In an attempt to describe the (physical) properties of the adjacent possible the closest parallel would be a vacuum. A vacuum, a space where there is little or no matter (“vacuum”). On earth, a vacuum is temporary in nature and quickly filled with matter once possible. The same is true for the space, which Stuart Kauffmann defined as the adjacent possible. It is uncharted space on the edges of current knowledge with nothing in it, yet the place where the new ideas emerge quickly. Within this sphere, innovation happens fast and so manifold that it is quite common to see multiple inventions at the same time emerging independently. This remarkable pattern was first investigated by the researchers William F. Ogburn and Dorothy Thomas in their 1922 paper Are Inventions Inevitable (Ogburn and Thomas 1922). Their paper listed 148 major inventions and discoveries that were made independently by two or more two or more groups at the same time. A similar study by Merton in 1961 led him to conclude that “the pattern of independent multiple discoveries in science is in principle the dominant pattern, rather than a subsidiary one” (Merton 1961 p.470). Today, there is even continuously expanding Wikipedia list of multiple inventions (“list of multiple discoveries”). Examples include the atomic bomb, the jet engine, cosmic background radiation, and quantum cryptography—just to name a few.
Since the adjacent possible defines a narrow space of potential first-order combinations of exiting knowledge, it is not a coincident that new ideas emerge within these boundaries. This phase sphere is the breading ground of new ideas. It is where innovation almost magically emerges in great number and often do so at the same time yet independent from each other. Regarding the emergence of new ideas alone, however, does miss the important constraints of innovation. Considering the definition of innovation, developed in the first passage, there are three distinctive components of an innovation: the idea, the realization, and the exploitation. Each component is a prerequisite for innovation success, so the breakthrough of an innovation can be stalled or spurred by either one. Take the invention of the light bulb as a classic example. In 1840, British scientist Warren de la Rue developed an efficiently designed light bulb using a coiled platinum filament in place of copper, but the high cost of platinum kept the bulb from becoming a commercial success. Around 40 years later, American inventor Thomas Edison developed a cheaper and more durable version and is ever since credited as the inventor of the light bulb. Consequently, the viability of a technology as well as the successful diffusion of a technology within a society is an equally important factor for successful innovation than the idea creation. Consequently, the true breading ground of innovation, which will further be referred to as the opportunity vacuum (OV) has three dimensions which can be thought of as layers stacked on each other’s. The three dimensions of the OV are the following:
The Adjacent Possible (Technology)
The Adjacent Viable (Economy)
The Adjacent Acceptable (Society).
It is important to notice that there is no inherent order within these dimensions. It could be argued that without technical feasibility there is no such thing as viability or social acceptance. Thus, dimension 1, the adjacent possible is often considered the breeding ground of innovation. Successful innovations, however, only occur if there is an intersection among the boundaries of possibility within all three dimensions. This overlap can be initiated by changes in every single dimension. If, for instance, sequencing an entire personal DNA in order to develop personalized medication is technically possible and it is socially acceptable to get personal DNA sequenced, then innovation in this area is a function of the economic viability of DNA sequencing. Likewise, social developments might spur research in certain technical areas, which lead to an expanded adjacent feasible or adjacent viable as a result.
Dimension 1: adjacent possible (technology)
In our connected world, inventors stand on the shoulders of giants. The accelerated development of technology in the past century is a result of the increasing body of knowledge, continuously amassed by researchers around the world, and increasingly made available to others by the use of modern communication tools. As a well-known example take the development of the transistor in 1947, which consequently enabled the development of the integrated circuit in 1958, which in turn led to development of what we now know as modern computing. The invention of computers enabled software as a new field of further development and created the founding basis for the internet. In the internet domain, the invention of html and later Java led to the development of flash video with enabled Chad Hurley, Steve Chen, and Jawed Karim to start their video sharing platform YouTube in 2005. From a backward perspective, every single step of this development was crucial for the invention of YouTube. However, other paths may have led to different outcomes of what we know now as video sharing.
The basic mechanic, however, is quite simple. Each new technology opens the adjacent possible for new technologies based on this and all other available technologies. Each invention is based on preceding innovations which are in turn build on their predecessors. The adjacent possible can thus be described as a thin layer at the edge of current knowledge, which attracts new inventions and thus grows the body of exiting knowledge.
Figure 1 visualizes the basic conception of the adjacent possible form a technology perspective. Where A1 represents a first innovation, e.g. the transistor, which consequently enables various technologies B1 to Bn. These innovations, in turn, enable subsequently innovations C1 to Cn. The adjacent possible represents the current edge of this process, which is comprised of all possible first-order combinations of existing technology A1 to En.
Dimension 2: adjacent viable (economy)
If a technology or idea is within the adjacent possible, it can theoretically be realized. Whether it can be realized in reality depends on the economic viability of its realization concept. While it is currently theoretically possible to get your entire personal DNA sequenced in order to receive personalized medication, it is currently not economical feasible to do so. The sequencing of the first human genome costed in excess of US$3 billion and took 13 years to complete (Hayden 2014). However, due to advances in the field of genomics over the past quarter-century, substantial reductions in the cost of genome sequencing were achieved. The current costs for decoding the human genome are expected to shrink below US$1000 soon (Hayden 2014). Therefore, personal genome-based medication—as well as other related innovations—is gradually becoming economically feasible. As a result, we will see a lot of innovation based on sequenced personal DNA in the next years. We can describe the area of innovation development based on gradually emerging economic feasibility as a related form of the adjacent possible. Rather than becoming technically feasible, however, the trigger for innovation development here is the viability of its realization. The adjacent feasible defines an area of expected cost reduction for realization in a defined (future) time frame. It, again, creates a sphere of opportunity, attracting the rapid development of innovations in this field.
Extensive empirical research exists on the cost reduction of technology over time (Nagy et al. 2013). The most well known is certainly Moore’s law (1965), which postulates that the number of transistors on a chip should double in each technology generation resulting in a linear cost decrease over time. Because of the accuracy with which Moore’s law has predicted the cost reduction in semiconductors, similar studies have since been conducted on related industries from online commerce to genetic modifications (Mack 2011; Schaller 1997).
The underlying mechanic of this model is that the adjacent viable defines an area of potential innovation, based on an expected increase in viability within a defined time frame. The expected increase in viability can be of one technology—as in the example of genomic sequencing—but could also include multiple technologies. Figure 2 explains the basic mechanic of this model.
As an example, take the cost of digital storage and bandwidth for the realization potential of YouTube. Let A1 be the cost of data storage in 1998 at US$100 per gigabyte, which gradually decreased to A5 in 2005 at US$1. The cost of upload/download bandwidth decreased similarly from B1 in 1998 at US$1200 per gigabit to B5 in 2005 at US$75 per gigabit. A5 and B5 together enabled point F—The possibility to create an online video sharing platform like YouTube. The important point is that it was actually not viable, yet, to provide such a service in 2005. The income of advertising—even at today’s income rates—would not have been enough to compensate storage and especially bandwidth costs. But since both costs were expected to further decline, it created an opportunity vacuum for those who foresaw the future development of viability in this technological context. It is also important to note that innovations outside this trajectory are doomed to fail. If, for instance, YouTube would have tried to build the same service 5 years earlier, it would have been outright impossible due to the high bandwidth and storage costs, even if it would have been technically feasible.
Dimension 3: adjacent acceptable (society)
Changing people’s customs is an even more delicate responsibility than surgery in many cases (Rogers 2003, p.436).
Human behavior plays an important, if not the most important, role in the innovation process. Classical economic theory argues for consumers being “homo oeconomicus”, purely motivated by rational considerations. It has become well known, however, that consumers are not always rational, objective, and utility-maximizing. Instead, they tend to base their decisions on other, more subjective, beliefs about a new technology (Fishbein and Ajzen 2010). Different areas of technological and service advancements have shown that reasonable innovations fail or take longer than expected to reach wide-spread acceptance, despite their proven usefulness (Rogers 2003; Story et al. 2011). This paradox is generally explained by consumer resistance to change learned behavior (Planing 2015).
Being one of the forefathers of sociology and social psychology, French lawyer Gabriel Tarde was the first to observe and analyze how new ideas flourished within French society at around 1900. In his influential book Laws of Imitation, Tarde (1903) dealt with the central question of compatibility: that is, the goodness of fit between the attributes of a diffusing item and the social and psychological attributes of the potential adopter. In his classic book Diffusion of Innovations, Everett Rogers (1962) developed a common framework for the social acceptance of the new ideas and concepts. Since then, the scope of innovation acceptance research has broadened as more and more disciplines became involved. Early studies mainly focused on rural sociology, investigating the spread of the new farming techniques, but soon scholarly interest tailed off somewhat to other disciplines such as communication, public health, and marketing. Since around 1990, the number of diffusion studies strongly increased, with many focusing on the rapid spread of the new communication technologies like the internet and mobile applications (Rogers 2003). The common result of these studies is that the diffusion process develops because potential customers do not adopt an innovation directly after it becomes available to them, but only with a—varying—time gap. Plotting the adoption of an innovation over time on a frequency basis will result in a normal, bell-shaped curve or—if the numbers of adopters are cumulated over time—in an S-Shaped curve of adoption (Rogers 2003). Figure 3 gives an overview of Roger’s diffusion process.
By using some products repeatedly over a long period of time, consumers form habits and routines. In general, they aim to preserve these habits and strive for consistency and status quo rather than to continuously search for and embrace new behaviors (Bamberg et al. 2003; Bagozzi and Phillips 1982). If a new idea is not compatible with existing behavior, the perceived relative advantage of the new idea must be large enough to offset the perceived complexity of adopting to a new behavior. A theoretical model, frequently used for evaluating this relationship, is the Technology Acceptance Model (TAM) developed by Davis et al. (1989) The TAM is a tool for predicting and explaining user acceptance of an innovation and postulates that the acceptance decision can be reduced to two factors: the perceived usefulness (PU) and perceived ease of use (PEU). In essence, perceived ease-of-use (PEU) reduces uncertainty about the cause-effect relationship involved in the innovation’s capacity to solve an individual’s problem, while perceived usefulness (PU) describes the anticipated positive effect of using this technology. Bagozzi (2007, p.244) stated that more than 700 empirical applications of their original paper proved its validity for the adoption of different technologies. In particular, empirical research within this framework showed that both factors are heavily influence by the compatibility with existing beliefs and prior experience with comparable technologies (Karahanna et al. 2006, p.787).
Based on a meta-study of 1500 diffusion studies, Rogers (2003) developed a comparable framework for innovation adoption. The study showed that most of the variance in the rate of adoption of innovations, from 49 to 87%, is explained by only five attribute categories: (1) relative advantage, (2) compatibility, (3) complexity, (4) trialability, and (5) observability (Rogers 2003, p.222). Changing from a combustion car to an electric car, for instance, requires a behavior change. The perceived advantages of electric driving need to offset the perceived disadvantages, i.e. perceived complexity and incompatibility with current driving and fueling patterns. If the observability is high, i.e. individuals can see others using electric cars and if it is easy to try the technology on a limited basis, the chances of adoption are higher.
We can conclude that consumer habits change over time according to an S-shaped curve of adoption and new technologies are the driver of change. It is important to note that the adoption time of new technologies has significantly decreased in the recent decades (Hohberger 2016; van den Bulte 2000). Thus, the stretch of S-shaped curves, on average, becomes shorter over time. While there is a multitude of factors that influence the adoption speed of new technologies, the compatibility with existing behavior and beliefs is the single most important driver for the acceptance of innovation (Bamberg et al. 2003). Along the lines of the adjacent possible, socially accepted behavior builds on the stack of currently diffused ideas and concepts. If compatibility with existing behavioral patterns is too small, it is almost impossible to initiate social change immediately. Rather, gradual steps need to be taken. Using a fully autonomous vehicle, for instance, is out of the accepted behavior range for most people. Yet, by the increasing use of driver-assistance systems, individuals will get more used to the idea of handing over driving control to a computer system. The adjacent acceptable thus represents a small area on the current edges of socially accepted behavior, which currently only innovators embrace but soon will reach the early majority of technology adopters. Figure 4 shows the resulting conceptual model of the adjacent acceptable.
As an example for adjacent acceptable consider the diffusion of video streaming. Watching movies online was a behavior that was unheard of in 2005 when YouTube started. After 10 years of watching short video clips on YouTube, it became more natural to most people to use the internet as a channel for TV movies and series. Watching movies online reached the early majority on the S-shaped adoption curve. Other required behaviors became socially acceptable in parallel, most importantly paying online using PayPal and credit cards. This opened up the adjacent acceptable for offers like Netflix or Amazon Prime Video. In the visualized model, take the adoption of personal computers as A1. The full adoption of this technology was a prerequisite for the adoption of the internet access B1 and later high-speed internet B2. This was the prerequisite for wide-spread adoption of video streaming C1, online payment C2, and online Commerce C3, which brought payed, video-streaming (F) into the adjacent acceptable.
The opportunity vacuum
We have seen that the occurrence of innovations, despite its randomness, follows certain patterns and generally occurs within the defined boundaries of the adjacent possible, the adjacent viable and the adjacent acceptable. It is only, however, by bringing all three dimensions together, that we can really explain the origin of innovations. In retrospective, the origin of every innovation can be pinpointed to a moment when all three dimensions achieved an intersecting area. A moment, when it was technically feasible to realize an idea, financially viable to do so and when the early majority of the society was ready to adopt the idea. Figure 5 shows the combined three dimensions, with the intersecting area, which we refer to as the opportunity vacuum (OV).
The intersecting area among all three dimensions at a given point of time yields a flow of emergent possibilities which enable innovation to emerge. While the area itself can neither be measured nor described in detail, the emergent innovations can be traced with a relatively high accuracy in retrospective. The occurrence of a new idea within the adjacent possible can be attributed to its preceding ideas it was built upon. The economic resources needed for the realization of an idea within the adjacent viable can be measured over time as they approached a viable level for realization and exploitation. The adoption curve of technologies within the adjacent acceptable, finally, can be traced with increasing accuracy, mainly due to new communication tools and connectivity of products. In hindsight, we are thus capable of explaining how an innovation has evolved throughout this model with relatively high precision.
Application of the OV framework to predict future economic novelty
The OV framework is a valid method for explaining the origin of innovations in retrospective. Whether the model could also be used to predict future occurrence of innovation by defining the area of the OV and, in particular, its boundaries, however, is a more complex question. Considering the first dimension, the adjacent possible, Felin et al. (2014) compare the problem to the myriad functionalities and uses of any technological object, which cannot be prestated or captured. A simple technology, such as an electrical engine, can theoretically be applied to myriad uses which are both indefinite and unadorable, since most would not provide any value. A prediction of where the development of an electric engine will lead us is thus not feasible, since it is simply not possible, a priori, to list the number of uses of an electric engine. As a result, no full account or set of algorithms can be given about all possible, actions, uses, and functions of an electric engine. This algorithmic incapacity to compute developments with the adjacent possible does not mean that the adjacent possible defies any explanation. Other than in Newtonian and Laplacian physics, we cannot determine and predict motion or direction in the first dimensions of this model. Nonetheless, the evolution of new technologies is not fully random. Its randomness is canalized within the sphere of the adjacent possible. The set of adjacent possible directions is extremely large but not infinite. While we cannot predict the exact development of the future technically feasible, we can predict the area of where future technology will become feasible.
In order to apply the OV framework to predict future innovation, we therefore use an example, which currently enjoys high attention by the scientific as well as practitioner community: autonomous, electrical multicopters aimed at passenger transportation. Experts argue that these systems, often called passenger drones, will enable individual sky transport in the near future (see Lidynia, Philipsen, and Ziefle 2017). While we cannot predict how the development of passenger drones will exactly happen, we can use the model to derive an idea in which direction the development will go and where an opportunity for innovation will occur. Let us consider at the adjacent possible first. First prototypes based on either electric drones with multiple rotors were already presented to the public. However, technically, there are some major restrictions. The development of a passenger drone would require certain technology breakthroughs; first, more efficient battery technology. In order to increase the currently around 20-min flight time to about 2 h without adding weight (which would reduce flight time again), would require battery efficiency to increase drastically. Experts predict that the specific energy of current conventional lithiom-ion batteries, which is around 150 Wh/kg, needs to be brought to around 1.800 Wh/kg (National Academies of Sciences, Engineering, and Medicine 2016). At the same time, cognitive computing needs to improve as well. In order to enable fully autonomous flight, these systems need to combine visual data with other sensors, such as RADAR and LIDAR. This will be a prerequisite for navigating safely and efficiently through the air, based on real-time sensor data (Valavanis and Vachtsevanos 2014). While we cannot predict the boundaries of the adjacent possible dimension exactly, we can nonetheless conclude that currently at least two major building blocks are missing: more efficient battery technology and improved cognitive computing systems, who enable fully autonomous flying.
In the next step, the adjacent viable is considered. While it is already possible to buy a private helicopter, significant cost reduction will be a prerequisite to make individual sky transport a mass-market phenomenon. Passenger drones provide the opportunity to make that fundamental shift. The main factor in this regard will be the cost of electric energy storage, or in other words, the battery prices. Current battery costs in electric cars are ~US$150 to ~US$200 per kilowatt-hour, making them the most cost efficient battery products, well below the industry average pack costs of ~US$350 per kilowatt-hour. In order to make flying passenger drones a viable business, we need to see these prices drop to somewhere around US$20 per kilowatt-hour (Henbest et al. 2015). Based on the historical cost developments, a prediction could be derived, when this point could be reached.
Finally, the adjacent acceptable dimension is considered. This new form of mobility is far off anything consumers have experienced in the past and will require a substantial behavior change. As with any other new technologies, we will see an S-Shaped curve of technology diffusion, with a slow incline in the first years of availability. At first it will be only the innovators and early adopters, who are ready to accept the non-perfect product in terms of range, conform, and maybe even safety. The early majority (the part of the mass market more open to innovation) is pragmatic, in the sense that they will wait for the technology to proof its benefits and safety. Therefore, we can expect that once technological possibility and economic viability are reached, it will take several years of proof of concept before the early majority of customers will be ready to go on-board. Once this happens, however, diffusion theory predicts an exponential acceptance rate increase, which would mean that the technology could potentially conquer the world in only a couple of years.
There is no opportunity vacuum for passenger drones, yet. However, we can see developments in all three dimensions that could potentially lead to the rise of an OV in the next couple of years. For a technology expert, or anyone interested in either starting a business or investing in a business in this area, it is therefore wisely to monitor closely the developments in each of the three dimensions. From a scientific point of view, the framework enables to make a prediction about the rise of the OV, which could in a longitudinal study be checked against reality.
In sum, the OV is aimed at describing the origin of innovations in retrospective. It does not provide an exact solution for the prediction of future economic novelty, since there is no algorithmic way to specify the yet “empty” set of possibilities that are adjacent to the existing phase spaces. However, the framing approach along the three dimensions of the OV provides a reliable method to evaluate known paths of innovation. Once the path is known, e.g. the technology is within the adjacent possible of dimension 1, the OV framework can be used to predict where future entrepreneurial activity is likely to occur.