1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
MAKING SENSE OF TECHNOLOGY TRENDS IN THE IT LANDSCAPE:
A DESIGN SCIENCE APPROACH FOR DEVELOPING CONSTRUCTS
AND METHODOLOGIES IN IT ECOSYSTEMS ANALYSIS
Gediminas Adomavicius, Jesse C. Bockstedt, Alok Gupta
Department of Information and Decision Sciences, and MIS Research Center
Carlson School of Management, University of Minnesota
Minneapolis, MN 55455
{gadomavicius; jbockstedt; agupta}@csom.umn.edu
Phone: (612) 624-8030
Robert J. Kauffman
Center for Advancing Business through Information Technology (CABIT)
W. P. Carey School of Business, Arizona State University
Tempe, AZ 85251
rkauffman@asu.edu
Phone: (480) 965-2613
Last revised: September 21, 2007
Currently Under Review at MIS Quarterly (third round)
i
3
MAKING SENSE OF TECHNOLOGY TRENDS IN THE IT LANDSCAPE:
A DESIGN SCIENCE APPROACH FOR DEVELOPING CONSTRUCTS
AND METHODOLOGIES IN IT ECOSYSTEMS ANALYSIS
4
5
6
7
8
9
10
11
12
13
14
Acknowledgments. The authors would like to thank the Digital Technology Center and the Carlson
School of Management, University of Minnesota, for providing joint financial support of this research.
Rob Kauffman thanks the MIS Research Center, the Center for Advancing Business through Information
Technologies, and the W. P. Carey Chair for partial support. The authors would also like to thank the
organizers and participants of the “Innovation and Innovation Management” mini-track at the 2006 Hawaii International Conference on System Sciences for valuable comments and discussion on an early version of this manuscript. We also thank IT experts who willingly provided their valuable time for evaluating constructs and methods developed in this research. The special issue guest co-editors, Sal March and
Veda Storey, were instrumental in helping us to develop this research. We also are indebted to the
anonymous associate editor and three reviewers, who offered useful assistance and guidance for improving the quality of our manuscript.
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
Author Biographies
1
2
Gediminas Adomavicius (gadomavicius@csom.umn.edu) is currently a Digital Technology Center Assistant Professor of Information and Decision Sciences at the Carlson School of Management, University
of Minnesota. He received his Ph.D. in Computer Science from New York University. His research has
been published in several leading Information Systems and Computer Science journals, including Information Systems Research, ACM Transactions on Information Systems, IEEE Transactions on Knowledge
and Data Engineering, INFORMS Journal on Computing, and Data Mining and Knowledge Discovery.
He received the NSF CAREER award in 2006 for his research on personalization technologies.
Jesse C. Bockstedt (jbockstedt@csom.umn.edu) is a doctoral candidate in the Information and Decision
Sciences Department at the Carlson School of Management, University of Minnesota. His research interests include digital entertainment goods, technology evolution, and the impact of new ITs on markets,
products, and consumer experiences. He has published in Communications of the ACM, International
Journal of Electronic Commerce and Information Technology and Management. He has presented at the
INFORMS Conference on Information Systems and Technology and the Hawaii International Conference
on System Sciences, where his work was nominated for the best paper award.
Alok Gupta (agupta@csom.umn.edu) is the department chair and Carlson School Professor of Information and Decision Sciences at the University of Minnesota. He received his Ph.D. in MSIS from UT Austin in 1996. His research has been published in various information systems, economics, and computer
science journals such as Management Science, ISR, MISQ, CACM, JMIS, Decision Sciences, Journal of
Economic Dynamics and Control, Computational Economics, DSS, IJEC, IEEE Internet Computing, and
JOC. He received NSF CAREER award in 2001 for his research on online auctions. He serves on the
editorial boards of ISR, JMIS and DSS and various national an international research centers.
Robert J. Kauffman (rkauffman@asu.edu) is the W.P. Carey Chair in Information Systems at the W.P.
Carey School of Business, Arizona State University, and past Director of MIS Research Center, and Professor, Information and Decision Sciences at the Carlson School of Management of the University of
Minnesota. His research interests span the economics of IS, technology adoption, competitive strategy
and technology, IT value, strategic pricing and technology, supply chain management, and theory development and empirical methods for IS research. He has won numerous research awards, including the
2006 “best research award” for modeling and strategic decision making research on embedded standards
in technology-based products from the IEEE International Society for Engineering Management. His recent article on value-at-risk in IT services was the runner-up for best paper award at the 2007 Hawaii International Conference on Systems Science. His publications appear in Management Science, Organization Science, ISR, JMIS, MISQ, Decision Sciences, and other leading journals.
ii
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
MAKING SENSE OF TECHNOLOGY TRENDS IN THE IT LANDSCAPE:
A DESIGN SCIENCE APPROACH FOR DEVELOPING CONSTRUCTS
AND METHODOLOGIES IN IT ECOSYSTEMS ANALYSIS
ABSTRACT
An important problem for firms making IT investment and development decisions is analyzing
the complex IT landscape to understand trends in IT development. The sheer number of available technologies and the complex set of relationships among them make IT landscape analysis
extremely challenging for innovators, senior managers, and policymakers. Most IT consuming
firms rely on third parties and suppliers for strategic recommendations on IT investments which
can lead to biased and generic advice. Based on the design science research paradigm, this article defines a new set of constructs and methodologies by conceptualizing an ecosystem model of
technology evolution to help practitioners address this important business problem. The objective of these artifacts are: (1) to provide a formal problem representation for the analysis of technology development trends in the IT landscape, and (2) to suggest reusable structures which ensure that the complex technological environment is considered in the analysis of technology
trends. We adopt a process theory perspective and use a combination of visual mapping and
quantification strategies to develop a new empirical method to identify patterns of technology
evolution and a state diagram-based methodology to represent evolutionary transitions over time.
We illustrate the qualitative application of our proposed constructs and methodologies using the
example of digital music technologies. We also develop an associated quantitative empirical
methodology, which is illustrated with the example of wireless networking technologies. We
evaluate the utility of our proposed constructs by conducting in-depth interviews with several IT
industry experts and demonstrating the complementarities between our proposed approach and
existing techniques for technology forecasting.
Keywords: Design science, ecosystem model, evolutionary patterns, IT landscape analysis,
management of technology, paths of influence, technology ecosystem, technology evolution.
iii
MAKING SENSE OF TECHNOLOGY TRENDS IN THE IT LANDSCAPE:
A DESIGN SCIENCE APPROACH FOR DEVELOPING CONSTRUCTS
AND METHODOLOGIES IN IT ECOSYSTEMS ANALYSIS
1
2
3
4
1. INTRODUCTION
5
6
7
The landscape of information technology (IT) is in a constant state of change. There are an
8
overwhelming number of technologies available for use by organizations and that set continu-
9
ously grows at a steady pace. Additionally, the economic scope of IT investments is typically
10
very large, which adds significant pressure in the IT investment and development decision-
11
making process. Most IT managers indicate that it is difficult, if not impossible, to accurately
12
forecast advances and trends in IT. Nevertheless, senior managers need to understand the nature
13
of technological change and be able to accurately interpret the IT landscape 1 to position their
14
firms’ high-value technology investments and to achieve success with emerging market opportu-
15
nities.
16
Through interviews with IT industry experts we have learned much about how firms analyze
17
the IT landscape. We have found that this analysis and the IT strategic decision-making process
18
itself are often outsourced to third parties.
19
20
21
“It's kind of a funny thing. In IT it's okay to outsource your future decision-making, but with the business
you would never do that...[Companies] outsource their IT decision making just because it's so complicated.” -- VP of Global IT Infrastructure at a Fortune 500 company.
22
The difficulty of making these decisions typically requires skills and expertise beyond the capa-
23
bilities of most firms. As a result, many firms rely on reports produced by consulting companies
24
such as Gartner, Forrester, and IDC, and advice provided by their existing IT partners and sup-
1
Throughout the paper we use the terms IT landscape and IT ecosystem quite often. IT landscape is commonly
used by practitioners to describe the overall IT environment. IT ecosystem is a term that we operationally define as
the IT landscape centered around a specific set of technologies in a specific context that is the subject of analysis
using the methods and artifacts that we will shortly propose.
1
1
pliers. We recognize that such input can be helpful in the decision-making process; however,
2
our in-depth discussions with experts, whom we interviewed, identified two key concerns. First,
3
advice provided by existing partners and suppliers is often biased; existing suppliers have an in-
4
centive to encourage firms to continue their investment. Second, reports produced by third par-
5
ties are usually too general and often lack formal analysis of the IT landscape, relying primarily
6
on expert opinion and simple extrapolations to make recommendations.
7
8
9
10
“The main challenge is the dynamic nature of the whole rapidly changing IT environment...what we need
are more formal frameworks and tools to help see more clearly the current and potential future technology
landscapes. There is definitely room for improvement in developing these types of tools for managers.” -Senior Academic Researcher and IT Industry Consultant
11
To address these problems, we propose a new theory-based conceptual approach, a set of
12
new constructs, and a novel methodology for formally analyzing the IT landscape and identify-
13
ing trends in IT evolution. Adomavicius et al. (2007) proposed a new ecosystem model of tech-
14
nology evolution for understanding the dynamic and complex nature of technological evolution.
15
Building on this model, we follow a technological determinism viewpoint (Smith and Marx
16
1994) and use a process theory approach (Mohr 1982, Langley 1999) to develop a new set of
17
problem representation constructs and a novel methodology for identifying patterns of technol-
18
ogy evolution in the IT landscape. We incorporate the temporal aspect of technology evolution
19
and visually represent transitions between patterns of technology evolution over time. The goal
20
of our new approach and methodology is to complement existing techniques and provide firms
21
that make IT investment and development decisions with a more formal technique for analyzing
22
specific IT ecosystems and the interdependent relationships among the technologies they contain.
23
In developing and evaluating this new set of constructs and methodology, we adhere to the
24
principles of design science research, as outlined by Hevner et al. (2004) and March and Smith
25
(1995). Design science involves the construction of new artifacts to solve important organiza-
2
1
tional IT problems. Hevner et al. (2004) note that the definition of the IT artifact, as it applies to
2
design science, is both broader and narrower than the definitions outlined in prior literature.
3
Specifically, they include constructs, models, and methods in their definition, but do not include
4
people or elements of organizations. We present a new set of constructs, a model for represent-
5
ing relationships between ITs, and a new methodology for identifying and representing patterns
6
of technology evolution. This set of artifacts constitutes a new process for evaluating the IT
7
landscape, which is aimed to address the key business problem described earlier. This research
8
aims to explain the structure of relationships among ITs in a technology ecosystem, and to help
9
practitioners analyze the changes that occur in this structure over time. Whinston and Geng
10
(2004) argue that research that clearly impacts the IT artifact is important for furthering the IS
11
research field. As noted by Benbasat and Zmud (2003), the IS discipline includes methodologi-
12
cal practices and capabilities involved in the planning, construction, and implementation of IT
13
artifacts. This research follows in that vein by providing a new set of tools for analysts involved
14
in the IT investment and development decision-making process.
15
An important aspect of design science is the evaluation of the proposed artifacts, i.e., the util-
16
ity of the proposed artifacts must be demonstrated. To perform this evaluation we conducted in-
17
depth semi-structured interviews with a set of knowledgeable IT industry experts. These inter-
18
views were conducted to provide: (1) further understanding and motivation for the business and
19
organizational IT problem we address; (2) an evaluation of the utility of our proposed approach
20
in real-world settings; and (3) suggestions for improvements and future work. We constructed
21
structured interviews for four groups of IT industry experts to provide robustness to our utility
22
evaluation and capture differences of opinion. These groups were: IT industry senior managers
23
and executives, IT industry consultants, IT industry research staff and analysts, and senior aca-
3
1
demic researchers with expertise in the IT industry. We further demonstrated the utility of our
2
proposed approach by drawing insights from a qualitative example and quantitative empirical
3
analysis on two real-world IT ecosystems. We also provide in-depth discussion on the comple-
4
mentarities between our proposed approach and existing techniques that firms employ for evalu-
5
ating the IT landscape.
6
The remainder of the article proceeds as follows. Section 2 briefly reviews the foundations
7
of the research, and discusses sense-making from process data and the ecosystem model of tech-
8
nology evolution. Section 3 defines our analytical constructs and outlines the new problem rep-
9
resentation structure for exhibiting relationships among ITs within an ecosystem. Section 4 pro-
10
vides some examples of patterns of technology evolution, based on the constructs and model pre-
11
sented in Section 3. In Section 5, we illustrate the application of the constructs with a qualitative
12
example on the digital music industry and present a state-diagram-based visualization approach.
13
Section 6 presents a quantitative application of our constructs using a new empirical method for
14
identifying temporal changes in an IT ecosystem. The empirical method is used to analyze a
15
dataset containing over 3,000 new product certifications by the Wi-Fi Alliance (www.wi-fi.org).
16
Section 7 provides an evaluation of the utility of our proposed approach using in-depth inter-
17
views with IT industry experts. It also offers a detailed discussion of the complementarities be-
18
tween our approach and existing techniques that are used by firms and their consultants for tech-
19
nology forecasting. Section 8 provides the summary of main contributions as well as discusses
20
the limitations of our proposed approach and opportunities for future work.
21
2. CONCEPTUAL FOUNDATIONS
22
Decision-making and justification for IT investments is of strategic importance for modern firms
23
and can be difficult for even the most-seasoned and knowledgeable managers in the presence of
4
1
technology, organizational, and market complexity (Clemons and Weber 1990, Bacon 1992). An
2
important aspect of the IT investment decision-making process is analyzing the market landscape
3
to identify trends in IT development, which is often helpful for investment planning and product
4
development strategies. Formal analysis in this domain has traditionally been difficult primarily
5
due to the sheer number of available technologies and the complex inter-relationships among
6
them. Compounding this problem is the fact that practitioner knowledge of the historical drivers,
7
relationships, and patterns of technology evolution is often limited, and rarely is it well-
8
structured knowledge outside the realm of the IT forecasting and consulting firm ‘gurus.’ The
9
objective of this research is to address this problem by providing industry practitioners with a
10
new set of tools for analyzing the IT landscape. Our artifacts consist of a representation structure
11
and a set of constructs for defining a technology ecosystem, and a new methodology for identify-
12
ing trends within an ecosystem, consistent with the design science research paradigm outlined by
13
Hevner et al. (2004). We build upon existing theory related to technology evolution and IT in-
14
novation, and use a process theory approach for defining the constructs upon which we formu-
15
late and develop our tools.
16
2.1. The Process Theory Perspective for Design Science Research
17
Process theory provides the ideal theoretical lens for developing tools to analyze the IT land-
18
scape. Understanding the complex IT landscape and changes that occur within it over time re-
19
quires sensemaking (Weick 1979) of an environment that consists of many technologies and rela-
20
tionships. Process theories are concerned with explaining how outcomes evolve or develop over
21
time (Mohr 1982, Markus and Robey 1988), and, therefore, the process theory approach is a
22
proper base for developing tools for evaluating changes in the IT landscape.
5
1
The process theory approach has been used extensively in IS research, most notably as a base
2
for structuration analysis (Orlikowski and Robey 1991, Orlikowski 1993), and for modeling se-
3
quences of events (Abbot 1990, Newman and Robey 1992). In each of these cases, the process
4
theory approach was applied to inform the development of new techniques for analyzing com-
5
plex process data. Process data are difficult to analyze and manipulate because they deal mainly
6
with sequences of events, often involve multiple levels and units of analysis, vary in terms of
7
temporal precision and duration, and tend to focus on eclectic phenomena such as changing rela-
8
tionships (Langley 1999). This description matches the type of data used to analyze trends in an
9
IT landscape very well – event data (e.g., new technology introductions) consisting of multiple
10
types of technologies with differing attributes and various temporal measures. Our objective, as
11
noted earlier, is to develop tools that help practitioners understand how technologies evolve over
12
time and why they evolve in a certain way, which is also a key objective of process-data-related
13
research (Van de Ven and Huber 1990).
14
Langley (1999) outlines seven strategies for sensemaking and theorizing with process data,
15
and our current research uses a combination of two of them – the quantification strategy and the
16
visual mapping strategy – to develop its theoretical perspective and empirical methodology. The
17
quantification strategy, as exemplified by the research of Van de Ven and Poole (1990), involves
18
the systematic coding of events according to predetermined characteristics. It further involves
19
gradually reducing the complexity of the process data to a set of quantitative time series that can
20
be analyzed using empirical methods (e.g., Garud and Van de Ven 1992, Romanelli and
21
Tushman 1994). The visual mapping strategy, which can be used for the development and veri-
22
fication of theoretical ideas, involves producing graphical representations of process data to pre-
23
sent large quantities of data in relatively little space (Miles and Huberman 1994). Visual map-
6
1
pings allow simultaneous representations of a large number of dimensions and can be easily used
2
to show precedence, parallel processes, and the passage of time (Langley 1999). A diagram can
3
provide an intermediate step between raw data and theory development, and the comparison of
4
several diagrams can help researchers look for common sequences of events, patterns, and pro-
5
gressions in the data (Langley and Truax 1994). Both organization researchers (e.g., Meyer
6
1991, Meyer and Goes 1988) and decision researchers (e.g., Mintzberg et al. 1976, Pentland
7
1995) use the visual mapping strategy to develop process maps.
8
Our research combines the principles and sensemaking strategies of process theory with the
9
formal guidelines of design science research (Hevner et al. 2004) to develop new tools for mod-
10
eling and analyzing technology evolution. Our tools are designed to provide structure and a
11
means of analyzing the complex IT landscape by producing graphical representations of patterns
12
and trends. We use a quantification strategy to define a set of constructs for coding technology
13
innovation types and a methodology for empirically identifying patterns of technology evolution.
14
We also use a visual mapping strategy to represent these patterns over time, which is illustrated
15
through two examples of IT evolution. The two strategies complement each other by providing
16
theoretical and methodological underpinnings for the new artifacts that we present.
17
2.2 The Technology Ecosystem
18
A technology ecosystem view is a useful approach for representing the many technologies and
19
relationships that make up the IT landscape. Following this analogy, an IT ecosystem is com-
20
prised of a set of technologies that interact with and impact the success of one another. Hannan
21
and Freeman’s (1977 and 1989) seminal work on organizational ecology has sparked the in-
22
creased use of an ecological analogy in business and organization research. The theoretical per-
23
spective of organizational ecology is used to examine: (1) the environment in which organiza-
7
1
tions compete, and (2) the birth and death processes of firms. Strategy researchers have also
2
adopted a biological ecosystem model in the analysis of business relationships and strategic deci-
3
sion-making (Iansiti and Levien 2002, 2004). Managers and academicians are recognizing the
4
value of the ecosystem metaphor for understanding the complex network of business relation-
5
ships within and across industries (Harte et al. 2001). Most recently, IS researchers have also
6
begun to adopt an ecosystem perspective: Quaadgras (2005) used network modeling techniques
7
to define the RFID business ecosystem and forecast firm participation, Nickerson and zur
8
Muehlen (2006) analyzed Internet standards creation using a population ecology model, and
9
Funk (2007) presented a hierarchy of relationships between technologies to determine the timing
10
of dominant technology designs.
11
Although the ecosystem view is proving to be important in business and research, in most
12
cases the ecosystem perspective has been used merely as means of starting discussion. There has
13
been a lack of development of analytical tools that provide real value to practitioners based on an
14
ecological perspective. Additionally, previous research incorporating the ecosystem analogy has
15
focused primarily on industrial ecosystems and relationships between firms and organizations.
16
The current research applies the ecosystem analogy to the task of understanding the relationships
17
among technologies in the IT landscape. Two recent papers provide insights for modeling trends
18
in technology evolution using an ecosystem view. Lyytinen and Rose (2003) developed a model
19
of disruptive IT innovation that considers the interrelationships between technological innova-
20
tions at the systems development, infrastructure, and IT service levels. Considering cross-level
21
impacts of innovation is a key step in understanding the relationships between technologies as an
22
ecosystem. Adomavicius et al. (2007) developed a formal ecosystem model of technology evolu-
23
tion for the purposes of representing relationships that occur with technology changes. We build
8
1
on this model to develop a new set of constructs and methods for identifying trends in techno-
2
logical changes within an IT ecosystem.
3
The term technology ecosystem emphasizes the organic nature of technology development
4
and innovation that is often absent in standard forecasting and analytical methods. The tradi-
5
tional notion of an “ecosystem” in biological sciences describes a habitat for a variety of differ-
6
ent species that co-exist, influence each other, and are affected by various external forces. In the
7
ecosystem, the evolution of one species affects and is affected by the evolution of other species.
8
By considering the technology ecosystem as an interrelated set of technologies, a manager can
9
more successfully identify factors that may impact innovation, development, and adoption of
10
11
new technologies—and ultimately the success of the business activities that use the innovations.
The ecosystem model of technology evolution (Adomavicius et al. 2007) integrates the
12
strengths of many modeling methods and theoretical frameworks in economics, engineering, and
13
organizational theory. Three key research streams are combined to provide a comprehensive
14
conceptual model of evolution within a technology ecosystem. First, the population perspective
15
(Saviotti and Metcalfe 1984, Saviotti 1996) proposes that multiple interrelated technologies
16
should be viewed as a system or population whose characteristics and members change over
17
time. This concept of viewing technologies as an interrelated system is also supported by Dosi’s
18
technology paradigms (1982), Nelson and Winter’s technology regimes (1982), Laudan’s tech-
19
nology complexes (1984), and Sood and Tellis’ platform of innovation (2005). Second, complex
20
systems of technologies can be organized in hierarchies (Clark 1985, Rosenkopf and Nerkar
21
1999), which leads to the definition of specific roles played by technologies in the ecosystem.
22
Three levels of the hierarchy are typically considered. In particular, component-level technolo-
23
gies combine to form product-level technologies, and products are then combined to form a sys-
9
1
tem of use. Co-evolution of technologies in this model occurs both within and across levels in
2
the hierarchy (Campbell 1990, Rosenkopf and Nerkar 1999). Finally, technologies tend to fol-
3
low specific trajectories (Dosi 1982) and patterns of innovation (Sahal 1981 and 1985) through
4
the process of technology evolution. Baldwin and Clark (1997 and 2000) argue that, in the “age
5
of modularity,” specific design rules govern the common patterns of technological innovation.
6
These include the combination of two technological modules into one, or the augmentation of an
7
existing module into a new module.
8
We next build on technology ecosystem theory and use a process theory approach to develop
9
our model and constructs. We develop a means of representing the IT landscape analysis prob-
10
lem in terms of an IT ecosystem. We then present a methodology for identifying and analyzing
11
trends in technological change within an IT ecosystem over time.
12
3. MODEL AND CONSTRUCTS
13
The core of our model is a set of roles and relationships that are used to code technology innova-
14
tions and represents patterns of technological change in an ecosystem. In this section, we discuss
15
our use of theory (Gregor 2006) in support of our effort to design artifacts and define constructs
16
that provide foundations for a new visual representation and empirical approach for modeling
17
technology evolution patterns over time.
18
3.1 Roles in the Technology Ecosystem
19
The hierarchical nature of technologies within a population leads to the identification of specific
20
roles that technologies can play within the ecosystem (Adomavicius et al. 2007). By acting
21
through these roles, classes of technologies can influence each other’s evolution and develop-
22
ment through common patterns of innovation. Specifically, the three roles that technologies play
10
1
within an ecosystem are: product and application, component, and support and infrastructure, as
2
defined in Adomavicius et al. (2007).
3
The product and application role describes technologies that are built up from a set of com-
4
ponents, and are designed to perform a specific set of functions or satisfy a specific set of needs.
5
This role includes the focal technology and other technologies in direct competition with it in the
6
given context and associated with a specific application or use. By focal technology, we mean
7
the technology in which the analyst is most interested for the purpose of analysis. By a given
8
context, we are referring to some capability, use or application of the focal technology. For ex-
9
ample, in the digital music technology ecosystem, an MP3 player (a focal technology) plays a
10
product role because it is composed of several components and is designed to provide a specific
11
service to its user: storing and playing digital music files (a given context). Additionally, MP3
12
players can compete with related technologies, such as CD players and satellite radio devices.
13
The component role describes technologies that are used as components or subsystems in
14
more complex technologies. Components are necessary for the product and application role
15
technologies to perform their functions in the given context of use. For example, there are sev-
16
eral technologies that act as components for the personal computer: microprocessors, RAM
17
chips, hard disk drives, etc. Individually and in combination, these components provide different
18
functionality to the product role technologies. This is an important relationship in the ecosystem
19
because individual technologies can act as components in multiple technologies and contain
20
components themselves. Consider the hard disk drive. It acts as a component in PCs, MP3 play-
21
ers, and many other devices. However, the hard disk drive also has a set of component technolo-
22
gies itself, including DC spindle motors, actuators, and platters. So, in our approach, the as-
23
signment of technology roles and the scope of an ecosystem are based on relevance to the focal
11
1
2
technology and the specific context of analysis.
The support and infrastructure role describes technologies that work in conjunction or col-
3
laboration with (or as a peripheral to) product and application role technologies. The distinction
4
between the component role and the support and infrastructure role is that components are neces-
5
sary for the design and are part of the physical structure of another more complex technology,
6
while support and infrastructure technologies simply work in combination with other technolo-
7
gies. A key point about the support and infrastructure role is that technologies add value to the
8
technologies they support. For example, a printer is not physically necessary for the design and
9
use of a PC, but it supports the PC’s functionality, expands the PC’s system of use, and together
10
they provide additional value and services to their users.
11
3.2 Paths of Influence in the Technology Ecosystem
12
The decision to invest in or develop a new technology often requires the manager to understand
13
trends in technological change to capitalize on market opportunities. Since technologies change
14
over time, any practical model of technological evolution must consider the temporal aspects of
15
such change.
16
Technological determinism and social construction of technology are two competing theories
17
that attempt to explain the change in technology over time. Technological determinism posits
18
that technological development drives social and cultural changes (Smith and Marx 1994), while
19
social construction argues the opposite: society and culture determine technological development
20
(Bjiker et al. 1987). A related, yet slightly different debate exists in the economics and manage-
21
ment literature. Demand-side forces, such as consumer and market needs, drive technological
22
development (e.g., Adner and Levinthal 2001, Clark 1985, Malerba et al. 1999). Another per-
23
spective though is that supply-side forces, such as firm capabilities and R&D, are responsible
12
1
(e.g., Dosi 1982, Sahal 1985). Below we model technological change as relationships between
2
specific technologies over time, which closely relates to the supply-side and technology deter-
3
minism perspectives. Our model is focused on the technical forces that drive technological
4
change (Nelson 1995) and does not explicitly model external forces such as society, culture, and
5
the demand environment. We recognize that society and culture impact the development and
6
evolution of technology. However, our model is targeted towards domain experts that are aware
7
of market forces and their target audience. In addition, in most firms that are consumers of IT,
8
IT investment decisions are for internal purposes and social construction is less of a concern.
9
While our long-term goal is to incorporate these forces and develop more comprehensive set of
10
interactions, in this paper we focus on reducing technological complexity while increasing the
11
understanding of the IT landscape for the purpose of technology forecasting. For this study, we
12
conjecture that domain experts will be able to define their contexts with appropriate considera-
13
tion for market and social forces – as demonstrated by the digital music example in Section 5.
14
Boland et al. (2003) argue that understanding the information technology adoption requires
15
an integrated view of the innovation process. In particular, their work highlights the importance
16
of history and the effects of time in understanding innovation (Arthur 1989) and viewing it as a
17
continuous “path creation” process. Similarly, to represent the influence that current technolo-
18
gies have on future technologies, the technology ecosystem model of Adomavicius et al. (2007)
19
defined paths of influence, which occur within or across the component, product and application,
20
and support and infrastructure technology roles within the ecosystem.
21
Technological evolution and development is complex and can take many paths through the
22
layers within a technology ecosystem. Therefore, paths of influence are possible from any of the
23
current states to any of the future states of the technology roles within an ecosystem. For exam-
13
1
ple, the current component technologies can potentially influence the development of new prod-
2
uct technologies, representing a specific path of influence: Component Role (C) Æ Prod-
3
uct/Application Role* (P*). Here the asterisk (*) is used to indicate a future state of a technol-
4
ogy role in the ecosystem, and C, P, I are used as abbreviations for component role, product and
5
application role, and support and infrastructure role, respectively. Roles without an asterisk rep-
6
resent current states in the ecosystem.
7
Paths of influence represent the influence one technology role has on another in the evolution
8
of a set of technologies in the ecosystem. For example, the success of the DVD player has
9
helped drive the development of new DVD component technologies. These include recordable
10
DVD ROMs, multi-layer DVD ROMs, and new blue-ray and HD technologies (PÆC*). Simi-
11
larly, the evolution of infrastructure technologies can drive the development of new product
12
technologies. For example, the wide cellular phone network provides infrastructure for new
13
phone services and applications such as text messaging and picture mail (IÆP*). Paths of influ-
14
ence follow the design science paradigm (Hevner et al. 2004), as they are a set of constructs that
15
provide a problem representation structure for analyzing technological interdependencies in
16
technology evolution. Table 1 presents a 3 x 3 matrix that classifies possible paths of influence.
17
INSERT TABLE 1 ABOUT HERE
18
We have discussed the roles technologies can play in an ecosystem and the paths of influence
19
possible between technology roles. These constitute the constructs of our model, and in the next
20
section we discuss several examples of patterns of technology evolution using these constructs.
21
We derive these constructs directly from the theory developed in Adomavicius et al. (2007). We
22
develop new methodologies for analyzing technology evolution based on these constructs and
23
using strategies of sensemaking from process data. In Section 4, we use these constructs as the
14
1
basis for a visual mapping strategy that generates graphical representations of evolutionary pat-
2
terns, and in Section 5 we demonstrate this using a qualitative evaluation of the digital music
3
technology ecosystem. In Section 6, we demonstrate how these constructs can be used to de-
4
velop a quantification strategy that empirically identifies evolutionary patterns using a data set of
5
over 3,000 wireless networking technologies.
6
4. REPRESENTING PATTERNS OF TECHNOLOGY EVOLUTION
7
Sahal (1981 and 1985) defines an analytical framework for understanding patterns of technologi-
8
cal evolution. The author identifies several specific patterns, including invention, innovation,
9
and diffusion, and recognizes that technology development follows an evolutionary process.
10
Additionally, the author notes that the temporal evolution of any system that is subject to a col-
11
lection of random changes is characterized, in general, by prolonged oscillations. Therefore, the
12
existence of recurring fluctuations in innovation and technology evolution should be expected.
13
Worlton (1998) investigates the recent evolution of high-performance computing and identifies
14
patterns of technological change. He notes that technological change typically occurs in four
15
stages: invention, innovation, diffusion, and change of scale. Rosenberg (1994, p. 74) shows that
16
“technological forces exist which may lead to cyclical behavior in certain industries, where ma-
17
jor innovations come to substitute for one another sequentially in time.” Also, Baldwin and
18
Clark (1997 and 2000) argue that, due to increased modularity, specific design rules can lead to
19
predictable innovation patterns.
20
4.1 Coding and Representation
21
We next define a visual mapping strategy for representing patterns of technological change based
22
on the roles technologies can play and the paths of influence among them within an ecosystem.
23
Visual mapping strategies provide a means for simultaneously representing multiple dimensions
15
1
and can help researchers identify common patterns, sequences, and progressions in process data
2
(Langley 1999, Langley and Truax 1994). For example, Nickerson and zur Muehlen (2006)
3
demonstrate the use of a population ecology perspective to map out the complexities in Internet
4
standards making. They use a visual mapping strategy to represent the “space-time network” of
5
the migration of ideas generated during Web services standards development. Similarly, Lyyti-
6
nen et al. (2003) use a visual mapping strategy to represent path creations in industries as a result
7
of IT-led innovations.
8
As technologies evolve, some new technologies are introduced, and some existing technolo-
9
gies die out. An ecosystem’s form and content change as well as the patterns of evolution occur-
10
ring within the ecosystem. The component, product, and infrastructure roles can be used to code
11
technologies according to their position within an ecosystem. Within a specific time period, the
12
quantity of technologies in each role determines the dominance of a role in the ecosystem. Tran-
13
sitions from one set of dominant roles in the ecosystem to another set can then be represented by
14
paths of influence and over time evolutionary patterns can be represented as collections of the
15
paths of influence occurring in the technology ecosystem. At any given point in time, any col-
16
lection of paths of influence can be at work in the ecosystem. Table 1 provides a pattern tem-
17
plate, and Figure 1 provides examples of several evolutionary patterns, each one represented by a
18
collection of certain paths of influence occurring at the same time.
19
INSERT FIGURE 1 ABOUT HERE
20
An alternative representation of the patterns can be achieved using a graph-based approach.
21
The nodes in Figure 2 represent the collection of component (C), product (P), and infrastructure
22
(I) technologies at each time in the evolutionary process. The edges between nodes represent the
23
individual paths of influence as outlined earlier. Finally, the evolutionary patterns are repre-
16
1
sented by the set of edges in each time period.
2
INSERT FIGURE 2 ABOUT HERE
3
An aspect of the power of coding technologies into the component, product, and infrastruc-
4
ture roles is the complexity that emerges from a seemingly simple model. Three roles may seem
5
to offer a rather simple representation, however, nine possible paths of influence emerge from
6
these roles and dozens of possible patterns of technology evolution emerge as various combina-
7
tions of different paths of influence. The three technology roles provide a simple set of con-
8
structs that is strongly based on a synthesis of prior literature and provide a means for represent-
9
ing a large number of complex patterns of technological change.
10
4.2 Example Patterns of Technology Evolution
11
To demonstrate the mapping technique described above, we present five examples of technology
12
evolution patterns. (See Figure 1.) They are based on observations of real-world instances of
13
technology evolution and development, and some of these patterns can also be traced back to in-
14
novation theory outlined at the beginning of this section.
15
Product Development. The first example pattern of technology evolution describes the ini-
16
tial stages of the evolution of a particular product technology. In this pattern, the ecosystem may
17
exhibit many different product and component technologies, as designs are being developed and
18
accepted. We call this the product development pattern because the primary paths of influence at
19
work in the ecosystem deal with the impacts of new technology components and products on
20
each other. (See Figure 1a.) For example, in the personal computing ecosystem the introduction
21
of the first personal digital assistants (PDAs), such as the Apple Newton and the Palm Pilot, re-
22
quired the simultaneous development of products, the PDAs themselves (PÆP*), and compo-
23
nents, microprocessors, micro hard disk drives, touch screens, etc. (CÆC*). Component tech-
17
1
nologies are developed and refined to meet requirements of new products (PÆC*) and, similarly,
2
product designs are modified and refined based on the newly available components (CÆP*).
3
Product and Infrastructure Alignment. A second pattern – product and infrastructure
4
alignment – depicts a scenario where a set of infrastructure technologies already exists. (See
5
Figure 1b.) Infrastructure technologies provide incentives for the ongoing use and adoption of
6
product technologies in an ecosystem and enhance the utility of existing technologies. New
7
products, in turn, leverage existing infrastructure to provide new services and functionality. For
8
example, in the European mobile phone ecosystem there was a mutual development of 3G wire-
9
less networks and the mobile phones that would operate on them. New network capabilities
10
drive innovations in mobile phones and the desire to provide more functionality in phones drives
11
the development of the network. This demonstrates a collection of paths of influence, (e.g.,
12
PÆI*, I*ÆP), working in parallel to ensure interoperability of products and infrastructure.
13
Feed-Forward Evolution. Two more patterns describe what Campbell (1990, 1994) calls
14
cross-level co-evolution. The paths of influence in the upper right section of the 3 x 3 matrix are
15
feed-forward paths of influence. In each case, evolution of current technologies impacts tech-
16
nologies at higher levels of the technology hierarchy in the future state. (See Figure 1c.) The
17
introduction of an evolved component technology can drive evolution of the product technolo-
18
gies that will use this new component. When new components are combined to create a new
19
product, this path of influence reflects the process of design and compilation. For example, in
20
the digital camera ecosystem the development of color LCDs, high-capacity solid-state storage
21
devices, and CMOS sensors for image capture provided the component technology enhance-
22
ments to create a digital camera product technology (CÆP*).
23
Feed-Back Evolution. The paths of influence in the lower left section of the 3 x 3 matrix re-
18
1
fer to feed-back paths of influence in technology evolution. (See Figure 1d.) Often the use of a
2
technology impacts the development and evolution of the technologies on which it depends. For
3
example, in the web programming ecosystem the development of many new Internet component
4
technologies such as Dynamic HTML or Active Server Pages (ASP) resulted from the continu-
5
ous growth of the World Wide Web (IÆC*). Also, once a support or infrastructure technology
6
is in place, it provides opportunities for a new product technology to leverage the services and
7
facilities it provides. The widespread adoption of instant messaging (IM) paved the way for new
8
products and applications, such as IM-based video conferencing, file transfer, and online game
9
software, leveraging the existing infrastructure (IÆP*).
10
Incremental Evolution. The last example pattern, incremental evolution, represents within-
11
level co-evolution and continuous development and refinement of technologies within each role.
12
(See Figure 1e.) For example, Moore’s Law explains that processing power of integrated chips
13
will double approximately every eighteen months, and ongoing research and development con-
14
tinues to uphold this law (CÆC*). In the mobile phone ecosystem, camera phones are an exam-
15
ple of the integration of two existing technologies, digital cameras and mobile phones, to create a
16
new product (PÆP*) and continuous expansion and refinement in the cellular network is an ex-
17
ample of incremental infrastructure evolution (IÆI*). Incremental evolution can be considered a
18
type of steady-state evolution in which expected incremental technological progress occurs
19
within a role.
20
Although the ability to identify single patterns of evolution in technology ecosystems pro-
21
vides additional insights for managers making technology-related decisions, a more significant
22
contribution of the ecosystem model is its ability to provide a systematic approach for describing
23
the temporal changes (transitions) in the ecosystem using these evolutionary patterns. In the next
19
1
section, we use an example to illustrate the utility of our approach by examining how a series of
2
evolutionary transitions can be represented by connecting multiple patterns over time, which can
3
lead to the identification of some intrinsic patterns (e.g., in certain cases representing evolution-
4
ary cycles) that can be utilized in technology forecasting.
5. QUALITATIVE APPLICATION: EVOLUTIONARY TRANSITIONS IN
DIGITAL MUSIC TECHNOLOGIES
5
6
7
8
We next use the landscape of digital music technologies evolution to provide a demonstration of
9
a qualitative approach for applying the constructs and patterns discussed in Sections 3 and 4.
10
This example demonstrates the design and use of our constructs and communicates their rele-
11
vance for both IS research and practice. We identify technology evolution patterns, and use the
12
visual mapping strategy (Langley 1999) to create a state diagram to map the patterns of techno-
13
logical change and cycles of innovation that emerge from patterns of evolution discussed in Sec-
14
tion 4.
15
5.1 Qualitative Analysis Approach
16
We use the evolution of digital music technologies to demonstrate the effectiveness of identify-
17
ing transitions between common evolutionary patterns in technology ecosystems. The digital
18
music technology ecosystem is an ideal setting for analysis. It includes many different compo-
19
nent, product, and infrastructure technologies. Furthermore, most digital music technologies
20
have only existed since the mid-1990s, yet there has been a significant amount of technological
21
change in this ecosystem. Demand for these technologies has skyrocketed over the past several
22
years, so we expect to see many new product introductions and new entrants in the market. Ad-
23
ditionally, digital music technologies span the consumer electronics, entertainment, and com-
24
puter industries, which suggest that there is an underlying complexity in their design and rela-
25
tionships within the ecosystem. Furthermore, digital music technologies have revolutionized the
20
1
consumption of music and other forms of media, and most people can relate to these technolo-
2
gies since they likely own or use them. We will show that, in its young life, the digital music
3
ecosystem has passed through at least two evolutionary cycles.
4
In developing this qualitative example, we follow the descriptive approach outlined by Hev-
5
ner et al. (2004) to demonstrate the application of our proposed constructs. We used LexisNexis
6
and Internet search tools to gather announcements, news stories, and historical records related to
7
technologies in the digital music ecosystem between 1989 and 2006. In total, we gathered in-
8
formation on approximately 100 related technologies (e.g., flash-based storage, LCD screens,
9
MP3 players, digital music services). These were coded into examples of new technologies in
10
the component, product and infrastructure roles. Using information on the timing of technology
11
releases, we developed a rich qualitative interpretation of technology trends in the ecosystem.
12
While we do provide insights on the nature of digital music technology evolution, the objective
13
of this analysis is to illustrate the use of our artifacts for analyzing an IT ecosystem qualitatively.
14
5.2 Technology Evolution in the Digital Music Ecosystem
15
The demand for digitally-formatted music files, players, and services has grown steadily over the
16
past five years. In fact, a new digital music market has developed with many technological in-
17
novations and rapid consumer adoption. Since it was originally patented in Germany in 1989,
18
the MP3 audio compression format has had a significant impact on the traditional music indus-
19
try. In 1999, peer-to-peer (P2P) file sharing networks gained rapid acceptance, sparking legal
20
battles and the development of new encryption and file-tracking technologies. In February 1999,
21
Sub Pop Records became one of the first labels to begin releasing music in the MP3 format
22
(Wired News 1999). Since then, the introduction of mass storage digital music players and on-
23
line digital music retailers has transformed the music business. Table 2 and Figure 3 provide
21
1
multiple illustrations of the timeline and two cycles of the evolution of digital music technolo-
2
gies.
3
4
INSERT TABLE 2 AND FIGURE 3 ABOUT HERE
The first cycle of digital music technology evolution started with the introduction of the MP3
5
compression format and software applications for playing MP3-encoded music files. The birth
6
period of the digital music industry was characterized primarily by the initial product develop-
7
ment pattern of technology evolution, where component technologies (such as the MP3 compres-
8
sion format) and product technologies (digital music software, such as WinAmp) were being re-
9
fined as they gained more attention. Activities in this era included the refinement of the MP3
10
format by integrating it into MPEG-1 in 1992 and MPEG-2 in 1994. Once MP3 files reached a
11
reasonable level of adoption, feed-forward patterns of technology evolution took over as new
12
product and infrastructure technologies were introduced based on the MP3 encryption format.
13
The first portable MP3 player, the 32MB MPMan device from Eiger Labs, was released in mid-
14
1998 (Van Buskirk 2005), and P2P networks were introduced with Napster’s inception in May
15
1999. Both technologies emerged because of the popularity of the MP3 compression format.
16
As popularity increased for the technologies in the digital music ecosystem, additional infra-
17
structure technologies were developed, including refinements to P2P networks and the introduc-
18
tion of new MP3 standards, such as Microsoft’s WMA and Apple’s AAC. As a result of the con-
19
tinued growth in popularity of digital music products and technologies, feed-back patterns of
20
technology evolution took hold, and new components and products, such as higher capacity flash
21
storage based players, were developed. At this point the majority of MP3 players were flash-
22
storage-based and virtually all MP3 file distribution occurred over P2P networks.
23
A second evolutionary cycle began when innovations in components, such as high-capacity
22
1
micro hard disk drives, led to the initial product development of hard disk drive-based MP3
2
players, such as the Apple iPod and the Creative Nomad Jukebox. These new HDD-based play-
3
ers sparked a new feed-forward patterns of evolution that resulted in the introduction and adop-
4
tion of new online music services, such as iTunes and Napster 2.0, as well as a slew of accesso-
5
ries for portable MP3 players, such as FM transmitters and voice recorders. With the presence of
6
multiple online music providers and portable MP3 players, technology evolution became focused
7
on the alignment of infrastructure and product technologies. The wide adoption of the second-
8
generation digital music technologies led to feed-back patterns that included introduction of new
9
products using new components such as color LCD screens.
10
5.3 Mapping the Analysis Back to the Constructs
11
The events that occurred in the digital music technology ecosystem can be represented as pat-
12
terns of technological change using the roles and paths of influence defined in Section 4. Figure
13
4 provides a visual representation of the two evolutionary cycles that were described above. This
14
represents a series of evolutionary transitions that represents connections between multiple evo-
15
lutionary patterns over time. Coding the technologies into roles allowed us to identify paths of
16
influence, represented by the arrows in Figure 4, and multiple patterns of technology evolution,
17
represented by the collection of arrows in each time period. As the figure shows, we can then
18
depict the sequence of 3 x 3 matrices representing paths of influence as a state diagram. We be-
19
lieve that both qualitative and quantitative data can be used to develop these evolutionary paths
20
and cycles, enabling an analyst to understand and predict next generation of technologies in the
21
desired context. Our development is supported by prior research, which shows that innovation
22
typically occurs in specific patterns (Sahal 1981 and 1985, Baldwin and Clark 2000, Worlton
23
1998) and in some cases cycles (Rosenkopf and Nerkar 1999, Worlton 1998). It also shows that
23
1
2
3
new innovations typically replace existing ones (Rosenkopf and Nerkar 1999).
INSERT FIGURE 4 ABOUT HERE
Based on the previous cycles, an analyst might forecast that the next cycle of technology evo-
4
lution in the music industry will begin as new components are introduced that allow for even
5
more advanced features in product technologies. For example, the evolution of components that
6
are used across multiple ecosystems may result in the convergence of hand-held computing de-
7
vices (e.g. PDAs, cellular phones, MP3 players, digital cameras). In fact, Motorola introduced
8
one of the first MP3-enabled mobile phones (Shillingford 2005), and the Sony/Ericsson Walk-
9
man MP3 phone grew sales by 33% in the second quarter of 2006 (Ewing and Burrows 2006), as
10
demand expanded in the presence of falling prices. Microsoft released a new multimedia-
11
playing (audio, video, and software) hand-held device called a “Zune” in November 2006, and
12
Apple released its “iPhone,”an iPod-Phone hybrid device, in mid-2007.
13
Currently, the ecosystem is rapidly evolving, with reconsideration being given to embedded
14
digital rights management (DRM), extensions involving web services on hand-held devices, GPS
15
built-in functionality for location-based shopping, and component innovations that will eventu-
16
ally support mobile TV. These all reflect digital convergence with technology components that
17
have been developed outside the hand-held sphere of technological influence. The ecosystem
18
view allows the manager to model these types of evolutionary patterns as well as track and ana-
19
lyze their progression over time (including the identification of transitions and cycles), which
20
will provide a better understanding of the dynamic nature of technology evolution.
6. QUANTITATIVE APPLICATION: EVOLUTIONARY TRANSITIONS
IN WI-FI TECHNOLOGIES
21
22
23
24
To further substantiate the application of our constructs and model, we developed a new empiri-
25
cal methodology to identify technology evolution patterns by combining a quantification strategy
24
1
with the visual mapping state-diagram-based approach for sensemaking from process data
2
(Langley 1999, Van de Ven and Poole 1990). A quantitative approach for analyzing the IT land-
3
scape provides a strong complement to the qualitative approach we demonstrated in Section 5.
4
When sufficient data on the introduction of technologies are available, a quantitative approach
5
provides additional rigor to the identification of evolutionary patterns. We follow guidelines in
6
Hevner et al. (2004) and use an analysis of real data to demonstrate our empirical methodology.
7
6.1. Data
8
The wireless networking ecosystem provides an appropriate context for applying our empirical
9
methodology for several reasons. Similar to the digital music technology ecosystem example,
10
wireless networking technologies are relatively young but have experienced a large amount of
11
technological change; in addition, these technologies also fit into the computer and consumer
12
electronics industries, so complexity is high. However, unlike digital music technologies, wire-
13
less networking technologies have had clearly-defined generations based on IEEE standards.
14
The existence of these standards suggests that recognizable patterns may exist, and our empirical
15
methodology can be validated by identifying those patterns. Furthermore, wireless networking
16
technologies are used not just by individuals but also by firms and organizations. They can also
17
be certified by the Wi-Fi Alliance (www.wi-fi.org), which maintains a database of product certi-
18
fications and makes the data publicly available:
19
20
21
22
“The Wi-Fi Alliance is a global, non-profit industry trade association with more than 200 member companies devoted to promoting the growth of wireless local area networks (WLAN). Our certification programs
ensure the interoperability of WLAN products from different manufacturers, with the objective of enhancing the wireless user experience.”
23
This provides an opportunity to compare and contrast different ecosystems, and to evaluate our
24
constructs and visual mapping strategy with both qualitative and quantitative analyses.
25
We collected data on over 3,000 certifications for new wireless networking (802.11) tech-
25
1
nologies awarded by the Wi-Fi Alliance. The member companies of the Wi-Fi Alliance include
2
3Com, Apple, Dell, Intel, Linksys, and many others. Certifications are awarded for ten different
3
technology categories: access points, cellular convergence products, compact flash adapters, em-
4
bedded clients, Ethernet client devices, external cards, internal cards, PDAs, USB client devices,
5
and wireless printers. Technologies can be certified based on IEEE communication standard
6
(802.11a, b, g, d, and h), security (e.g., WPA, and WPA2), authentication protocol (e.g., EAP,
7
and PEAP), and quality of service (e.g., WMM). 2
Generally, historical product data that includes comprehensive technical specifications and
8
9
dates of release is difficult to obtain. However, the Wi-Fi Alliance certifications have been
10
awarded to a substantial number of products, with most certified prior to their commercial re-
11
lease. For this reason, we have used the date of certification as a proxy for the date of innovation
12
for a new technology, and the type of certification as a proxy for the technical specifications of
13
the product. Both are readily observed, and the former is likely to occur close to the date of in-
14
novation, and so they represent acceptable empirical proxies.
15
We coded the Wi-Fi certification categories into the ecosystem roles (component, product,
16
and infrastructure) based on our operationalization of the theory of the ecosystem model of tech-
17
nology evolution. Compact flash adapters, internal cards, external cards, and USB client devices
18
were coded as component technologies, because each clearly acts as a component by providing
19
wireless capabilities for product devices. We coded access points, Ethernet client devices, and
20
wireless printers as infrastructure, because these technologies either form or extend the network
2
WPA (Wi-Fi protected access) is a standard for wireless network security. For more information, see
en.wikipedia.org/wiki/WPA. EAP (extensible authentication protocol) is a universal authentication framework frequently used in wireless networks, and PEAP is an open-standard authentication framework based on EAP proposed
by Cisco, Microsoft, and RSA Security. See en.wikipedia.org/wiki/Extensible_Authentication_Protocol for additional information. WMM (Wi-Fi multimedia, also known as WME – wireless multimedia extensions) is a standard
that provides basic quality of service (QoS) for wireless networks by prioritizing traffic according to the following
access categories: voice, video, best effort, and background. en.wikipedia.org/wiki/WMM offers details.
26
1
infrastructure necessary for wireless communication. Finally, we coded PDAs, embedded cli-
2
ents, and cellular convergence technologies as products, because each represents a product de-
3
vice that provides fully functioning wireless networking capabilities to the end user. 3 Coding
4
the wireless technologies into appropriate roles will lead to the identification of dominant roles
5
and the paths of influence between roles. The collections of these paths of influence at different
6
time periods will represent patterns of technology evolution in the ecosystem.
7
6.2. Empirical Methodology
8
Following a quantification strategy similar to Van de Ven and Poole (1990), we present a meth-
9
odology for reducing the complexity of technology evolution process data to a set of time-based
10
quantitative data that can be used to empirically identify patterns. Table 3 provides a high-level
11
description of the steps in the methodology.
INSERT TABLE 3 ABOUT HERE
12
13
First, raw technology evolution data must be coded according to the component, product, and
14
infrastructure roles within a specific ecosystem (Step 1 in Table 3). As noted above, the wireless
15
networking data were coded based on the product category assigned to a technology in the Wi-Fi
16
Alliance certification. Technical specifications for wireless networking technologies exhibit a
17
natural progression over time. For example, the IEEE 802.11b standard was introduced prior to
18
the 802.11g standard and, therefore, new technology introductions are distributed accordingly,
19
based on their technical specifications. We use the technical specifications of IEEE communica-
20
tion standard (802.11b versus 802.11g) and the basic security standard (WPA1 versus WPA2) to
21
identify different generations of wireless technologies. We independently analyzed each 802.11
3
Cellular convergence certifications are awarded to cellular phones that are Wi-Fi enabled. Embedded clients are
certifications for PCs, laptops, and other end user devices that are certified with internal Wi-Fi capabilities.
27
1
and WPA generation (i.e., two generations in each category) and then made comparisons across
2
generations to identify cycles of technology evolution.
3
Next, to derive a baseline for the number of new technologies introduced over time, we esti-
4
mate a function of the frequency of all technologies introduced across all roles (Step 2 in Table
5
3). This function provides an approximation for the total innovative activity in the technology
6
ecosystem over time. A wide variety of approximation techniques may be used for this purpose.
7
For example, for the wireless network data we used a five-month moving average of the fre-
8
quency counts (i.e., to eliminate random monthly fluctuations and obtain the underlying trend)
9
and estimated the frequency curve using a polynomial approximation function. In this specific
10
case, a sixth-degree polynomial provided a good fit with R2 values over 90%. Figure 5 depicts
11
the estimation for the frequency of technology introductions in the 802.11b generation.
12
INSERT FIGURE 5 ABOUT HERE
13
We next derived threshold frequency functions for each role using the frequency function es-
14
timated for all technology introductions (Step 3 in Table 3). If we assume that the introductions
15
of new technologies are independent of the role to which they belong and there are no interde-
16
pendencies across roles, we would expect to see the number of technology introductions in each
17
role over time be proportional to the total number of technology introductions in the ecosystem.
18
With this in mind, we derive estimated frequency functions for each role based on the proportion
19
each role has of the total number of technology introductions. For reference, the set of technolo-
20
gies released in the 802.11b generation is 54.4% components, 10.4% products, and 35.2% infra-
21
structure, and the set released in the 802.11g generation is 45.9% components, 8.0% products,
22
and 46.1% infrastructure. In Figure 6, the top curve represents the estimated frequency function
23
for all technologies and the three curves below represent the proportional frequency estimates for
28
1
components, infrastructure, and products, respectively from the top.
2
INSERT FIGURE 6 ABOUT HERE
3
Estimating the proportional frequency curves is necessary in order to take into account scale
4
differences in the number of technologies introduced in each role. In the context of the wireless
5
networking data, the total number of product certifications is significantly lower than the number
6
of component and infrastructure certifications. There are several possible reasons for the lower
7
number of product certifications. In particular, one certification for an embedded client or cellu-
8
lar convergence technology is often applied to multiple product models using the same technol-
9
ogy. For example, Dell may certify one laptop-embedded client and then apply the certification
10
to multiple laptops using the same client. Also, not all product technologies in this ecosystem
11
need to be certified. For example, a laptop that uses a wireless adapter is a product technology in
12
this ecosystem; however, the laptop itself is not certified; only the adapter is.
13
The proportional frequency functions are used as thresholds for determining the dominant
14
technology roles over time (see Step 4 in Table 3). If the number of actual technologies released
15
for a certain role is above (below) the threshold, then one can argue that there is proportionally
16
more (less) innovative activity occurring in that role than expected under the assumption of inde-
17
pendent technology introductions and no interdependencies among roles. Using error bars, in
18
this case exogenously set at ±5% of the threshold curve value, actual frequencies of technology
19
introductions in each role are compared to the threshold (plus or minus error) to determine which
20
roles are dominant at what times. Figures 7 and 8 present this comparison for the 802.11b and
21
802.11g wireless technology generations. In the figures, the solid line with error bars is the thre-
22
shold curve, the dotted line represents the actual frequency counts per month, and the smoothed
23
line represents a five-month moving average of the frequency counts. From these plots it is ap-
29
1
parent that over time the dominant technology roles vary. For example, for the 802.11b genera-
2
tion it is clear that component and infrastructure technologies either trace the threshold or surpass
3
it for the first half of the generation, but they begin to lag in the second half as product technolo-
4
gies begin to dominate. Similar patterns are apparent in the 802.11g figure.
5
INSERT FIGURES 7 AND 8 ABOUT HERE
6
The results of the threshold comparisons discussed above can be represented using the visual
7
mapping strategy discussed previously. By identifying the dominant technology roles in each
8
time period, a state diagram can be created to represent the transitions across technology evolu-
9
tion patterns (see Step 5 in Table 3). The next section provides the examples of state diagrams
10
obtained from the Wi-Fi certification data using the proposed methodology.
11
6.3. Mapping the Analysis Results Back to the Constructs
12
Based on the analysis presented in Figures 7 and 8 for the 802.11b and 802.11g wireless tech-
13
nologies, the state diagram in Figure 9 was generated, which allows several general trends to be
14
observed. The empirical method described above identifies the dominant technology type within
15
each time period (represented as nodes in Figure 9), and the expert can then define the transitions
16
from one time period to the next (represented as arrows in Figure 9) using contextual information
17
from the domain. In particular, it is apparent that innovations in product technologies clearly lag
18
the introduction of new component and infrastructure technologies. As 802.11b component and
19
infrastructure innovation intensity begins to drop off, an increase in 802.11b product certifica-
20
tions develops as well as the initial certifications for 802.11g components and infrastructure (i.e.,
21
the next innovation cycle begins). In addition, product innovations initially lag component and
22
infrastructure innovations, but for the second generation this lag is shorter, likely because
23
802.11g products are backwards-compatible with 802.11b components and infrastructure.
30
1
Manufacturers are able introduce the next generation (802.11g) of wireless product technologies
2
more quickly without having to wait for the widespread development of 802.11g components
3
and infrastructure.
4
5
INSERT FIGURES 9 AND 10 ABOUT HERE
A state diagram for the WPA1 and WPA2 generations in the same wireless networking data
6
is presented in Figure 10. In this case, it is also apparent that component technology innovations
7
predate product and infrastructure technology innovations. In the WPA2 generation it is very
8
clear that the progression of technological innovation was from components to infrastructures to
9
products, while in the WPA1 case there is an initial component precedence followed by a domi-
10
nance of infrastructure and products.
11
Using the information provided by these two cycles of technology evolution in the Wi-Fi
12
ecosystem, an analyst may be able to forecast that the lag between component and infrastructure
13
technology innovations and product technology innovations will continue to reduce, and eventu-
14
ally simultaneous innovation across all technology roles will occur. In 2006, Linksys demon-
15
strated routers (infrastructure) and Internal cards (components) that operate on the emerging
16
802.11n standard (Garcia 2006) and Dell Computer shipped an 802.11n laptop client (product)
17
using Broadcom chipsets (Corner 2006). As of mid-2007, there has not yet been widespread
18
adoption of 802.11n-ready infrastructure capabilities and clients, and sentiments in the current
19
marketplace suggest some lag in next-generation wireless product technologies, as infrastructure
20
capabilities catch up and consumers become aware of the benefits (Thornycroft 2007).
21
6.4 Comparison of Digital Music and Wireless Networking Technologies
22
The qualitative example of digital music evolution and the quantitative analysis of the wireless
23
network technologies demonstrated that different patterns of technological change can occur in
31
1
different ecosystems. The difference in the digital music and Wi-Fi evolutionary patterns might
2
be explained in part by the influence of infrastructure technologies as either supporting or ena-
3
bling other technologies within the ecosystem. Specifically, in the digital music ecosystem, in-
4
frastructure technologies typically play a supporting role – they are not required for the use of
5
digital music products but provide additional value (e.g., online digital music stores, FM trans-
6
mitters). In contrast, in the Wi-Fi ecosystem, infrastructure technologies had to be developed
7
first simply to make wireless networking possible for product technologies. Then, as the ecosys-
8
tem developed, new infrastructure technologies, such as wireless printers, supported the product
9
technologies by providing additional value to their use. The existence of two different types of
10
infrastructure roles – supporting and enabling – provides a possible explanation of different evo-
11
lutionary cycles across different ecosystems.
12
There are also other possible explanations for the different patterns in these ecosystems. For
13
example, digital music technologies are typically purchased by individuals while wireless net-
14
working technologies are purchased by both individuals and firms. The social construction of
15
technology view would argue that different social and cultural environments around the use of
16
each technology lead to different patterns of innovation. Similarly, demand-driven theories of
17
innovation would argue that the consumer and market will demand different functionality from
18
the technologies in each of these ecosystems, and therefore their evolution should be different.
19
7. UTILITY EVALUATION OF THE PROPOSED ARTIFACTS
20
The utility of a new artifact must be demonstrated using well-formulated and well-executed
21
evaluation techniques. To demonstrate the utility of our proposed artifacts, we follow Hevner et
22
al. (2004), who suggested seven evaluation methods, two of which are appropriate for the con-
23
text we have studied. The first of these is the observational approach, which is exemplified by
32
1
case study and interviewing methods. We used this approach to evaluate the effectiveness of our
2
proposed artifacts in the business environment. We accomplished this through a number of one-
3
hour structured interviews with IT industry experts. The second is the descriptive approach. We
4
employed the informed argument method, an instance of the descriptive approach, by using in-
5
formation from the knowledge base of our research domain to build arguments for the utility of
6
our proposed artifacts. We accomplished this by assessing common techniques that are used in
7
practice for analyzing the IT landscape, and pointing out how our artifacts complement these
8
techniques to improve analysis capabilities and achieve utility.
9
10
11
7.1 Interviews with IT Industry Experts
12
1987, Eisenhardt 1989) and is one of the most important data gathering tools in qualitative re-
13
search (Myers and Newman 2007). We chose to use interviews as our primary technique for
14
evaluating the utility of our proposed artifacts for two reasons. First, our proposed artifacts are
15
novel and not directly comparable to existing IT landscape analysis techniques on any specific
16
quantitative performance measures, so we must rely on qualitative evaluation techniques. Sec-
17
ond, interviews provide a means for capturing extremely rich data and, in this case, can be used
18
to evaluate the potential utility of our proposed artifacts in a business setting by allowing us to
19
capture the informed opinions of IT industry experts.
Conducting interviews is a key technique for performing IS case study research (Benbasat et al.
20
The Interview Process. Our interview approach was based on an interview script that was
21
developed ahead of time and pre-tested to ensure that the questions we asked would be under-
22
stood and properly interpreted, that they would yield the appropriate kinds of insights, and that
23
they would be scoped to encourage open-ended input while they delivered answers that would
24
help us to gauge the utility of the technological artifacts in our research. The interviewer en-
33
1
sured that all questions in the script were covered during the interview; however, related topics
2
of discussion were permitted in order to increase the richness of the information we captured.
3
The interview included an opening, introduction, key questions, and a closing (Myers and New-
4
man 2007). The interview script and additional information about the development and imple-
5
mentation of interviews are provided in an appendix at the end of this paper. The opening was
6
used as a means of introducing the interviewer and interviewee and capturing basic background
7
demographic information about the interviewee. The introduction was used to explain the pur-
8
pose of the interview.
9
The interviewee was then asked questions about: (1) the business and organizational problem
10
of analyzing the IT landscape for technology investment and development decision-making; (2)
11
the utility of our proposed artifacts, based on their strengths and weaknesses in a context of use;
12
and (3) potential improvements that might be appropriate to our proposed artifacts. Between the
13
first two sets of questions, the interviewee was given a three-page handout that summarized our
14
proposed approach. This was the first time the interviewee was introduced to our proposed arti-
15
facts. The interviewer subsequently spent, on average, twenty to thirty minutes explaining and
16
demonstrating the proposed approach, using the handout as a guide. The interviewer answered
17
questions raised by the interviewee during this discussion. The closing of the interview was a
18
simple debriefing and included a request for follow-up interviews, as needed. The interviews
19
took a little less than one hour each.
20
Interview Subjects. We interviewed a set of IT industry experts with participants from four
21
distinct populations: (1) IT industry senior managers and executives, (2) IT industry consultants,
22
(3) IT industry research staff and analysts, and (4) senior academic researchers with expertise on
23
the IT industry. We interviewed a total of twelve experts, three in each group. Each participant
34
1
had well over ten years of experience in the IT industry or academia, and over ten years of ex-
2
perience in a management or decision-making role. All participants evaluated themselves to
3
have a high level of understanding of the landscape of current and past ITs, which validated our
4
participant selection process. The participants hailed from Fortune 500 companies, technology
5
research and government organizations, and well-known research universities.
6
Question Coverage. To evaluate our proposed approach, we created questions in our inter-
7
view script to evoke discussion of the interviewee’s opinions about four key aspects of our arti-
8
facts for analyzing and understanding the IT landscape. (See Table 4.) First, we evaluated the
9
usefulness of the proposed constructs in the ecosystem model. In particular, we asked if the use
10
of the component, product, and infrastructure roles was a useful model for representing tech-
11
nologies within an ecosystem. We also queried their opinions about the use of paths of influence
12
to classify temporal relationships. Second, we evaluated the logic of the qualitative and quanti-
13
tative methodologies for identifying and visualizing trends in the IT landscape. Here our ques-
14
tions directed the discussion to the soundness of the methodologies and the insights produced by
15
following them. Third, we evaluated the utility of the information produced by the proposed
16
qualitative and quantitative methodologies. Here we directed the discussion to the interviewee’s
17
opinions about the usefulness of the information about technology trends to practitioners and or-
18
ganizations involved in the IT investment or development decision-making process. We also
19
captured opinions about the format and understandability of the output graphs and diagrams.
20
Fourth, we captured the interviewee’s opinions about how our methods complement existing ap-
21
proaches for analyzing the IT landscape. The interviewees’ feedback regarding these four issues
22
led us to identify several key dimensions of utility for our proposed artifacts, which we describe
23
in the remainder of Section 7.
35
1
INSERT TABLE 4 ABOUT HERE
2
7.2. Results of the Interviews: Key Dimensions of Utility
3
Responses to the first set of interview questions provided motivation and shaping of the business
4
and organizational problem our proposed artifacts address. Several key insights came out of this
5
part of the interview. The majority of the participants noted that the staff and management of
6
most IT-consuming companies do not have the time or expertise to perform the necessary analy-
7
sis of the IT landscape, and so they must outsource this process to third parties. Additionally,
8
every participant noted the reliance of IT organizations on reports produced by the companies
9
like Gartner, Forrester, and IDC. Multiple participants also noted that current IT investments
10
and partnerships play a significant role in future investments, and often IT investment decisions
11
are outsourced to partners and suppliers. This yielded important information for us on the locus
12
of likely value for our proposed artifacts – not necessarily through direct use by IT managers
13
within a company, but instead through the consultants, partners, and agents with whom they
14
work in the implementation of their strategic, operational, and product-related use of IT.
15
In general, all of the participants found value in our proposed artifacts for evaluating the IT
16
landscape. Four key dimensions about the utility of our proposed artifacts consistently emerged
17
in their opinions regarding the value of our research. In particular, these artifacts support com-
18
plexity reduction, help to structure investment decisions, provide a formal method for quantify-
19
ing technology ecosystem evolution, and support the identification of the locus of value for post-
20
investment evaluation. We discuss each of these in succession, and provide our respondents’
21
reactions to illustrate our arguments about utility.
22
Complexity Reduction. The general consensus of the experts we interviewed was that the
23
use of technology roles and the paths of influence provide a novel and useful way of reducing the
36
1
complexity of the IT landscape while maintaining the important relationships between technolo-
2
gies. Our interviewees’ commented:
3
4
“[The roles and model] are very clever because you compress the universe of possibilities and make the
ecosystem understandable.” -- Senior Technology Analyst at a Fortune 500 transportation company
5
6
7
“This is a very good way to think about the problem ... it explains the technology ecosystem very well and is
nice way of trying to break up very complex phenomena.” -- Managing Director of a Government IT Organization
8
The exercise of defining the technology ecosystem provides two useful insights to the user of
9
the proposed constructs from the points of view of our respondents. First, it forces the analyst to
10
consider interdependencies among technologies and realize the complexity of the technology
11
ecosystem. Second, it provides structure (based on the concepts of technology roles and paths of
12
influence) to reduce this complexity using a system view of the IT landscape and captures the
13
technology ecosystem from the analyst’s point of view. Each of these aspects enhances the us-
14
er’s ability to understand the nature of relationships in the IT landscape.
15
Structure for IT Investment Decision-Making. The interviewees also reported:
16
17
“This approach provides structure to the conversation and decision-making process for IT investment." -Director of Business Development at a major university IT-related research center
18
19
20
21
“This [approach] brings the ability to work on [the IT investment decision-making problem] interdisciplinarily within an organization. You could present this to the CEO, engineering guys, marketing guys, and
they would all know what you were talking about. They may ask different questions, but they would all find
it useful.” -- Former VP of a Fortune 500 technology company, current IT industry private consultant
22
Through the interviews we discovered evidence of a lack of structure in how firms go about ana-
23
lyzing the IT landscape. They typically rely on third-party reports and advice from suppliers and
24
partners, as we noted earlier, but this apparently is still not sufficient. Our proposed methodol-
25
ogy provides a means for generating representations of the IT landscape and associated technol-
26
ogy trends that are relevant to the firm’s interests and business contexts. A majority of the par-
27
ticipants also noted that the proposed approach is a useful tool for decision-makers across differ-
28
ent functional roles in the organization. The participants felt that senior managers and strategic
37
1
planners, as well as technical managers and engineers, could all benefit from understanding the
2
IT landscape and technology trends in terms of the proposed technology ecosystem model. In
3
general, the consensus of the participants was that the proposed methods should be useful in the
4
IT investment and development decision-making process.
5
Formal Method to Quantify Technological Change. The proposed methods provide a for-
6
mal technique for quantifying trends in technological change within the IT landscape. Our inter-
7
viewees noted that the techniques most commonly used by firms to analyze the IT landscape are
8
informal and ad hoc:
9
10
11
12
"Most work on this problem is informal and this is one of only a few formal approaches I have seen. Attempts to formally quantify things are a good thing. This is a formal methodology to add some quantification to the analysis by [companies like Gartner]." -- Senior Consultant at a Fortune 500 technology company
13
14
“The systematic approach this provides is useful. […] Most strategic IT decisions are made using less formal types of analysis.” -- Senior Technology Analyst at a Fortune 500 transportation company
15
It turns out that only few firms – including those producing industry reports – use formal
16
quantitative means to produce technology forecasts, aside from simple linear extrapolation. All
17
interviewees found the quantification aspect of the proposed methods to be useful, and most also
18
commented that our formal approach will complement existing techniques well and will provide
19
firms with new and useful information for making IT investment and development decisions.
20
Locus of Value for the Artifacts. We also learned where the value of our proposed artifacts
21
will be the highest, which is another important aspect of their utility. This is similar to the locus
22
of value construct (Kauffman and Weill, 1989), which describes where value flows are most like-
23
ly to occur. A consultant and a senior manager offered the following comments:
24
25
26
27
“Companies that can benefit most from this are the technology producers, like for example IBM, Microsoft,
and Sun. These are the ones defining the future technologies. By looking at a systematic way of how technology got to where it is today it may help [technology producers] determine what types of technologies
are needed.” -- Senior Consultant at a Fortune 500 technology company
28
29
“For analysis purposes, this sort of model is very good and should definitely help analysts at Gartner or
Forrester produce reports for managers.” -- IT Manager at a Fortune 500 retail company
38
1
Based on interviews, an interesting finding for us was that, in terms of the locus of value, the
2
interview participants differentiated between the utilities of different artifacts: (1) the utility of
3
the proposed model, constructs, and the information produced by our methods (i.e., resulting
4
graphs and diagrams of specific ecosystems), and (2) the utility of methodologies themselves for
5
conducting IT landscape analysis and producing various patterns of technology evolution. The
6
participants indicated that the information produced by our proposed methods would be useful to
7
decision-makers in both IT-consuming and IT-producing firms. On the other hand, the majority
8
of the interview participants felt that using the proposed approach to actually conduct the analy-
9
sis of the IT landscape would be most beneficial to firms that either produce IT or produce the
10
industry reports on trends in IT. Understanding the trends in technology evolution that led to the
11
current state of the IT landscape should prove vital in determining what directions IT develop-
12
ment initiatives should follow in the future. Furthermore, the reality of IT landscape analysis is
13
that IS and corporate strategy staff members at most IT-consuming firms have neither the time,
14
the resources, nor the technology and market knowledge to conduct formal analyses. So, even if
15
the techniques for analyzing the landscape improve, IT-consuming firms will still likely rely on
16
third parties to conduct their technology assessments and analyses for them. As a result, new
17
formal approaches for analyzing the IT landscape, such as what we propose, should add value to
18
both the firms producing the forecasts and the firms consuming them.
19
Through our interviews we also captured recommendations and suggestions for enhancing
20
the proposed artifacts. We review the most important suggestions in Section 8, where we elabo-
21
rate on the limitations and future directions for this research.
22
7.3. The Fifth Dimension: The Complementary Value of the Proposed Artifacts
23
24
“This [approach] should be very useful for helping educate analysts about the [IT] landscape. It is complementary to other existing approaches.” -- Senior Researcher at a Fortune 500 technology company.
39
1
An additional aspect of utility that was suggested by most of the interview participants was that
2
our proposed methods will complement well existing techniques for analyzing the IT landscape.
3
To delve deeper into the potential complementarities, we evaluated the strengths and weaknesses
4
of many common approaches for technology forecasting and IT landscape analysis and discuss
5
how our approach specifically complements each of them.
6
Table 5 provides an outline of common technology forecasting and planning techniques used
7
in industry, including trend analysis, expert opinion, modeling and simulation, and scenario
8
analysis. Although specific methods are most often proprietary, firms, such as Gartner, use some
9
version and/or combination of these techniques to generate their IT forecasts and reports. The
10
constructs and methods presented in this article are intended to complement firms’ proprietary
11
approaches, and provide an additional means for viewing technology evolution. Our approach
12
utilizes elements of trend analysis and modeling techniques, enables mapping of the historical
13
relationships among technologies, and can provide useful insights regarding the next possible
14
evolutionary steps within an ecosystem. In particular, at a given point in time, our approach may
15
complement existing technology forecasting methods (as noted in the last row of Table 5) by
16
providing structured input and formal analysis of the past and current states of the IT landscape.
17
18
INSERT TABLE 5 ABOUT HERE
Another relevant industry analysis approach is a technology roadmap. A technology road-
19
map is a tool that is typically used for planning purposes, such as in product, strategic, ser-
20
vice/capability, and process planning (Kostoff and Schaller 2001, Rinne 2004, Phaal et al. 2004).
21
Technology roadmaps provide a way to identify, evaluate, and select strategic alternatives by
22
mapping structural and temporal relationships among R&D, technologies, potential products, and
23
markets (Kostoff and Schaller 2001, Rinne 2004). The process of generating a technology
40
1
roadmap follows a visual mapping strategy not unlike we have presented. Our methods com-
2
plement technology roadmaps by providing a problem representation vocabulary that extends
3
current road-mapping techniques, and provides a more formal quantitative method for identify-
4
ing trends in technological change. Technology roadmaps are designed to outline the set of pos-
5
sible future strategies for a specific firm (Kostoff and Schaller 2001), and our technique adds to
6
this by providing a method for historically evaluating the evolution of an entire set of interrelated
7
technologies (represented by technology roles within the ecosystem) using similar visual repre-
8
sentation techniques.
9
As noted in Table 5, all technology forecasting methods have inherent assumptions, and the
10
accuracy of these assumptions influences the predictive accuracy of the forecast. Most of these
11
methods, with the exception of some regression and econometric approaches, are not predictive
12
in the sense of classical variance theory (Mohr 1982), where a set of predictor variables is used
13
to predict the level of some outcome variable. The basic assumption in all forecasting tech-
14
niques, including the ones that a variance theory would suggest, is that the historical trends and
15
patterns will continue into the future following the same dynamics. Our approach follows the
16
same basic assumption: if technological change will continue to occur following the same pat-
17
terns identified using our methods, then we can make reasonable forecasts about the future.
18
8. CONCLUSION, LIMITATIONS, AND FUTURE WORK
19
Following the design science research paradigm, the major contribution of this research is the
20
development of a new set of artifacts for: (1) codifying technological innovations based on the
21
role they play within an ecosystem of interrelated technologies, (2) identifying dominant tech-
22
nology types within an ecosystem using real-world data, and (3) visually representing patterns of
23
technological change over time based on dominant technological roles. These artifacts comprise
41
1
a new technique for analyzing the IT landscape to inform the IT investment and development
2
decision-making process. This article contributes to the IS research field in several additional
3
ways. We provide a review of relevant IS and organizational science research on technology
4
evolution and construct a theoretical perspective that integrates and builds upon previous ideas.
5
We incorporate strategies for sensemaking of complex data (e.g., quantification and visual map-
6
ping) from process theory. In addition, we provide insightful analysis on the evolution of tech-
7
nologies in the digital music and wireless networking ecosystems. We employ interviews with
8
senior technology managers, consultants, and industry researchers as a means to demonstrate
9
utility of our proposed artifacts, and we also review existing technology forecasting methods and
10
theoretical perspectives on technological change.
11
To develop the methodologies presented in this article, we used a combination of the visual
12
mapping and quantitative strategies for sensemaking from process data (Langley 1999). These
13
strategies do have their limitations. Process mapping and visual representations may exclude
14
some dimensions of data ambiguity, and graphical forms may be biased toward the representa-
15
tion of certain types of information over others. The conclusions derived may sometimes have
16
rather mechanical qualities since these representations deal more with the surface structure of
17
activity sequences than the underlying forces (Langley 1999). On the other hand, since the goal
18
of quantification strategies is to reduce complexity, their use may sometimes lead to a loss of
19
richness in the process data (Langley 1999). A common criticism of the process theory in gen-
20
eral is that, although it can produce enriched understanding and explanation, it often lacks pre-
21
dictive power (Van de Ven 1992).
22
23
To address the limitations of process theory, Langley (1999) suggests that both the quantitative and visual mapping strategies should be used in combination with other approaches. We
42
1
chose to use these two in combination because they complement each other well. Visual map-
2
ping provides additional contextual information that may be lost in quantification, while quanti-
3
fication provides an opportunity to apply empirical rigor that is missing in visualization. Fur-
4
thermore, we demonstrated the use of our model’s constructs in both qualitative (digital music
5
technologies) and quantitative (Wi-Fi technologies) investigations. The constructs were used
6
successfully to identify evolutionary patterns in both scenarios thus suggesting their usefulness.
7
All models are abstractions of the real world and, therefore, depend on the assumptions used
8
in their construction. In this research we relied on the assumption, based on our synthesis of
9
prior literature and our observations of the real world, that the common roles technologies play in
10
an ecosystem are components, products, and infrastructure. Another choice we make in this
11
model is to follow the technological determinism approach (Smith and Marx 1994) and currently
12
exclude the role of external forces such as market dynamics, the demand environment, society,
13
and culture. Multiple interview participants recommended expanding our model in future re-
14
search to include more external forces and context-specific factors. Excluding these factors may
15
result in a loss of contextual richness; however, by limiting the number of factors considered in
16
the model we gain control and specificity, and reduce complexity for the user of our methods.
17
An objective of our approach is to demonstrate that the patterns of technological change can
18
be identified using a model based solely on relationships between technologies. We do recog-
19
nize, however, the potential contribution of including these external factors in the analysis of the
20
IT landscape. We plan to expand the current model and methodology to include more aspects of
21
the external environment, such as market demand and social forces. Combining the technologi-
22
cal determinism view with the social constructivism view will likely provide many new insights
23
and opportunities for further improving techniques for analyzing the IT environment.
43
1
Although an important goal was to develop general constructs and methods for mapping out
2
evolutionary patterns in technology ecosystems, our research provided insights about possible
3
specific innovation and evolution drivers in specific technology ecosystems. We point to some
4
of these in our quantitative and qualitative analyses in Sections 5 and 6 (e.g., the possible differ-
5
ence in evolutionary trends when you have supporting vs. enabling types of infrastructure). Us-
6
ing our ecosystem-based approach as a basis, in future research we will analyze ecosystem-
7
specific drivers of innovation within several IT ecosystems and explore their similarities and dif-
8
ferences.
9
The interview participants who evaluated our proposed artifacts provided three additional
10
important recommendations and comments for expansions to the current research. First, there
11
was a suggestion that exploring the directionality of paths of influence may provide interesting
12
insights on technological development. In the current model, we looked primarily at positive
13
relationships – an innovation of a technology in one role provides an opportunity for the devel-
14
opment of a new technology in another role. There may be negative relationships between tech-
15
nologies though. For example, a new technology may make a series of existing technologies ob-
16
solete, thus effectively exterminating a portion of the ecosystem. Considering the directionality
17
of the relationships between technologies also reinforces the ecological analogy in which both
18
birth and death processes occur.
19
Second, the time scale may be used more effectively in the quantitative analysis to identify
20
lags in transitions between patterns in technology trends. Quantifying such lags may provide a
21
predictive tool for forecasting the occurrence of future trends. Third, two of the interview par-
22
ticipants noted that they would expect data collection for performing the proposed analysis to be
23
difficult for many firms. We recognize the importance of this comment and note that many
44
1
technology services companies are investing significantly in new business intelligence tools for
2
extracting quantifiable data from the seas of information available on and off the Internet. As
3
these tools evolve, rich data on the IT landscape should become more readily available, and we
4
plan to investigate opportunities for integrating our proposed approach with these tools.
5
References
6
7
Adner, R. and Levinthal, D. “Demand heterogeneity and technology evolution: Implications for
product and process innovation,” Management Science, (47:5), May 2001, pp. 611-628.
8
9
10
Adomavicius, G., Bockstedt, J., Gupta, A., and Kauffman, R.J. “Technology roles and paths of
influence in an ecosystem model of technology evolution,” Information Technology and Management, (8:2), April 2007, pp. 185-202.
11
12
Arthur, W.B. “Competing technologies, increasing returns, and lock-in by historical events,”
Economic Journal (99:394), March 1989, pp. 116-131.
13
14
Bacon, C.J. “The use of decision criteria in selecting information systems/Technology investments,” MIS Quarterly (16:3) September 1992, pp. 335-353.
15
16
Baldwin, C.Y. and Clark, K.B. “Managing in the age of modularity,” Harvard Business Review
(75:5), September 1997, pp. 84-93.
17
18
Baldwin, C.Y. and Clark, K.B. Design Rules: The Power of Modularity, The MIT Press, Cambridge, MA, 2000.
19
20
Benbasat, I., Goldstein, D.K., and Mead, M. “The case research strategy in studies of information systems,” MIS Quarterly (11:3) September 1987, pp. 369-386.
21
22
Benbasat. I. and Zmud, R.W. “The identity crisis within the IS discipline: Defining and communicating the discipline’s core properties”, MIS Quarterly, (27:2), June 2003, pp. 183-194.
23
24
25
Boland, R.J., Lyytinen, K. and Yoo, Y. “Path creation with digital 3D representation: Networks
of innovation in architectural design and construction,” Sprouts: Working Papers on Information Environments, Systems and Organizations, (3:1), Winter 2003, pp 1-28.
26
27
28
Campbell, D.T. “Levels of organization, downward causation, and the selection-theory approach
to evolutionary epistemology,” in Theories of the Evolution of Knowing, G. Greenberg and E.
Tobach (eds.), Lawrence Erlbaum, Hillsdale, NJ, 1990, pp. 1-17.
29
30
31
Campbell, D.T. “How individual and face-to-face group selection undermine firm selection in
organizational evolution,” in Evolutionary Dynamics in Organizations, J.A.C. Baum and J.V.
Singh (eds.),Oxford University Press, New York, NY, 1994, pp. 23-38.
32
33
Clark, K.B. “The interaction of design hierarchies and market concepts in technological evolution,” Research Policy, (14:5), October 1985, pp. 235-251.
34
35
Clemons, E.K and Weber, W. “Strategic Information Technology Investments: Guidelines for
Decision Makers,” Journal of Management Information Systems (7:2) Fall 1990, pp. 9-28
45
1
2
Corner, S. “Dell puts next gen WiFi, 802.11n, into laptops,” ITWire, July 18, 2006, available at
www.itwire.com.au/content/view/4995/127/ (last accessed September 17, 2007).
3
4
Dosi, G. “Technological paradigms and technological trajectories,” Research Policy, (11:3), June
1982, pp. 147-162.
5
6
Eisenhardt, K.M. “Building theories from case study research,” The Academy of Management
Review, (14:4) October 1989, pp. 532-550.
7
8
9
Ewing, J. and Burrows, P. “Will Apple answer Sony’s call?” Business Week, July 31, 2006,
available online at www.businessweek.com/magazine/content/06_31/c3995016.htm (last accessed March 31, 2007).
10
11
Funk, J. L. “Technological change within hierarchies: The case of the music industry,” Economics of Innovation and New Technology, (16:1), January 2007, pp. 1-16.
12
13
Garcia, A. “First 802.11n products show standard’s promise,” eWeek, April 24, 2006, available
at www.eweek.com/article2/0,1895,1952659,00.asp (last accessed September 17, 2007).
14
15
Garud, R. and Van de Ven, A.H. “An empirical evaluation of the internal corporate venturing
process,” Strategic Management Journal, (13: Special Issue), Summer 1992, pp. 93-109.
16
17
Gregor, S. “The nature of theory in information systems,” MIS Quarterly, (30:3), September
2006, pp. 611-642.
18
19
Hannan, M.T., and Freeman, J. “The population ecology of organizations,” American Journal of
Sociology, (82), March 1977, pp. 929-964.
20
21
Hannan, M.T. and Freeman, J. Organizational Ecology, Harvard University Press, Cambridge,
MA, 1989.
22
23
24
Harte, J., Shireman, B., Burke, A., and Scarlett, L. “Business as a living system: The value of
industrial ecology (a roundtable discussion),” California Management Review, (43:3), Spring
2001, pp. 16-25.
25
26
Hevner, A.R., March, S.T., Park, J., and Ram, S. “Design science in information systems research,” MIS Quarterly, (28:1), March 2004, pp. 75-105.
27
28
29
Iansiti, M. and Levien, R. “The new operational dynamics of business ecosystems: Implications
for policy, operations and technology strategy,” Working paper 03-030, Harvard Business
School, Cambridge, MA, 2002.
30
31
32
Iansiti, M. and Levien, R. The Keystone Advantage: What New Dynamics of Business Ecosystems Mean for Strategy, Innovation, and Sustainability, Harvard Business School Press, Boston, MA, 2004.
33
34
35
36
Kauffman, R.J., and Weill, P. “An evaluative framework for research on the performance effects
of information technology investment,” in Proceedings of the 10th International Conference on
Information Systems, J.I. Degross, J.C. Henderson and B.R. Konsynski (eds.), Boston, MA,
December 4-6, 1989, pp. 377-388
37
38
Kostoff, R.N. and Schaller, R.R. “Science and technology roadmaps,” IEEE Transactions on
Engineering Management, (48:2), May 2001, pp. 132-143.
39
40
Langley, A. “Strategies of theorizing from process data,” Academy of Management Review,
(24:4), October 1999, pp. 691-710
46
1
2
Langley, A. and Truax, J. “A process study of new technology adoption in smaller manufacturing firms,” Journal of Management Studies, (31:5), September 1994, pp. 619-652.
3
4
Laudan, R. (ed.) Nature of Technological Knowledge. Are Models of Scientific Change Relevant? Reidel, Dordrecht, Netherlands, 1984.
5
6
7
Lyytinen, K. and Rose, G.M. “The disruptive nature of information technology innovations: The
case of Internet computing in systems development organizations,” MIS Quarterly, (27:4), December 2003, pp. 557-595.
8
9
10
Malerba, F., Nelson, R., Orsenigo, L., and Winter, S. “History-friendly models of industry evolution: The case of computer industries,” Industrial and Corporate Change, (8:1), March 1999,
pp. 1-36.
11
12
March, S.T and Smith, G.F. “Design and natural science research on information technology,”
Decision Support Systems, (15:4), December 1995, pp. 251-266.
13
14
Markus, L. and Robey, D. “Information technology and organizational change: Causal structure
in theory and research,” Management Science, (34:5), May 1988, pp. 583-598.
15
16
Meyer, A.D. “Visual data in organizational research,” Organization Science, (2:2), May 1991,
pp. 218-236.
17
18
Meyer, A.D. and Goes, J. “Organizational assimilation of innovations: A multilevel contextual
analysis,” Academy of Management Journal, (31:4), December 1988, pp. 897-923.
19
Miles, M.B. and Huberman, A.M. Qualitative data analysis, Sage, Newbury Park, CA, 1994.
20
21
Millet, S.M and Honton, E.J. A Manager’s Guide to Technology Forecasting and Strategy Analysis Methods, Battelle Press, Columbus, OH, 1991.
22
23
Mintzberg, H., Raisinghani, D., and Theoret, A. “The structure of unstructured decision processes,” Administrative Science Quarterly, (21:2), June 1976, 246-275.
24
Mohr, L.B. Explaining Organizational Behavior, Jossey-Bass, San Francisco, CA, 1982.
25
26
Myers, M.D., and Newman, M. “The qualitative interview in IS research: Examining the craft,”
Information and Organization, (17:1) January 2007, pp. 2-26.
27
28
Nelson, R.R. “Recent evolutionary theorizing about economic change,” Journal of Economic
Literature, (33:1), March 1995, pp. 48-90.
29
30
Nelson, R.R. and Winter, S.G. An Evolutionary Theory of Economic Change, Belknap Press of
Harvard University Press, Cambridge, MA, 1982.
31
32
Newman, M. and Robey, D. “A social process model of user-analyst relationships,” MIS Quarterly, (16:2) June 1992, pp. 249-266.
33
34
Nickerson, J.V. and zur Muehlen, M. “The ecology of standards processes: Insights from Interenet standard making,” MIS Quarterly, (30:Special Issue), August 2006, pp. 467-488.
35
36
Orlikowski, W.J. “CASE tools as organizational change: Investigating incremental and radical
changes in systems development,” MIS Quarterly, (17:3), September 1993, pp. 309-340.
37
38
Orlikowski, W.J. and Robey, D. “Information technology and the structuring of organizations,”
Information Systems Research, (2:2), June 1991, pp. 143-169.
47
1
2
Pentland, B.T. “Grammatical models of organizational processes,” Organization Science, (6:5),
September 1995, pp. 541-556.
3
4
5
Phaal, R., Farrukh, C.J.P., and Probert, D.R. “Technology roadmapping – A planning framework
for evolution and revolution,” Technological Forecasting and Social Change, (71:1), January
2004, pp. 5-26.
6
7
Porter, A. L., Roper, A. T., Mason, T. W., Rossini, F. A., and Banks, J. Forecasting and Management of Technology, Wiley-Interscience, New York, NY, 1991.
8
9
10
Quaadgras, A. 2005. “Who joins the platform? The case of the RFID business ecosystem,” in
Proceedings of the 38th Hawaii International Conference on System Sciences, R. H. Sprague
(ed.), Big Island, HI. January 3-6, 2005, IEEE Computing Society Press, Los Alamitos, CA.
11
12
Rinne, M. “Technology roadmaps: Infrastructure for innovation,” Technological Forecasting
and Social Change, (71:1), January 2004, pp. 67-80.
13
14
Romanelli, E. and Tushman M. “Organizational transformation as punctuated equilibrium,”
Academy of Management Journal, (37:5), October 1994, pp. 1141-1166.
15
16
Rosenberg, N. Exploring the Black Box: Technology, Economics, History, Cambridge University
Press, London, UK, 1994.
17
18
19
20
Rosenkopf, L. and Nerkar, A. “On the complexity of technological evolution: Exploring coevolution within and across hierarchical levels in optical disk technology,” in Variations in Organization Science: In Honor of Donald T. Campbell, J.A.C Baum, and B. McKelvey (eds.),
Sage Publications, Thousand Oaks, CA, 1999, pp. 169-183.
21
Sahal, D. Patterns of Technological Innovation. Addison-Wesley, Reading, MA, 1981.
22
23
Sahal, D. “Technological guideposts and innovation avenues,” Research Policy, (14:2), April
1985, pp. 61-82.
24
25
Saviotti, P.P. Technology Evolution, Variety and the Economy, Edward Elgar Publishing, Cheltenham, UK, 1996.
26
27
Saviotti, P.P., and Metcalfe, J.S. “A theoretical approach to construction of technological output
indicators,” Research Policy, (13:3), June 1984, pp. 141-51.
28
29
Shillingford, J. “Sony Ericsson launches Walkmans,” BBC News, February 14, 2005, available at
news.bbc.co.uk/1/hi/business/4265777.stm (last accessed September 17, 2007).
30
31
Smith, M.R. and Marx, L. (eds.) Does Technology Drive History? The Dilemma of Technological Determinism, MIT Press, Cambridge, MA, 1994.
32
33
Sood, A. and Tellis, G.J. “Technological evolution and radical innovation,” Journal of Marketing
(69:3), July 2005, pp. 152-168.
34
35
36
Thornycroft, P. “Designing for speed: Preparing for the 802.11n adoption wave,” The Edge,
Aruba Networks, Sunnyvale, CA, August 6, 2007, available at edge.arubanetworks.com/ article/designed-speed-preparing-802-11n-adoption-wave (last accessed on September 17, 2007).
37
38
39
Tushman, M.L. and Rosenkopf, L. “On the organizational determinants of technological change:
Toward a sociology of technological evolution,” in Research in Organizational Behavior, B.
Staw and L. Cummings (eds.), JAI Press, Greenwich, CT, 1992, pp. 311-347.
48
1
2
3
Van Buskirk, E. “Bragging rights to the world’s first MP3 player,” CNET News, available at
news.com.com/2010-1041_3-5548180.html (last accessed September 17, 2007), January 15,
2005.
4
5
Van de Ven, A.H. “Suggestions for studying strategy process: A research note,” Strategic Management Journal, (13:Special Issue), Summer 1992, pp. 169-188.
6
7
Van de Ven, A.H. and Huber, G.P. “Longitudinal field research methods for studying processes
of organizational change,” Organization Science, (1:3), August 1990, pp. 213-219.
8
9
Van de Ven, A.H. and Poole, M.S. “Methods for studying innovation development in the Minnesota Innovation Research Program,” Organization Science, (1:3), August 1990, pp. 313-335.
10
Weick, K. The Social Psychology of Organizing, Addison-Wesley, Reading, MA, 1979.
11
12
13
Whinston, A.B. and Geng, X. “Operationalizing the essential role of the information technology
artifact in information systems research: Gray areas, pitfalls, and the importance of strategic
ambiguity,” MIS Quarterly, (28:2), June 2004, pp. 149-159.
14
15
Wired News. “Sub Pop enters MP3 domain,” February 24, 1999, available at www.wired.com/
news/culture/0,1284,18107,00.html (last accessed September 17, 2007).
16
17
18
Worlton, J. “Some patterns of technological change in high performance computers,” in Proceedings of Supercomputing 1988, IEEE Computing Society Press, Orlando, FL, November
14-18, 1998, pp. 312-320.
49
1
Table 1. Paths of Influence in a Technology Ecosystem
COMPONENT
PRESENT STATE
(C)
PRODUCT
PRESENT STATE
(P)
INFRASTRUCTURE
PRESENT STATE
(I)
COMPONENT
FUTURE STATE (C*)
Component Evolution
Examples: microprocessors
and Moore's Law, digital camera mega- pixels.
Product-Driven Component
Development
Examples: Blue-ray DVD,
digital encryption technology
PRODUCT
FUTURE STATE (P*)
Design and Compilation
Examples: digital camera,
MP3 players, PCs
INFRASTRUCTURE
FUTURE STATE (I*)
Standards and Infrastructure Development
Examples: XML, RFID
Product Integration and
Evolution
Examples: camera phones,
Wi-Fi-enabled PDAs
Diffusion and Adoption
Examples: digital camera
infrastructure, software
applications designed for
Windows OS
Support Evolution
Examples: growth of mobile cellular phone network, Internet 2.0
Infrastructure-Driven
Infrastructure-Leveraging
Component Development
Product Development
Examples: Internet technologies, Examples: instant messaging
802.11g Wi-Fi equipment
services, picture mail
2
3
Table 2. Timeline of Digital Music Technologies
YEAR
1989
1996
1998
Feb 1999
May 1999
May 2000
Jan 2001
Jul 2001
Oct 2001
Mar 2002
Apr 2003
Oct 2003
Sep 2004
May 2005
Oct 2005
Sep 2006
Aug 2006
July 2007
Sep 2007
EVENT
German MP3 patent
US MP3 patent
First portable MP3 player (32 MB)
Sub Pop distributes MP3 music
Napster founded
Transactional watermarking developed
Apple iTunes music applications released
Napster injunction
10 GB Apple iPod introduced
20 GB iPod for PC introduced
40 GB iPod introduced
Dell DJ introduced
iTunes online music store opens
MSN online music store opens
Yahoo online music store opens
First iPod with video capabilities
iTunes starts selling full length movies
160 GB 1.8 inch HDD introduced
iPhone (MP3 player/phone) introduced
160 GB Video iPod introduced
4
50
1
Table 3. Empirical Methodology Overview
STEP
1. Coding
2. Frequency
Estimation
3. Threshold
Determination
4. Dominant Role
Identification
5. Pattern
Identification
DESCRIPTION
Raw data on technology introductions is coded into the component, product,
and infrastructure roles.
A function of the frequency of all technology introductions over time is estimated.
Based on the proportion of technologies of each role within the overall number of technology introductions, a threshold function is derived for each role.
Actual frequency of technology introductions is compared to threshold function for each role to determine dominant roles in each time period.
Transitions between dominant roles in adjacent time periods are mapped out.
2
3
4
Table 4. Coverage of Interview Questions: Key Issues for Evaluation
ISSUE
Constructs
Logic of
Methodology
Information
Produced
Relationship to
Existing Techniques
DESCRIPTION
Is the ecosystem model, with its technology roles and paths of influence, a useful
approach for representing the evolution of ITs? Does this representation improve managerial capabilities for analysis? Does it aid the processes of IT investment and development decision-making?
Are the qualitative and quantitative methodologies we propose for identifying
trends in technology evolution sound? Do they produce new insights?
Is the information produced by the proposed artifacts useful to practitioners?
Does it aid in IT investment and development decision-making?
Do the proposed artifacts complement existing techniques to provide new insights and improved analysis of IT evolution?
5
51
1
2
Table 5. Overview of the Traditional Technology Forecasting/Modeling Methods and How
Our Approach Complements Them
3
TREND
ANALYSIS
EXPERT
OPINION
Description
Historical trends are
extended into the future
using mathematical and
statistical techniques.
Domain expert opinions
are collected and analyzed.
Examples
Extrapolation, time
series estimation, regression and econometrics, S-curve estimation
Past trends will continue into future.
Delphi method, interviews, questionnaires, idea
generation
Assumptions
Strengths
Quantifiable and databased forecasts, shortterm accuracy
Weaknesses
Requires a significant
amount of data, which
can be difficult to obtain. Can be inaccurate
for long time horizons.
Experts know significantly
more about a domain than
others. Group opinions are
better than individual
opinions.
Experts typically possess
detailed knowledge of
subject matter that produces high-quality forecasts.
Difficult to identify experts. Knowledge is typically implicit (internalized). Group forecasts
may be affected by social
and psychological factors.
A formal quantitative
approach and a representation of the past and current
IT landscape that can
structure discussion
among experts.
MODELING AND
SIMULATION
Simplified representation
of structure and dynamics
of real world created to
forecast or simulate future
outcomes.
Cross-impact analysis,
system dynamics analysis,
path and tree analysis
SCENARIOS
Plausible set of outcomes
for some aspect of future
is created and analyzed.
Complex structures and
processes can be captured
effectively by simplified
models.
Descriptive vs. normative
scenarios, baseline vs.
optimistic vs. pessimistic
scenarios
Imaginative descriptions
can reasonably capture
the full set of future possibilities.
Models reduce complexity
and highlight the most
important factors. Process
of building a model can
provide insights.
Models often ignore qualitative and contextual factors.
Effective way to communicate forecasts. Incorporate a wide range of
qualitative and quantitative data.
Can be highly speculative
and not firmly based in
reality.
A view of relationships
A representation of the
A formal representation
Our Apbetween multiple techstructure of the IT landof the past and present
proach
nologies
that
complescape
that
can
inform
the
ecosystem which can be
Complements and informs indevelopment of a more
used as quantifiable input
ments This
depth analysis of a sinrealistic simulation.
for generating scenarios.
by Method gle technology attribute.
Providing:
Note: Based on the technology forecasting methods discussed in Porter et al. (1991) and Millet and Honton (1991).
4
5
52
1
Figure 1. Example Patterns of Technological Change
C*
2
P*
C*
I*
P*
I*
P*
I*
C
P
I
C
P
I
a) Product Development
b) Product and Infra.
Alignment
c) Feed-forward
C*
P*
I*
C
P
I
C*
P*
I*
C
P
I
e) Incremental
d) Feed-back
3
4
C*
C
P
I
Figure 2. Transformation between Different State Diagram Representations
C*
C
P
I
P*
C
C
P
P
I
I
I*
5
6
53
1
2
Figure 3. Digital Music Technology Evolution Cycles
3
4
5
Figure 4. Digital Music Technology Graph-Based State Diagram
6
7
54
1
2
Figure 5. Estimating the Monthly Frequency of 802.11b Technology Certifications Using a
Polynomial Approximation Function
45
40
Actual Frequency
Polynomial Estimation
Certifications
35
30
25
20
15
10
5
3
4
5
6
6
Se
p0
-0
6
M
ar
5
Se
p0
-0
5
M
ar
4
Se
p0
-0
4
M
ar
3
Se
p0
-0
3
M
ar
2
Se
p0
-0
2
M
ar
1
Se
p0
-0
1
M
ar
0
Se
p0
M
ar
-0
0
0
Figure 6. Proportional Frequency Functions for the Total Number of Certifications and
Each Technology Role
35
Total
Components
Products
Infrastructure
Certifications
30
25
20
15
10
5
6
p0
Se
-0
6
M
ar
5
p0
Se
-0
5
M
ar
4
p0
Se
-0
4
M
ar
3
p0
Se
-0
3
M
ar
2
p0
Se
-0
2
M
ar
1
p0
Se
-0
1
M
ar
0
p0
Se
7
M
ar
-0
0
0
55
Certifications
4
-0
0
p0
M 0
ar
-0
Se 1
p0
M 1
ar
-0
Se 2
p0
M 2
ar
-0
Se 3
p0
M 3
ar
-0
Se 4
p0
M 4
ar
-0
Se 5
p0
M 5
ar
-0
Se 6
p06
Se
M
ar
Certifications
M
ar
-0
Se 0
p00
M
ar
-0
Se 1
p01
M
ar
-0
Se 2
p02
M
ar
-0
Se 3
p03
M
ar
-0
Se 4
p04
M
ar
-0
Se 5
p05
M
ar
-0
Se 6
p06
2
M
ar
-0
Se 0
p0
M 0
ar
-0
Se 1
p0
M 1
ar
-0
Se 2
p0
M 2
ar
-0
Se 3
p0
M 3
ar
-0
Se 4
p0
M 4
ar
-0
Se 5
p05
M
ar
-0
Se 6
p06
Certifications
1
Figure 7. Actual Frequencies of 802.11b Technologies Plotted against Threshold Functions
45
40
802.11b Components
35
Threshold
5-Month Moving Average
Actual Frequency
30
25
20
15
10
5
0
8
7
802.11b Products
6
5
4
3
2
1
0
3
25
802.11b Infrastructure
20
15
10
5
0
56
4
ec
-0
M 2
ar
Ju 03
n0
Se 3
pD 03
ec
-0
M 3
ar
Ju 04
nSe 04
pD 04
ec
-0
M 4
ar
Ju 05
nSe 05
pD 05
ec
-0
M 5
ar
Ju 06
nSe 06
pD 06
ec
-0
6
D
Certifications
ec
-0
M 2
ar
-0
Ju 3
nSe 03
pD 03
ec
-0
M 3
ar
-0
Ju 4
nSe 04
pD 04
ec
-0
M 4
ar
Ju 05
nSe 05
pD 05
ec
-0
M 5
ar
Ju 06
nSe 06
pD 06
ec
-0
6
D
Certifications
ec
-0
M 2
ar
-0
Ju 3
nSe 03
pD 03
ec
-0
M 3
ar
-0
Ju 4
nSe 04
pD 04
ec
-0
M 4
ar
-0
Ju 5
nSe 05
pD 05
ec
-0
M 5
ar
-0
Ju 6
nSe 06
pD 06
ec
-0
6
D
Certifications
1
Figure 8. Actual Frequencies of 802.11g Technologies Plotted against Threshold Functions
60
50
802.11g Components
45
Threshold
5-Month Moving Average
Actual Frequency
40
30
20
10
0
2
25
802.11g Products
20
15
10
5
0
3
50
802.11g Infrastructure
40
35
30
25
20
15
10
5
0
57
1
2
Figure 9. State Diagram for 802.11b and 802.11g Generations
(6-month periods, starting March 2000)
3
4
5
6
Figure 10. State Diagram for the WPA 1 and WPA 2 Generations
(6-month periods, starting March 2000)
C
C
C
C
C
C
C
P
P
P
P
P
P
P
I
I
I
I
I
I
I
C
C
C
C
C
C
P
P
P
P
P
P
I
I
I
I
I
I
11
12
13
WPA1
WPA2
7
7
8
9
10
Period
58
APPENDIX: INTERVIEW SCRIPT
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Participant name ____________________________________________
Participant organization ______________________________________
Participant title _____________________________________________
1) Present introductory paragraph (Handout 1) to participant.
a. Provide a summary of the introductory paragraph
b. Ask permission to record the interview (audio) and mark that you asked them
2) Gather demographic information (2 minutes)
a. Years of general experience
Less than 5 5-10 More than 10
b. Years of experience in IT industry or research
Less than 5 5-10 More than 10
c. Years of experience making IT related decisions (tech acquisitions, investments,
etc.)
Less than 5 5-10 More than 10
d. How would you rate your knowledge of the IT landscape
low
medium
high
e. Education ______________________________________________
3) Gather information on experience with technology forecasting, IT landscape analysis, etc.
(10 -15 Minutes)
Based on your experience, how are technologies chosen in the information technology (IT)
investment and development decision processes? i.e., How do organizations typically go
about analyzing the IT landscape to discover potential technologies to invest in?
ASK ALL OF THESE TOGETHER UPFRONT (HISTORICAL, CURRENT, FORECASTING)
How important do you think historical technology analysis of is for making decisions about
IT investment/development? (Rate 1-5 and then explain)
1
2
3
4
5
Very Unimportant
33
34
35
36
37
Neither important or
unimportant
Somewhat important
Very important
Please explain
How important do you think evaluating the current landscape of IT is for making investment/development decisions? (Rate 1-5 and then explain)
1
2
3
4
5
Very Unimportant
38
39
Somewhat Unimportant
Somewhat Unimportant
Neither important or
unimportant
Somewhat important
Very important
Please explain
59
1
2
3
How important do you think predicting the future landscape of IT (technology forecasting) is
for making decisions about IT investment/development?
Please Rate
1
2
3
4
5
Very Unimportant
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
Somewhat important
Very important
What types of techniques, methods, or tools have you or your company/organization used (or
are familiar with) to analyze the past, current and future landscape of information technology?
Based on your experience, what are common challenges in IT landscape analysis, forecasting
and the IT investment/development decision making process? In what ways, if any, do you
think existing tools/methods/techniques could be improved?
How effective do you feel existing tools/methods/techniques are for analyzing the technology
landscape and performing technology forecasting?
1
2
3
4
5
Somewhat ineffective
Neither effective or
ineffective
Somewhat effective
Very effective
Please explain
4) Provide handouts H2-1, H2-2, and H2-3 and discuss our proposed theory and method
(10-15 minutes)
a. Discuss the objective/motivation for the proposed technique
b. Go over constructs (with specific examples)
c. Go over paths of influence (with specific examples)
d. Discuss the Music case study
e. Discuss the method of identifying trends
f. Discuss the Wi-Fi example case study
5) Gather reactions to methodology (10-15 minutes)
a. Goal of methodology
How effective do you feel the proposed methodology is in providing a new technique for identifying and analyzing patterns in technology evolution?
1
2
3
4
5
Very ineffective
39
Neither important or
unimportant
Please explain
Very ineffective
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Somewhat Unimportant
Somewhat ineffective
Neither effective or
ineffective
Somewhat effective
Very effective
Please explain
60
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
b. Roles
Do you feel the use of the component, product, and infrastructure roles help in
understanding the nature of technology evolution? If so, how do you think they
create value?
Are there any missing or unnecessary roles? Please Explain
c. Relationships
Do you feel the method for identifying relationships between technology roles is
provides better understanding about the evolution of focal technology? Please
Explain
Do you feel the classification of paths of influence is complete? Please Explain
What do you think are the strengths and weaknesses of the technology roles and
paths of influence?
d. Logic of the method
Do you feel the method (qualitative and quantitative) for identifying patterns of
technology evolution is sound? Please Explain
Are there any assumptions or steps in the method that seem unreasonable? Please
Explain
e. Information output
Do you feel the information generated using the proposed methodology is useful?
Please Explain
Do you feel the graphical representation of trends is useful? Please Explain
What do you think are the strengths and weaknesses of the information generated
by the proposed methodology?
f. Overall utility of the approach
What is your general reaction to the proposed methodology and its utility for analyzing technology trends?
61
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
If you feel the proposed methodology and approach is useful, how is it useful for
organizations?
Do you feel the proposed conceptual approach and methodology would benefit
organizations making IT forecasts? Please Explain
Do you feel the proposed conceptual approach and methodology would aid in the
IT investment/development decision making process? Please Explain
Overall, what do you feel are the strengths and weaknesses of the proposed conceptual approach and methodology?
Who do you think is the most appropriate user of the proposed conceptual approach and methodology, if any? Please Explain
6) Debriefing
a. Provide the participant with the debriefing handout.
b. Give the participant a summary of the debriefing handout
c. Ask if they have any questions or additional comments
d. Point out the contact information
e. Discuss possibility of a follow up
f. Thank them and give them a thank you gift
62
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Handout 1: Description of Study and Purpose of Interview
We are a group of information systems researchers from XXXXXXXXXXX XXXXXXXXX
XXXXXXXXXXXX. We are currently developing new methods for analyzing the evolution of
information technology (IT) with the goal of providing useful tools to IT professionals to aid in
the IT investment decision making process. We are conducting interviews with IT industry
managers, IT industry consultants, IT industry researchers, and academic researches to gather
impressions and evaluations of our proposed conceptual approach and methods. This interview
will take approximately 45 minutes and will consist of four parts. First, we will ask for some
background information about you, your organization, and your experience and understanding of
information technology landscape analysis and forecasting. Second, we will present you with a
brief document outlining our proposed conceptual approach and methodology. We will explain
our conceptual approach and methodology using the document as a supplement; we will allow
time for any questions you might have. Third, we will ask for your reactions and evaluation of
the proposed theory and methodology. Fourth, we will perform a short debriefing. Thank you
for your participation. You may quit the interview at any time. Your personal information will
not be shared or published. The input of industry experts and academics is necessary to produce
new tools that are useful, practical, and accessible to our intended users and we highly value your
input.
Do you give us permission to record the audio of this interview?
Yes _________ No _________
63
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Handout 2-1: Overview of theory and constructs
Motivation
• The IT landscape is complex with many technologies and interdependencies.
• This makes the information technology investment decision making process is difficult.
• Can we create a methodology that systematically identifies patterns in technology evolution to
help inform managers making these decisions?
Technology Ecosystem
• A theoretical construct that is used to represent multiple associated ITs and the interdependent relationships among them over time.
• A technology ecosystem is centered on a focal technology in a certain context and consists of a
set of technologies acting in specific roles and influencing each other through paths of influence.
Focal Technology and Context
• The focal technology and its context are the center of the technology ecosystem. Example: mobile phone in the context of wireless entertainment.
• A technology ecosystem representation is flexible, choosing a different focal technology and/or
context changes the representation.
Technology Roles
• Components are the elemental pieces or subsystems that make up complex technologies. Examples: RAM chip, microprocessor, CMOS sensor.
• Products are built up from a set of components, and are designed to interact with a user to perform a specific set of functions. Examples, MP3 player, digital camera, personal computer.
• Infrastructure technologies work in conjunction or collaboration with (or as a peripheral to)
products to add value to users. Examples: laser printer, Ethernet, network attached storage.
Technology Relationships
• There are clear relationships between the components, products and infrastructure (e.g., digital
camera ecosystem
• Over time, technologies in each role change and evolve which impacts related technologies.
• The Paths of Influence matrix classifies these interactions.
• An evolutionary pattern is identified from multiple paths of influence over time.
COMPONENT
FUTURE STATE (C*)
PRODUCT
FUTURE STATE (P*)
INFRASTRUCTURE
FUTURE STATE (I*)
COMPONENT
PRESENT STATE (C)
Component Evolution
Design and Compilation
Standards and Infrastructure Development
PRODUCT
PRESENT STATE (P)
Product-Driven Component Development
Product Integration and
Evolution
Diffusion and Adoption
INFRASTRUCTURE
PRESENT STATE (I)
Infrastructure-Driven
Component Development
Infrastructure-Leveraging
Product Development
Support Evolution
64
1
Handout 2-2: Overview of qualitative method and example
YEAR
1989
1996
1998
Feb 1999
May 1999
May 2000
Jan 2001
Jul 2001
Oct 2001
Mar 2002
Apr 2003
Oct 2003
EVENT
German MP3 patent
US MP3 patent
First portable MP3 player (32 MB)
Sub Pop distributes MP3 music
Napster founded
Transactional watermarking developed
Apple iTunes music applications released
Napster injunction
10 GB Apple iPod introduced
20 GB iPod for PC introduced
40 GB iPod introduced
Dell DJ introduced
iTunes online music store opens
Sep 2004 MSN online music store opens
May 2005 Yahoo online music store opens
2
3
4
5
6
65
Handout 2-3: Overview of quantitative methodology and example
STEP
1. Coding
2. Frequency
Estimation
3. Threshold
Determination
4. Dominant Role
Identification
5. Pattern
Identification
3
4
DESCRIPTION
Raw data on technology introductions is coded into the component, product,
and infrastructure roles.
A function of the frequency of all technology introductions over time is estimated.
Based on the proportion of technologies within each role within the overall
number of technology introductions, a threshold function is derived for each
role.
Actual frequency of technology introductions is compared to threshold function for each role to determine dominant roles in each time period.
Transitions between dominant roles in adjacent time periods are mapped out.
Threshold Plots
45
8
802.11b Components
40
25
20
15
20
6
Certifications
Certifications
Actual Frequency
30
802.11b Infrastructure
7
5-Month Moving Average
35
25
802.11b Products
Threshold
Certifications
1
2
5
15
4
3
10
2
10
5
1
5
0
M 1
ar
-0
2
p0
M 2
ar
-0
Se 3
p0
M 3
ar
-0
Se 4
p0
M 4
ar
-0
Se 5
p05
M
ar
-0
Se 6
p06
Se
-0
1
p0
M
ar
Se
0
-0
0
p0
M
ar
Se
M 0
ar
-0
1
p0
M 1
ar
-0
Se 2
p0
M 2
ar
-0
Se 3
p0
M 3
ar
-0
Se 4
p0
M 4
ar
-0
Se 5
p0
M 5
ar
-0
Se 6
p06
Se
-0
0
p0
M
ar
0
Se
p01
M
ar
-0
Se 2
p02
M
ar
-0
Se 3
p03
M
ar
-0
Se 4
p04
M
ar
-0
Se 5
p05
M
ar
-0
Se 6
p06
0
-0
1
M
ar
Se
p0
M
ar
5
6
7
Se
-0
0
0
Pattern Identification
8
66
1
2
3
4
5
6
7
Handout 3: Debriefing
Interviewer: XXXXXXXXXXXXXXXXX
Date/Time of Interview ___________________________
Research Team:
XXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXX
8
9
XXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXX
10
11
XXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXX
12
13
XXXXXXXXXXX
XXXXXXXXXXXXXXXXXXXXXXXXXX
14
15
16
17
18
19
20
21
22
23
24
If you wish to follow up with our research team for any reason please contact:
XXXXXXXXXXXXXX
XXXXXXXXXXX
XXXXXXXXXXX
XXXXXXXXXXX
XXXXXXXXXXX
XXXXXXXXXXX
XXXXXXXXXXX
67