2.0 Discovery and Invention
The starting point for making proposals that increase the rate of innovation
in firms is to discuss closely related social processes of discovery and invention.
In this section we will first define discovery and invention. Then, we will
examine the incentives that promote these activities. Finally, we will consider
the interactions between the two.
Definitions in this chapter will be indicated by "italics ". The first two definitions to consider are discovery and knowledge. A discovery is a new increment to knowledge 1 that is defined as understanding the behavior of observable natural phenomena as well as understanding the structure of logical relationships. These observable natural phenomena consist of all physical, biological, and social processes that can be directly or indirectly observed. Logical relationships, on the other hand, are described by the study of pure logic, mathematics, statistics and computer science.
Knowledge of behavior can be further partitioned into theoretical models and empirical relationships. Theorists create models to explain and forecast behavior and empiricists observe behavioral relationships. Their pursuit of knowledge varies from the most theoretical to the most practical, or alternatively, from deep to surface. For example, theorists create models which vary from intuitive qualitative models to formal mathematical models, whose predictive capacity depends on how well the behavior being modeled is understood. Similarly, empiricists make observations using methodology which varies from carefully controlled experiments to casual unstructured observations.
With definitions for discovery and knowledge, the relationship between these two concepts can be clarified. Some examples of discoveries are the creation of a new theory, the observation of a new behavioral relationship or the development of a new skill. Current knowledge is the sum of prior discoveries which are stored either in human memory or on a human record such as printed material. Thus, not all prior discoveries are part of current knowledge because some discoveries are forgotten. Examples of forgotten discoveries are skills associated with obsolete technology such as the techniques for constructing stone-age tools, which modern archaeologists are trying to recreate2.
The next definition to consider is an invention. An invention is a new manmade device or process. A new device which qualifies as an invention may take such forms as a new physical product, a new biological lifeform or a new piece of software. A process, on the other hand, is a chemical, physical, or biological chain of events that produces a product or service. To be patentable, an invention must meet a test of originality. But the fact that an invention may qualify for a patent, does not guarantee that the invention will be profitable to produce. Each year inventors create numerous inventions, of which only a small percent will be profitable to produce. Corporations, in fact, focus much inventive effort on making improvements to existing products and processes. These improvements will be considered minor inventions, regardless of whether or not they could be patented.
The incentive for discovery is fame. Scientists who make important discoveries or develop new theories become famous and simultaneously receive material rewards in the form or professorships and prizes. The pursuit of fame by researchers promotes the free flow of ideas as researchers compete to first present their work at conferences and in publications. Thus, new discovery is based on prior discovery and the rate of discovery is increased by the rapid dissemination of new results that anyone can use without charge.
In market societies, intellectual property in the form of patents, trade secrets, and copyrights have been developed to create better economic incentives for inventive and artistic endeavor. A patent gives an inventor exclusive use of an invention for a finite period of time. A trade secret enables a firm to sue for damages if an employee reveals economic secrets of the firm, and a copyright provides writers and musicians exclusive use of their creations for a finite period of time. Over time, innovations are needed to adjust these intellectual property rights to fit the changing needs of the political economy3. For example, in recent years copyright protection has been extended to software and integrated circuit masks.
These two incentive systems also have their weaknesses. Fame, for example, does not create an incentive system to finance basic research. And, while intellectual property creates strong incentives to invent, it creates a chasm between discovery and invention. Until inventors receive a patent on their invention, they are by necessity very secretive about any discoveries that they made in the process. Thus, their discoveries are not available to researchers in the field. We shall explore different aspects of this chasm throughout the chapter.
Once discovery and invention have been defined, the next step is to describe the two-way interactions between these activities. For example, discoveries frequently lead to inventions, but in many cases the full economic development of an invention requires major new discoveries. Also, invention can lead to a large increase in knowledge accumulation. Fortunately, because the knowledge required for invention can be compartmentalized a successful inventor has to understand only a portion of this knowledge accumulation. Finally, the difference in incentives between discovery and invention impedes the necessary interaction between discovery and invention.
To illuminate these relationships, let us start with the interactions within the process of discovery itself. An important component of discovery is the direct pursuit of basic knowledge by scientists and mathematicians. Science has become a specialized activity which advances through the interactions of specialists such as empiricists, theorists, mathematicians and engineers. Empiricists discover new phenomena, which stimulate theorists to explain with new models and theories. Theorists then use formal mathematical methods to deduce the implications of their models. Empiricists either confirm or reject these implications on the basis of experiments and hypothesis testing. In addition, engineers using new discoveries create new instruments such as new observation devices and computers, which in turn promote further empirical and theoretical studies. Similarly, mathematicians in discovering the structure of formal relationships, create new tools that theorists can use in constructing formal models of behavior. Through the interactions of specialists, then, new formal models for explaining and predicting behavior are created4 .
The next step is to describe the interactions between discovery and invention. Discovery frequently creates opportunities for invention; however, the development of an invention generally requires further discovery specific to that invention. Inventors can rarely invent a fundamentally new product as a pure exercise in engineering. That is, they can rarely design a product purely from known principles, since theory rarely provides answers to all the design questions that are likely to arise during the process of invention. For example, in designing the airplane, the Wright brothers had to conduct wind tunnel experiments to design an efficient airfoil. They could not simply apply the nascent aerodynamic theory that existed at that time.
The relationship between theoretical discovery and invention is two way. An example of a theoretical advance that led to invention is the case of nuclear power. With Einstein's development of a theory to explain the relationship between mass and energy, physicists and engineers made numerous applied discoveries first to build the atomic bomb and then to create atomic power. In contrast, the advance of the economic usefulness of an invention from satisfying the legal definition for patentability to widespread economic application frequently requires a major investment in theoretical discovery5 . The development of commercial aviation and military airforces, for instance, stimulated the advance of theoretical discoveries in aerodynamics. In like fashion, the current efforts to develop market applications of superconductivity will undoubtedly generate major new theoretical models of superconductivity.
The rate that applied discovery needed to make an invention a commercial success occurs depends on the methodology of invention. The simplest methodology for invention is trial and error experimentation by an inventor without any formal training. All other things being equal, a more efficient methodology is trial and error experimentation by a inventor with training in science and engineering because such an inventor can eliminate blind alleys from theoretical considerations. A scientifically trained inventor can also improve the speed of applied discovery by using good scientific and statistical methodology. Finally, if a theory is well understood it can be incorporated into a computer assisted engineering program so that the inventor can quickly, inexpensively analyze alternatives without having to build prototypes and perform tests.
In addition, the organization of invention and discovery affects the rate of advance. Separating invention from production into research and development laboratories means that inventors can devote all their time to applied research needed to create commercially successful inventions. Nevertheless, in some cases the separation of research and development activities from production can create a problem in transferring ideas from the research and development laboratory to the market. A classic example here is Xerox who developed the graphical user interface that Apple brought to the marketplace.
Next, let us consider how the chasm between university discovery and private invention impedes the interaction between these two activities. In the pursuit of fame university researchers have incentives to present their ideas to other researchers, but not to potential inventors. Thus, there is a need to create incentives to promote the transfer of ideas from pure research to inventive activity. Also, the research that takes place in private research and development laboratories is frequently of great interest to university researchers, but because of the desire to reap the benefits of intellectual property, firms prohibit its release.
Another problem is that university researchers are generally interested in basic research to promote their academic careers and managers of researchers in the firm are generally interested in applied research which promises to create products in a reasonable time horizon. There is no incentive to perform applied research that might produce useful inventions in the long run. Such research is not prestigious enough for university research and is too risky for research in firms.
In spite of the these problems, discovery to develop the inventions associated with a new technology and its applications frequently leads to a significant accumulation of knowledge in many fields. Consider, for instance, the inventions associated with integrated circuits, computers and software. Crowding more and more components into an integrated circuit advances knowledge of materials at the atomic level, and designing integrated circuits advances knowledge of silicone compilers, software used in the design process. Reducing the production cost of computers requires discoveries to create new production techniques such as surface mount technology. One aspect of software discovery are advances in artificial intelligence to make programmers more efficient. The development of new applications software is frequently based on discoveries in the respective applications discipline. For example, the development of the mathematics of linear programming led to the development of software for applications.
If inventors in an industry had to understand all the knowledge accumulated in the development of technology in that industry, the process of invention would be severely hampered by the knowledge requirements. But, because technology generally has a modular structure, the process of invention requires specialized knowledge and discovery. Consider, for instance, the new central processor unit, the Pentium Pro chip, created by Intel. An inventor using this chip to invent a new computer workstation needs to know only the operating characteristics of the chip and not the knowledge needed to design the chip, crowd the components onto the chip, or obtain economic yield rates. Similarly, the developer of the operating system needs only to know the operating characteristics of the new computer rather than information about how it is designed. Finally, the developers of applications software knowledge of the computer are limited to knowing the software language and the operating system. This modular specialization of technology greatly facilitates invention by limiting the knowledge requirements for invention.
Within the modular structure of technology the depth of knowledge required for effective invention varies greatly. An inventor has an incentive to understand theory to the extent that this knowledge reduces the applied empirical research needed to perfect an invention. Some inventive activity takes place by inventors who understand how the modules work and creatively combine them. One example is the creation of the Apple II personal computer by Steve Wozniak who used a microprocessor designed for intelligent appliances as one of his modules6 . On the other hand, it is doubtful that a quantum effects transistor could be developed by an inventor without a considerable formal study in quantum mechanics7.