to describe the rate of improvement in computer chips over time.
These findings might not impact the fate of new green tech presented at the 2013 Eco-products International Fair (March 14-16), but they will make an impact in the future. Photovoltaics, which convert light into electricity at the atomic level, have an excellent future based on the new findings. This information comes at an important time in the green tech race. As the graph below from National Geographic Daily News reveals, our energy consumption is growing, perhaps not exponentially, but enough to make a big impact on our resources.
Last month, the president of the World Bank, Dr. Jim Yong Kim, urged governments to address climate change. “These are not just risks. They represent real consequences….serious consequences for the economic outlook,” he said. He cited damages and losses as a result of natural disasters, which have more than tripled over the past 30 years, often wiping out years of development efforts in a matter of days or even minutes. Now is the time for green tech to thrive, and if companies can’t allow it to do so simply on the premise that it is more friendly to our environment and consequently us, then perhaps they’ll be persuaded by the numbers after all.