Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 18
--> 2 The Sources of Commercial Technological Innovation Don E. Kash George Mason University The debate over the role of research in the innovation of commercial technologies has demonstrated one thing. As a general rule, it is very difficult to identify and prove the contribution of research, and particularly fundamental research, to commercial innovation. Common sense tells us that the role of research vis-à-vis commercial technological innovation varies greatly from one area of technology to another. It also varies greatly over the lifetimes of particular technologies. What seems clear is that research is central to the innovation of some technologies—for example, pharmaceuticals, chemicals, and biotechnologies. It has marginal value in the innovation of other technologies. In seeking insight into the processes that deliver innovation, I found it useful to divide technologies into two groups: those that are simple and those that are complex. A simple technology can be understood in full detail by an individual expert sufficiently well, so that that individual can communicate all of the details of the process or product across time and distance to other experts. Alternatively, a complex technology is one where that kind of understanding and communication is not possible. Economic rewards from the innovation of simple technologies commonly flow from the ability to gain legal monopoly protection for the intellectual property involved. This is a pattern one sees manifested in the pharmaceutical and the chemical industries. In addition, simple technologies are commonly derived from research. Complex technologies manifest a different pattern. Economic benefits flow much more heavily from the ability to carry out repeated incremental innovations. It is difficult and sometimes simply not possible to use patents to protect complex technologies. The route to economic payoff is to incrementally enhance the complex technologies ahead of or in parallel with one's competitors. This is a pattern manifested in technologies ranging from computers and telecommunications technologies to corn planters. For very large numbers of complex technological innovations, the capacity to carry out systems integration is critical to successful innovation. My colleagues and I like to characterize this innovation process as involving synthesis—that is, complex technologies appear to benefit primarily from the ability to put diverse components and subsystems together and obtain a synergistic result. The technol-
OCR for page 19
--> ogy that comes out of this synthesis ordinarily has either performance or quality advantages or reduced production costs; under the best of circumstances, the innovation produces all three. A key point with regard to complex technologies is that there is no capacity for understanding them in detail. Certainly there is no such understanding of how the innovation of complex technologies occurs. Complex technologies are largely the result of trial and error. They build substantially on an accumulation of knowledge within an industry or a technological area, and they benefit especially from organizational arrangements with established organizational routines and heuristics that inform and guide the process of incremental innovation. It is generally true that incremental innovations of complex technologies involve a process of learning by organizational networks. The way the organizations interact in the networks is at least as important as the research, and in many instances a good bit more so. To set this in context it is useful to use a series of trajectories to illustrate what commonly occurs with regard to complex technologies. Complex technologies tend to be launched either by what are called radical innovations—that is, innovations that are first of a kind—or by what we call trajectory transitions. A trajectory transition occurs when the basic design or the technological platform at the center of a continuing series of incremental innovations changes fundamentally. In Figure 2.1 are three S-curves defining the trajectory of audio technologies. The bottom left curve indicates the audio technology that started with Edison's cylinder and then moved incrementally to the FIGURE 2.1 Trajectory of audio technologies.
OCR for page 20
--> long-playing record discs that we know. The second S-curve resulted from a trajectory transition triggered by the arrival of analog magnetic tapes. The third trajectory is the arrival of compact digital discs. It is commonly recognized that in the case of complex technologies, being the innovator that produces a new trajectory is not necessarily the same as being the winner in the economic competition. What is quite clear is that if you are first and you continue to incrementally innovate ahead of your competitors, then you do well. But if you are first and you stop being innovative, you don't do well at all. This distinction between simple and complex technologies is particularly important as we think about the future of commercial innovation. The importance of complex technologies in the economic marketplace is accelerating rapidly. Simple technologies represent a declining percentage of the value of exports. Complex technologies, on the other hand, are becoming more and more important. There are two trends at work. More and more of our commercial technologies are becoming complex. In addition, those that are already complex are becoming more so. Recall that success in the incremental part of the innovation of complex technologies comes from the ability to carry out synthesis. Next I will show what happens if we use this simple-versus-complex categorical distinction to look at exports. In Figures 2.2 and 2.3, we took the 30 most valuable goods exported worldwide and the 30 most valuable manufactured goods exported over the last 25 years and classified them by type of product and the type of process used to produce it (Figure 2.2). We classified each of these exports and the process used to produce it as either simple or complex. This has been done by asking those who were experts with regard to these technologies how they would classify each using our rule-of-thumb distinction. What we essentially have are four categories of technologies into which we could classify the 30 most valuable goods exported and the 30 most valuable manufactured goods exported over the last quarter-century. Those categories are simple/simple, simple/complex, complex/simple (that is complex process, simple product), and complex/complex (Figure 2.4). I might note for you in this connection that the 30 most valuable goods exported represent nearly half of the total value of all exports in both 1970 and 1995—specifically, 48 percent in 1970 and 46 percent in 1995. FIGURE 2.2 Thirty most valuable goods exported.
OCR for page 21
--> FIGURE 2.3 The 30 top manufactured goods exported. Simple Process/Simple Product Simple Process/Complex Product Complex Process/Simple Product Complex Process/Complex Product FIGURE 2.4 Four categories of technology for classification of high-value exported goods. In 1970 the simple/simple group represented 56 percent of the value of the 30 products and the complex/complex group represented 32 percent. By 1995 the complex/complex category included 54 percent of the total value of the top 30 product exports, while the simple/simple category declined to 20 percent. From this it is clear that those who want to be where the money is in the future want to be in complex/complex technologies. The fact of course is that over the last 25 years simple technologies have become complex technologies. If non-manufactured exports are taken out of the picture and only manufactured exports are examined, what is being taken out is crude oil, we see a similar pattern. That is, even with regard to manufactured goods in 1970, 52 percent of the top 30 manufactured goods exported were in the simple/ simple category (see Figure 2.3). By 1980 that category had dropped to 30 percent, while the complex/ complex category had grown from 37 percent to 54 percent. By 1995, the simple/simple group of manufactured goods made up 12 percent of the market, and the complex/complex group represented 62 percent. Again, if we look to the future, it is the capacity to innovate complex technologies (particularly the capacity to successfully carry out incremental innovations) that is the key to economic success. The next question is, How is the United States doing if we apply these categories? What we find is peculiar. It is an anomaly and is inconsistent with almost everything else in the literature. If we look at
OCR for page 22
--> FIGURE 2.5 The U.S. and Japanese trade balances, 1995. the trade surplus or deficit of the United States using these three categories and we do the same thing for Japan, what we find is that in 1980 the United States had a $12 billion trade surplus in complex/complex technologies. Japan had a $40 billion surplus. If we look at the situation in 1995, what we find is that the United States had a $50 billion deficit and Japan had a $170 billion surplus (Figure 2.5). It is important to note that this is a period in which the value of the dollar vis-à-vis the yen has decreased by roughly 50 percent. In the 1990s the United States has been in economic expansion and Japan has been in what is termed a recession. In truth, by almost every measure the United States is currently booming, while Japan is in the doldrums. Yet if we break out the top 30 technology product exports for the world and compare the performance of the United States and Japan using balance-of-payments data, this pattern is a particularly striking anomaly (Table 2.1). In truth, no matter how we look at it, the United States does not do very well in this complex/complex category. Table 2.1 lists the product categories in the U.N. trade data. The largest category is automobiles, in which the United States runs huge trade deficits. This deficit remains very large despite the recovery of the U.S. auto industry. Japan runs a deficit in a single category, aircraft manufacturing. I conclude by saying that we cannot explain the above findings by any study of relative research and development efforts in the two countries. It has to be explained in other ways. I think very clearly, one of the most important of those categories is clearly the nature of organizational relationships. Discussion Richard K. Koehn, University of Utah: I'm a bit worried about what you've just said. Although the conclusions appear to be fine, they depend on this distinction between simple and complex. What got me worrying about this is that you classified drugs and the pharmaceutical industry as simple. Related to this are two major points that you did not mention but which I believe are critical. One is the degree to
OCR for page 23
--> TABLE 2.1 Balance of Payments for Complex/Complex Technologies, United States and Japan, 1995 Product Category United States Japan 781 Pass Motor Veh excl. Buses -49,327,745 31,689,249 776 Transistors, Valves, etc. -5,245,303 28,586,562 752 Automatic Data Process Equip -12,345,386 6,733,480 784 Motor Veh Parts, Acc NES 2,660,637 18,201,864 764 Telecom Equip Pts, Acc NES -805,052 12,398,541 759 Office ADP Mach, Pts, Acc NES -6, 185,914 11,044,216 778 Electrical Machinery NES -1,190,106 11,443,713 792 Aircraft, etc. 19,225,396 -2,768,035 728 Other mach. for Spcl Indus 2,688,104 9,550,854 713 Internal Combus Piston Engine -2,603,547 13, l 19,205 874 Measuring Control. Equip 6,139,574 4,237,599 782 Lorries -4,874,498 9,615,997 793 Ships & Boats, etc. 1,243,569 10,932,151 744 Mechanical Handling Equip 424,584 4,850,386 Subtotal -50,195,597 169,627,782 SOURCE: United Nations, International Trade Statistics Yearbook, Vol. H: Trade by Commodity, Commodity Matrix Tables (New York: United Nations, 1996). which an industrial sector is regulated, and the other is the rate of innovation, which you characterize as a trial-and-error process. The reason you don't need a patent in the computer chip industry is because the product cycle is only 18 months and the company will be out of the market long before the patent is ever issued. So, a trade secret is used to control the information, and economic success is dependent on market penetration. You could do the same thing in the drug industry if the government didn't prevent you from killing people while you were carrying out the trial-and-error studies. But, of course, it does. So this industrial sector depends on patents because patenting drugs is the only way to recoup the $300 million investment in developing the drug. If a company could not obtain that monopoly, there would be no drug innovation, because there would be no economic benefit resulting from development of a drug. The above seems to go well beyond your classification of simple and complex, yet it seems to be critical to the conclusions that you have drawn. Don E. Kash: I don't disagree with anything that you have said. A proper classification system is very difficult to develop. What seemed to us to be the case is that there were very different patterns of innovation that occurred as we studied one technology area and then another. What we were looking for was a classification system that did not require us to consider every technology independently, while nonetheless ensuring that we did not discuss all technology as if it were the same. I would be interested in any suggestions as to how you might do it differently. Richard K. Koehn: First of all, I would do it industrial sector by industrial sector.
OCR for page 24
--> Don E. Kash: Categorizing technologies by industrial sector can be difficult also. One year when we were looking at the data, the fastest growing category was miscellaneous. Why was that? As technologies progress from being mechanically based to electromechanically based to electrooptical-mechanically based, knowing how to classify them becomes exceptionally difficult. So, the people who are collecting data at ports and filling out surveys in companies look for the right place to put them, and there is never an appropriate category. For example, is software a sector? Richard K. Koehn: Yes. Don E. Kash: I can tell you that nobody collects data on software, except the software industry. We have very good data on packaged software, but people in the software industry keep telling me that there is all kinds of software embedded in hardware. My problem with your sector-by-sector definition is that the sectors are continually undergoing metamorphosis, and that makes the data very difficult to track. Richard C. Alkire, University of Illinois at Urbana-Champaign: My question is in the nesting of S-shaped curves in Figure 2.1 on the development of audio technology. You made the point that research is very important at the initiation of a new curve. I wonder if you could consider scientific and engineering research separately, where engineering research determines whether the curve has an S shape or is a constant. Don E. Kash: First, in terms of asking companies questions about research, research was whatever the people in the company said it was. With this approach research can mean many different things. Second, much of what engineers do, which is terribly important in the innovation process, is not research and is not classified as research. It involves design, experimentation, and problem solving, which are never classified as research. I think this is a problem. I would expect this group to be very much like I am. Anything that contributes to innovation is research where innovation is, by definition, the first introduction to the market. Many people who are very sensitive about defending the role of research have a tendency to lump this type of engineering into the research category. Research is, if you look over time, one of those wonderfully expandable categories. For example, in a company, it shrinks and expands, in part, based upon tax definitions, but it also shrinks and expands depending on what you're trying to defend. Paul Anderson, DuPont-Merck Pharmaceuticals: I too was wondering what your definitions were for complex versus simple. I certainly would agree with you that if you look at a product from the pharmaceutical industry, then it's simple. But if you look at the process for getting to it, then it's complex/complex, because it requires at least eight different disciplines working together as a team over a long period of time (10 years is normal), even though very smart people are involved. In some of these areas it may help to look at the product versus the process to get to the product, when you classify the technology. Don E. Kash: We put the pharmaceutical industry in the complex process, simple product category. We put most of the chemical products in that same category, unless you go back to 1970. In 1970, an isopropyl alcohol plant tended to be run by people who turned valves on and off. Now the same plant is full of computers and sensors and controls and has become a very complex operation.
OCR for page 25
--> Thomas A. Manuel, Air Products and Chemicals Inc.: I think we may be getting into a semantic bog, and I have to apologize for taking us one step further, but I shared the groans about research being everything that led from the idea to the commercial product. Research is usually a close bedfellow of development, which tends to include much of the engineering community. I find it helpful to split them, because we start with science and we wind up with statistics on technologies. I would submit that, as a practical matter, there's something in between the science becoming a technology, which is that exercise we call development. That's certainly true in the chemical industry. To make one further comment in a different direction: I love S-curves, and I cook enough numbers to get them for performance. But it is important to realize that, if you have the same time axis, the slope of performance improvement is not the same as the slope of perceived value, which in turn means that the product life cycle doesn't have the same slope either. That's why in the chemical industry, our products have very long life cycles. If you look at the 50 top industrial chemicals of 1996, you will find a very great similarity to those of 1966, 1976, or some other time period. I'm a believer in some sort of sorting by sectors as well; otherwise there are no landmarks. Don E. Kash: Well, I don't really disagree with you at all on sorting by sectors. What we did in accumulating the numbers was to characterize each first as a sector. Then the effort was to try to collapse those numbers into meaningful units given the changes that are always taking place. When it comes to making a distinction between research and some other types of activities, I repeat that, for us, research is whatever the people in the industry tell us it is. It is perfectly feasible to have a definition of research—the federal government has a number of very precise definitions of research. People in industry and other areas make great distinctions between types of research: engineering research, scientific research, etc. However, the line between engineering research and design is frequently a very tough line to draw. Jack Halpern, University of Chicago: The point that Dr. Manuel made about the top products in the chemical industry being substantially the same, or very similar to what they were 50 or more years ago, has an important corollary with respect to the changing role and value of research in a given sector or industry with time. What happens with respect to any given product with time is that it improves, or the processes by which it's made improve, and you reach a point at which further innovation with respect to that product or that process has diminishing returns. When a chemical costs $1.00 a pound, there's lots of room for realizing substantial returns if you can drop that cost to 50 cents. When it gets down to 5 cents a pound, even if you could cut that in half, you're only saving pennies. So the investment in developing further improvements declines. One consequence that we're seeing in the chemical industry is a shifting to non-chemical fields that are at an earlier stage of their technology. Monsanto is an example of this. Monsanto, which was one of the major chemical companies, went out of the chemical business in recent years as a result of this. Monsanto is now a biotechnology company. There are products that are virtually impossible to improve upon. A friend of mine in the pigment industry, which is a large industry, points out to me that there are certain pigments (certain colors, that is) that are so good that it's absurd to think of investing in further research leading to improvements of the kind that sustained much of the chemical industry in the last century and that still goes on with respect to some other pigments and products. This is a point that's particularly important for this group.
OCR for page 26
--> Don E. Kash: Let me underscore one tentative conclusion that we have drawn. With regard to many of these complex technologies, the organizational arrangements that make it possible to access and integrate knowledge and expertise in diverse areas are as important as research. The ability to take a 35-millimeter camera and convert it into an electrooptical-mechanical system and enhance it on a continuing basis is obviously dependent on research. But if you look back into the process, it is this capacity to do the systems integration, to put things together, that leads to commercial success. I must confess to you that I've now been doing this for several years, and one of my standard games is to go around and ask people in industry to tell me what systems integration is. The reply is: Systems integration is what we do when we put different things together. Judith C. Giordan, International Flavors and Fragrances: I would like to make two points. When you take a look at the S-curves, is it just a result of scientific and engineering research? Now more than ever before there's a third component to research, and that is market research. Market research played far less of a role 40 or 50 years ago than it does now. When one debates the definition of research, one has to take this element into account. The other point was Dr. Manuel's comment about semantics, with which I agree completely. We don't want to get lost in that morass. By asking anyone in the organization to use whatever definition they want for research, as well as asking anyone in the organization the value of said research, I suspect that you can obtain widely different views of research and its value. Don E. Kash: Well, I can assure you that what your answer is depends on whom you ask. However, I have talked to no company that includes market research in their research budget. That is, when I ask people what the contribution of research is, I've never had anyone identify market research. Of course, for most of the complex technologies that I discussed, market research is very important—it's a major input to the innovation process. But I am sure that it doesn't show up in DuPont's research budget. Roland Hirsch, U.S. Department of Energy: I would like to go back to an anecdote that I think illustrates the dilemma that is at the heart of the issue we're discussing here. Although a Toyota executive is said to have stated that Toyota really doesn't do research, I would submit that Toyota's Model T would probably look a lot like Ford's original Model T if it weren't for research in all sorts of areas, fundamental and applied, and that in fact Toyota is probably paying as much attention to research now being done in a variety of industries and academic laboratories as it ever has. Otherwise, it won't be around 10 years from now. Don E. Kash: Well, the first point I should make about that anecdote is to tell you that that comment was made about 10 years ago. Toyota, in fact, spends a lot of money under the category of research now. So there's been a real change in the company as it has prospered and come of age. Toyota believes, as best as I can tell, that research needs to be done today. But I again repeat my cautionary note: One of the biggest problems that people get into when they try to make this linkage between research and commercial product and process payoff is claiming too much for research. What happens, if you're not very careful, is what happened to DuPont in the 1960s (see Professor Hounshell's paper in this volume). If you're not conscious of the process whereby production, marketing, and all of these other factors are major inputs into the innovation process, you can get into deep trouble. People who have responsibility for research have—regularly and consistently over time—made claims that I simply do not believe can be supported. However, I must also make the point on which there is general agreement, and that is, I haven't
OCR for page 27
--> found anyone who doesn't think research is critical. In no small part, it's what you bet on to be able to respond to the unknown. But with regard to complex technologies, there are other inputs that are important that are not research inputs. They're primarily organizational. Michael P. Doyle, Research Corporation: I submit that the problem that you're dealing with may be one that is limited by the data available. The data that you have that could speak to the issue may not, in fact, be a reliable measure of what you're trying to measure. Let me give you one point in fact. A 1970 study was produced by the Illinois Institute of Technology. This was a study commissioned by the National Science Foundation, and its purpose was to trace the origin of critical technologies. They picked out four technologies, two of which were transistors and birth control pills, and traced the fundamental development of each technology over a period of 100 years or so, compiling a list of the individual discoveries involved. This study clearly shows that research was a critical feature of the development of these technologies—that certain elements could have not been developed were it not for discoveries that were made in that period of time, both in terms of fundamental scientific discovery and in terms of discoveries that brought those materials to the marketplace. I submit that, were we to do such a study now, we would arrive at similar conclusions. In fact, a few years ago the Board on Chemical Sciences and Technology traced this same path for a set of critical technologies (although in a less direct way) and concluded that R&D, either as fundamental developments or as the basis for bringing those technologies to the marketplace, was fundamental to the process. So I wonder, why is there such difficulty in identifying those activities as being critical? I think it is the limitations in the data available to provide those types of measures rather than the realistic system in which we all work. Don E. Kash: Well, I start with the assumption that you cannot find data that will convince everyone. One of the masons that you have great difficulty is because this whole discussion is embedded in ideology. You, of course, know that TRACES was study number two. It was preceded by HINDSIGHT. In fact, some people believe that TRACES was a direct result of HINDSIGHT. HINDSIGHT was an attempt by the Department of Defense to define the role that its fundamental research funding had made to weapon systems. It wasn't very successful. The National Science Foundation, which has some interest in defending fundamental research, got the Illinois Institute of Technology to do TRACES. TRACES found that there was a relationship. If you operate in this town, is there a message? I suggest to you that that is what's important here. I don't think there is any argument about the importance of research. What we're dealing with is a set of circumstances where people are trying to force the agencies (and very frequently, corporations) to demonstrate a cause-and-effect quantitative linkage between research and commercial products. It seems to me that Professor Hounshell's point is the key one. If you try to do that, you immediately get yourself into trouble, and you start arguing over the process. There's a more important issue as far as I'm concerned. Innovation involves important inputs beyond research. The most important one of these today, I believe, is organizational flexibility. With regard to complex technologies, I would, for example, eliminate all antitrust regulations, because they're barriers to information flow. I think that simple act would be an important contribution to the innovation process.
Representative terms from entire chapter: