January 25, 2013
Researchers Develop New Means Of Predicting The Future Of Technology
Alan McStravick for redOrbit.com — Your Universe Online
Gordon E. Moore, the founder of microchip giant Intel and who was responsible for the formulation of the law named for him, postulated in the 1970s that processor speeds for computers will have doubled every 18 months. While Moore´s Law does not enjoy popularity among technicians employed by different computer companies, the rule is generally still accepted.
A more accurate explanation of the law applies to the number of transistors on an affordable CPU and how they would double roughly every two years. The number of transistors and the speed of the processor are not necessarily correlative. As an example, if one looks at the processor speeds from the 1970´s to 2010, it would appear Moore´s Law is nearing its limit, if it hasn´t achieved that limit already.
If we look solely at speed increases of the last decade, the difference between 2000 and 2009 was practically negligible, with a range of speed increases going from 1.3 GHz to 2.8 GHz, showing that speeds barely doubled over most of the decade. Viewing through a prism of total number of transistors, however, shows a remarkable increase. In 2000, the number of transistors in the CPU was approximately 37.5 million. Fast forwarding to the end of the decade showed a total number of transistors in the CPU equaling an astounding 904 million.
Earlier this year, theoretical physicist Michio Kaku spelled out the reasons why he believed Moore´s Law was approaching a form of critical mass, stating that a perpetual doubling of capacity was an unsustainable possibility using the current computing infrastructure. And Kaku is not alone in seeing that the end is nigh for this established law.
In their paper, entitled “Predicting the Path of Technological Innovation: SAW vs. Moore, Bass, Gompertz and Kryder,” Gareth James and Gerard Tellis, professors at the USC Marshall School of Business, conclude that Moore´s Law will no longer apply to most industries, including the PC industry. The paper, published in the current issue of Marketing Science, was also co-authored by Ashish Sood of Emory University and Ji Zhu at the University of Michigan.
They believe the formulation of a new method for predicting the evolution of future technologies is needed in order to save tech giants millions of dollars in research and development. A new method would also serve as a more appropriate indicator for technology analysts and venture capitalists, helping them to determine which companies are on the right track.
The researchers believe their new model could supersede Moore´s Law, along with other similar heuristics, in predicting the path of evolution of competing technologies and could help investors decide into which tech companies they may want to funnel their millions of dollars for future research and new product development opportunities. The authors claim the old models have become outdated and are thus inaccurate.
The team´s new model, which they have called ℠Step and Wait´ (SAW), more accurately tracks the progress of technology evolution across the six markets they tested. While Moore´s Law, along with other models such as Kryder´s Law and Gompertz's Law, have predicted a smoothly increasing exponential curve for the improvement in performance of various technologies, the research team shows how their model predicts most technologies will proceed in steps of large improvement with short wait periods in between.
Investors would, with this new model, be responsible for trying to identify the sweet spot, or the time when a significant financial investment would best be placed behind an emerging technology poised for a jump, or step, in performance.
"We looked at the forest rather than the trees and see 'steps' and 'waits' across a variety of technologies," Tellis said. The authors did concede that no one law was applicable to every market. However, the authors looked at 26 individual technologies across six markets, ranging from lighting to automobile batteries. Their results showed the SAW model worked in all six markets. The other competing models did not enjoy the same remarkable track record.
To explore a current scenario, Tellis explains how tablet and mobile phone manufacturers could effectively leverage data obtained from the SAW model. “Any manager has first to break down his or her products into components, find components for each technology, and then predict the future path of those technologies. For example, the mobile phone consists of three important technological components: memory, display, or CPU, the first two of which the authors analyzed. Similarly, [with] tablets, manufacturers could rely on the figures for display and memory technologies.”
The authors cite Sony as another prescient example of how their SAW model could have saved that company money, not only in the department of research and development, but also in lost market share. As the liquid crystal display (LCD) revolution was taking hold, Sony maintained their investment in cathode ray tube (CRT) technology. LCD began gaining in popularity in 1996. Despite that trend, Sony continued to produce the bulkier CRTs until 2005, with their WEGA series of flatter CRT televisions. A lost decade, especially in technology, is a difficult prospect to return from. Sony, in conceding the error, was forced to enter into a joint venture with Samsung in 2006 to begin the manufacture of their LCD televisions.
According to the authors, they believe had the SAW model been developed previously, Sony, along with other companies, might have changed their course of product development. “Prediction of the next step size and wait time using SAW could have helped Sony´s managers make a timely investment in LCD technology,” the authors claimed in their study.
Each author brought their specific and particular skill set to the study. Gerard Tellis is the Neely Chair in American Enterprise at the University of Southern California Marshall School of Business. He has also recently authored a new book, titled “Unrelenting Innovation: How to Create a Culture for Market Dominance.” Gareth James, also of the Marshall School of Business at the University of Southern California, is a professor of statistics.
Ji Zhu is an associate professor of statistics, Department of Statistics, at the University of Michigan, Ann Arbor. And Ashish Sood is assistant professor of marketing at the Goizueta School of Business at Emory University.