How Does a Technology Improve? - An Argument for Reinvesting in the Light Bulb

Published: October 05, 2005 in Knowledge@Emory

Each year, top minds in research and development are faced with a daunting task: To review scores of promising new technologies and determine which ones will receive funding—and which ones won’t. Their decisions not only affect their company’s bottom lines but may also influence the global economy—which makes it all the more unfortunate, says Ashish Sood, an assistant professor of marketing at Emory University’s Goizueta Business School, that often these R&D executives are making such critical choices based on faulty logic.

"These managers need to make decisions every year to determine what products to push and what products not to push,” Sood explains.  “If they had a better knowledge of the patterns of how new technologies evolve, they could possibly identify the more promising technologies and weed out the less promising technologies. In this way, their R&D dollars would be better spent."

By taking such a stand, Sood is challenging a long-standing, long-accepted theory in the field of technological development called the S-curve theory. This theory is so stubbornly entrenched in R&D circles as the way to measure a technology, Sood says, it’s hardly ever been questioned—even though there is little evidence to support it.

 

That is, until now.

In a new paper entitled “Technological Evolution and Radical Innovation,” published in July by the Journal of Marketing, Sood and his colleague, Gerard Tellis, a professor of marketing at the University of Southern California, contend the old S-curve theory is wrong. And Sood says once enough decision-makers are convinced of that, the ripples could be felt across the entire U.S. economy.

The essence of the S-curve theory

The existing S-curve theory states that a new technology invariably starts out less effective than its competition, but eventually surpasses the old technology on performance. The theory states the old and new technologies intersect just that once—as the new surpasses the old—and then proposes that the new technology will eventually reach a plateau of its own. Once that plateau is reached, so the theory says, that new technology has reached full maturity—and little, if any, improvement is possible beyond that point.

 

As Sood and Tellis write in their paper, proponents of the S-curve theory warn “even though managers might be able to squeeze out improvement from an old technology when it has reached a plateau, improvement is typically costly, short-lived and small.” As a result, R&D managers should “abandon a maturing technology and embrace a new one in an attempt to stay competitive.”

This ongoing cycle, Sood explains, has continued for decades, despite any substantial proof supporting the logic of the S curve. "These books [and the S-curve theory] have become so well-entrenched in the minds of technology managers that they always think of an S-curve," Sood says. "It seems almost intuitive—, 'Why wouldn't it evolve in the shape of an S-curve?' There were very few people questioning it, even though there was little empirical evidence backing it up.  The appealing simplicity of the model misleads people into believing that a complex phenomenon like technological evolution can be captured easily.

For their research, Sood and Tellis decided to finally test the theory. The researchers painstakingly collected and analyzed data on the development of 23 different technologies—from computer printers to monitors to memory devices, analgesics to data transfer technologies —throughout business history, studying such sources as technical journals, industry publications, white papers from R&D firms, annual reports from industry publications, press releases, museum records and other sources.

Of the technologies they studied, 80% to 90% did not follow the S curve. Some new technologies, they note, performed better than existing technologies immediately—a scenario the S curve does not allow for—while some older technologies did improve, even after reaching the so-called plateau that the S curve proposes marks a technology’s “maturity.” In fact, in some industries, the old technologies improved to be much better than the new technologies. For example, Sood points out in the research, “Optical memory was introduced in the desktop memory category with a massive advantage over magnetic memory (to an order of 30 times) but this technology was not able to maintain its lead over magnetic technology when in 1997-1998 magnetic memory regained leadership by offering much higher areal density (bits per square inch of recording media) than optical memory.” Sood and Tellis also say their research indicates that technologies can plateau and then improve at irregular intervals through their life cycle in the form of step functions.

In short, the work finds fault with much of the S curve theory’s components. The authors strikingly conclude “using the S curve to predict the performance of a technology is quite risky and may be misleading,” and add “an analyst expecting a single S curve may prematurely abandon a promising technology at the first plateau in performance.”

“Any manager that believes in the theory would stop investing in a technology as soon as it plateaus," Sood explains. "They would think, ‘The technology seems to have reached its limits, and it is time to push a new technology.' But that's premature. Our research showed that a lot of substantial improvement in performance comes before or after the first plateau. Investors need to be patient and managers need to persevere in order to bring a new technology to fruition. "


The conclusion has widespread implications for U.S. businesses, if not the entire economy, says Sood. That’s because managers who ditch the old theory may begin actually directing money back into existing technologies that would have previously been considered “mature.”

 

"The other major finding of our research is that new technologies perform better than existing technologies in almost half the cases right from the time of introduction. Moreover, new technologies come as much from new entrants as from large incumbents. This implies that relying on the status quo is deadly for any firm,” cautions Sood. As soon as people start removing their belief in the ‘Theory of S-curves,’ “it will change the way they invest in technology and how they assess competitive threats,” he says, adding, “We might even see a lot of conventional technologies start to improve, [as money is put into existing technology] rather than money being spent on far-fetched technology."

An example, Sood says, would be that old standby of the American workplace: the fluorescent light. Lighting companies dedicated much of their energies toward developing these lights in the early 20th century, but reduced spending on fluorescent technology around World War II. The result? Today's fluorescent lights are not much more efficient than their predecessors in the 1950s. Sood suggests spending on any existing technology might end specifically because R&D decision makers believe the technology had reached its plateau—and, as such, would not get any better, no matter what resources were thrown at it. Lighting companies instead turned their attention to such trendy technologies as sodium vapor lighting and others.

But Sood contends it is likely old technologies like fluorescents can be improved. If they were, it could save companies untold amount of money in years to come. And that’s just one small example, he says.

As convincing as Sood believes his work to be, however, he knows old theories die hard. This is especially true as "our paper pushes for development of a new theory, and dropping of the old theory." Consequently, there remain some old S-curve supporters who are skeptical of his research—"Some of the proponents of the old theory are asking, 'What are you talking about? How can you replace a theory with no-theory?'” Sood says—but he adds many others in his field are excited by the possibilities his research has created. "This opens up a lot of other research questions," Sood says.

Ideally, those questions would lead to more research and, eventually, a new theory that could guide R&D managers through their crucial decision-making process. This is even more important since the rate of technological change and number of new technologies is increasing all the time.

The evolution of technology—like the stock market—is very difficult to predict. "The evolution of a technology is quite unpredictable," explains Sood. "Researchers have been working for years but have not been able to create a good predictable model [for the stock market] and the next day's [stock] price is completely random. So I'm not sure what the new solution will be. But it's a very exciting time to improve our understanding of this phenomenon and even small improvements in forecasting the path of technological evolution would have massive benefits."

Print

Here's what you think...

Be the First to Comment on This Article.

 

Sign In to Join the Discussion

Email Address:   
Password: 

Not a member?
Sign Up for Knowledge@Emory

Tools

Print Send a Comment
Adjust font size:
8pt10pt12pt14pt

Knowledge@Emory