The Sweet Spot of Research & Development
Digital manufacturing covers a broad range of technologies and trying to predict their evolution and course can be an uncertain process. As a company, you can forgo the latest technologies as a passing fad and really miss out as your competitors embrace the cutting edge and advance. Think of how Blackberries were overtaken by smart phones. On the other hand, if a company is too eager to embrace the latest and greatest, you may be seen as foolish for embracing an unproven technology. Remember when we were all going to be driving fuel cell vehicles by now? General Motors invested millions only to see lithium ion batteries come out as the preferred power generation for cars.
So how do companies decide when to embrace technology and when to take the wait and see approach? One “law” that is used as guidance for computer chips is Moore’s Law, first stated by Gordon Moore, one of the founders of Intel. He observed that over the history of computing hardware, the number of transistors on integrated circuits doubles approximately every two years. Today chip performance usually doubles every 18 months but it nevertheless proved to be an accurate predictor, so much so that the “law” is used in the semiconductor industry to guide long term planning, set R&D targets and plan for capital investment.
However, these laws are not always reliable because progress usually isn’t predictable or smooth. A paper by Gareth James and Gerard Tellis, professors at the USC Marshall School of Business and their co-authors Ashish Sood, at Emory and Ji Zhu at the University of Michigan, concludes that Moore's Law does not apply for most industries, including the PC industry. The paper titled, "Predicting the Path of Technological Innovation: SAW vs. Moore, Bass, Gompertz, and Kryder," is in the current issue of Marketing Science.
Many companies in the high-tech arena use Moore's Law and other similar heuristics to predict the path of evolution of competing technologies and to decide when and where to invest in research and development or new product development. The paper's researchers claim that these models are outdated and inaccurate, so they’ve come up with their own model that is based on the tendency of technological developments to progress in fits and starts.
Their paper offers a new model, Step and Wait (SAW), which more accurately tracks the path of technological evolution in six markets that the authors tested. According to the researchers, Moore's Law and others such as Kryder's Law and Gompertz Law predict a smooth increasing exponential curve for the improvement in performance of various technologies. In contrast, the authors found that the performance of most technologies proceeds in steps (or jumps) of big improvements interspersed with waits (or periods of no growth in performance). The steps can be big or small and the waiting periods can be short or long.
The researchers postulate that greater support for innovation means new technologies will improve in larger and more frequent steps than old technologies have in the past. They chalked this up to higher R&D spending, the availability of better tools and the fact that countries are now investing more in research. They also noted that as the number of companies entering a new field increases the size of the step and the length of the wait tend to change.
The study looked at 26 technologies in six different markets: external lighting, desktop printers, display monitors, desktop memory, data transfer and car batteries. The choice proved to be a good sampling as it represented more than a century of diverse technological innovation. To conduct the study, they had the advantage of using historical records to track performance steps and waiting periods to obtain averages for each market. "We looked at the forest rather than the trees and see 'steps' and 'waits' across a variety of technologies," Tellis said.
Some examples of the study show that in lighting, the predicted step size as a percentage improvement in the performance for light emitting diode lamps was 0.34% with a mean waiting time between steps of 3.6 years. If we look at traditional incandescent lighting, the step was 0.11% with the wait time of almost 20 years. So who was the big performer? Optical fibers used in networking showed some impressive results with 2.19% per step with the waiting period between the steps of less than 2 years.
All this data allowed the researchers to obtain better predictions for technology progression than those using traditional “laws.” They suggest that the SAW can be used to predict the nature of the threat posed by a new, competing technology by using the model to more precisely identifying the steps and waiting periods involved. The sweet spot is in knowing which technology to back based on predicting when a new technology is going to have a jump in performance. This model might help companies and investors to reach those conclusions. So looking back, perhaps Sony wouldn’t have invested in cathode ray tube televisions when Samsung put their resources into LCDs.
