DARIUS MAHDJOUBI,

P.Eng

N ON -L INEAR M ODELS OF T ECHNOLOGICAL I NNOVATION

Linear Model of Technological Innovation another report of this series studied the origin, taxonomy
and deficiencies of the Linear (R&D) model of technological innovation in the wake of the Second
World War. The present report briefly studies the following non-linear innovation models.

Stephen Kline's Chain-Linked model

Ralph Gomory's Circle model
Alic-Branscomb's model
OECDs Oslo manual
John Ziman's Neural Net model

John Alic argues that four closely related themes characterize a non-linear understanding of the
technological process: the importance of downstream or non-R&D technical activities, the
pertinence of alternatives to the so-called pipeline view of technology generation and application,
the convergence of science and engineering, and the importance of borrowing and adapting
technology as well as generating it1. However, as the Linear Model of Innovation indicated, the
deficiencies of the Linear model go beyond the above four points
S T E P H E N K L I N E ' S C H A I N -L I N K E D M O D E L
The Chain-Linked Model, the most cited non-linear innovation model, was first reported in
Research, Invention, Innovation and Production: Models and Reality, Report INN-1, March
1985, Mechanical Engineering Department, Stanford University. Dr. Stephen Kline, an emeritus
professor of Mechanical Engineering, has further developed and revised the Chain-Linked model
in An Overview of Innovation2, Models of Innovation and Their Policy Consequences3. In
Styles of Innovation and Their Cultural Basis 4 Kline further elaborated the Chain-Linked model.
The Chain-Linked model has also been referred to in many other documents, for instance in the
Oslo Manual 5 , the OECDs Technology/Economy Program 6 and Canada and the National
System of Innovation7.

Darius Mahdjoubi, 1997 E-Mail: [email protected]

Kline argues that the Chain-Linked model is consistent with a detailed evaluation of the nature of
technology, the concept of innovation, and the failures of a simple linear model which are often
assumed, and the necessity that the linear model be replaced with a more complex model in order
to understand the nature of innovation. The Chain-Linked method emphasizes the socio-technical
nature of industry and technology and the necessity to look at it as a complex system. In the model
the first path of innovation process, central chain-of-innovation, begins with design and continues
through development and production to marketing. The second path is a series of feedbacks8.

Research

Existing Corpus of Knowledge

Potential Market

Invent and/or produce

analytic design

Detailed design and

test

Redesign and
procedure

Distribute and market

T H E C H A I N -L I N K E D M O D E L
In the Chain-Linked model, the general process starts with a market-finding phase followed by
design, production, marketing and distribution, and use phases. It differs from the Linear model in
a number of ways: there are multiple paths from which innovations may arise and many forms of
feedback. Research is not normally considered to be the initiating step (in fact, research occurs in
and contributes to all phases in the innovation process), and the primary source of innovation is
now held to be stored knowledge and technological paradigms9. It appears this model more closely
corresponds to the Japanese perception of the innovation process10.
The Technology/Economy Program - TEP by OECD (Organization for Economic Cooperation
and Development) addresses also the need to integrate policies for science and technology with
other aspects of government policy, particularly economic, social, industrial, energy, education
and labor policies. Technology and the Economy; the Key Relationships11, which is an outcome
of TEP and published in 1992, also refers to the Chain-Linked model. This report underlines that
the Chain-Linked model combines two somewhat different types of interaction. One concerns
processes within a given firm (or possibly a group of firms working in a tightly-knit network). The
second expresses the relationships between the individual firm and the wider science and
technology system within which it operates.
At the level of the firm, the innovation chain is visualized as a path starting with the perception of
a new market opportunity and/or a new science and technology-based invention; this is necessarily
followed by the analytic design for a new product or process, and subsequently leads to
development, production and marketing. Feedback relations are generated: short feedback loops
Non-Linear Models of Innovation

Section 5 Page 2

link each downstream phase in the central chain with the phase immediately preceding it and
longer feedback loops link perceived market demand and product users with phases upstream.
Problems identified by the processes of designing and testing new products and new processes
often spawn research in engineering disciplines but also in science.
The second set of relationships links the innovation process embedded in firms and industries with
the scientific and technical knowledge base and with research. In an industry-focused, interactive
approach to innovation, a useful analytical distinction can be made between the two different uses
of science and technology by firms, the use of available knowledge about physical and biological
processes, and the work undertaken to correct and add to that knowledge. Generally, innovation
takes place with the help of available knowledge. When corporate engineers confront a problem in
technical innovation, they will call first on known science and technology, most often in serial
stages. Only when those sources of information prove inadequate does a need arise for research.
This analysis of the role of industrial R&D in the innovation process applies directly to large firms.
Firms below a certain size cannot bear the cost of an R&D team
RALPH GOMORY'S CIRCLE MODEL
Dr. Ralph Gomory is a retired senior vice president of IBM and the current president of the Alfred
Sloan Foundation. Gomory has developed the Circle model of innovation as an alternative to the
linear model that he calls the Ladder model. Gomory, in many articles, describes the main aspects
of the Circle model12
In the Ladder paradigm, new things descend from the realm of science - step by step - into practice
and become the genesis of an industry. Well known examples in this respect include transistor and
molecular biology. The belief that this kind of scientific dominance should translate into product
dominance is probably, in many cases, a residue of the Second World War and the enormous
impression made by the science-led, science-developed process of the atomic bomb. After the
Second World War, a belief emerged that scientific dominance does translate into economic
dominance.
Scientists play the dominant role both in basic research and in the early phases of the industry
because they are the only people who understand what's going on in sufficient detail. So, in the
early stages of a new industry, the Ladder paradigm predominates. Everything revolves around the
new technology. The Ladder paradigm is really a paradigm for getting things started, not for
winning the longer race.
Much of the dominant thinking about innovation comes from the Ladder process because it has
been so visible and so spectacular, involving the emergence of a new scientific effort and new
products. We must make the cyclic development process more visible and to understand it better.
The cyclic process is a second relationship of technology and science. The cyclic development
process is a process of repeated, continuous, incremental improvement. It is that process of
following up what exists in manufacturing with the next model, which is designed, built and
prototyped, tested, redesigned for manufacturing, put into production, and then in turn starts the
next generation. This process is characteristic of the later (not the earlier) states of an industry. The
Non-Linear Models of Innovation

Section 5 Page 3

type of industry in this case is discrete manufacturing, of which automobiles and transistors are
very good examples. It is this cyclic development process that determines in the long run, then,
who will be dominant in this industry. It is not as glamorous as the breakthrough type of invention;
but, nevertheless, the progress which it causes is enormous.
The cyclic process is very different from the Ladder process. It is not science-based. It is based on
what is already there, the existing product and its restrictions. So this type of development is very
much restricted, not by a totally new idea, but rather by what is already there, whether that be the
plant, or the tools, or the engineering team, or what they understand. And if new technologies are
going to be part of this, they must fit into that very special world.
The Ladder model, is the one that has shaped most peoples thinking on the subject of science,
technology, and product. The Circle is less familiar. The Ladder is characteristic of the early stages
of an industry and the Circle is characteristic of later stages. However, it is the Circle, not the
Ladder, that has been decisive in the industries in the US which are in trouble - automobiles,
semiconductors, and TV. The notion of investing in R&D, consistent with the Ladder approach, is
that you put the R&D here and the product pops out there. That is very, very different from what is
really encountered if participating in a cycle development process. It is the industrial success
which causes the R&D spending, not the other way around.
The wrong mental picture hurts a great deal. Consider the picture brought to mind by the phrase,
Commercialization of new technology. Most people believe in the notion that commercialization means to take something new and make it commercial, whereas the essence of the cyclic
development process rests in defining something which is already commercial. When one talks
about the commercialization of scientific discovery, one is relegated to use of phase one, or the
Ladder process.
John Alic argues that the Circle approach, compared to the linear model, gives equal weight to
technical virtuosity in all of the functions - research, design, production, and marketing - and
produces a model that stresses the importance of close coupling among them13.
A L I C -B R A N S C O M B ' S M O D E L
Dr. John Alic, a senior associate of the Office of Technology Assessment of the US Congress, and
Dr. Lewis Branscomb, from Harvard University, have studied the evolution of the tacit US
technology policy in the years after the Second World War. In their books, Empowering
Technology: Implementing a US Strategy14 and Beyond Spin-off: Military and Commercial
Technologies in a Changing World, they have greatly contributed to the studies related to
specification and procedure of evolution of the Linear model which is tied to the post-war tacit US
technology strategy. They vividly delineate some characteristics of the new innovation procedures,
described below. However, their studies have not gone far enough to articulate a new innovation
model.
John Alic regards innovation as a social process involving the application of knowledge,
together with other inputs, to design, develop, create, and market some products. The output of
innovation can include intangible service products as well as physical objects and systems. The
Non-Linear Models of Innovation

Section 5 Page 4

artifactual product should be viewed as derivative, the consequence of research, design,

development, production, and marketing activities - of what people and organizations know
(declarative knowledge) and can do (procedural knowledge). Technological innovation, especially
in the later stages often called commercialization, is an activity pursued by business firms.
The core activity of commercialization and the culminating stage of innovation is engineering
design and development - an activity sufficiently different from R&D that we might call it D&D.
As an extension of research, as in R&D, development implies reducing new knowledge to practice
- verifying and validating experimental results and theoretical predictions, exploring specific cases,
and determining the accuracy and limits of mathematical models and methods. But when coupled
with design, development implies the steady refinement of concrete products, processes and
systems through an iterative process of conceptualization, preliminary design, analysis, testing
and redesign. This is the everyday technical work of private firms, often called product
development or product engineering (or process development, software development, and so on).
Instead of thinking of design and development (activities in which knowledge is applied) as D&D,
reserving R&D for activities that generate knowledge, the latter term has entered into common use
to stand for either or both. Although practitioners, understanding the distinctions, do not have
much trouble communicating among themselves, the broad range of activities encompassed by
R&D in its now-common usage sometimes leads to confusion among economists, policy analysts,
and other observers.
Innovation results from the artful combination of new knowledge with, typically, a great deal of
existing knowledge. New knowledge and methods originating in R&D may be critical. At the
same time, a successful innovation, leaving aside exceptions such as defense systems, is one that
meets the test of the marketplace. Successful innovation depends more directly on D&D, though
not of course to the exclusion of R&D. But the latter may be far removed in time and place from
the design and development that precedes marketplace success.
Alic and Branscomb argue that central to the national system of any advanced economy will be
mechanisms both for generating new knowledge through R&D and for applying knowledge, new
and existing. Less developed economies with fewer needs for new knowledge may nonetheless
benefit from R&D tailored specifically to their situation.
OECD S O S L O M A N U A L
Proposed Guidelines for Collecting and Interpreting Technological Innovation Data, code
named the Oslo Manual, first published in 1991 by OECD, in Paris. Since then the Oslo Manual
has been revised and further elaborated.
The Oslo Manual intends to develop new procedures to collect and interpret data on innovation
and technology development, and then substitute the Frascati Manual, another element of the
Linear model, but it does articulate and depict the model which is going to be used. The Oslo
Manual states that it is now commonly accepted that the development and diffusion of new
technologies are central to the growth of output and productivity. But our understanding of the
innovation process and its economic impact is still deficient in many areas.15

Non-Linear Models of Innovation

Section 5 Page 5

While in the context of the Linear paradigm expenditure on R&D was the key to innovation, the
Oslo Manual emphasizes that despite the considerable efforts of researchers, however, we are still
a long way from understanding all of the factors which shape the rate, direction and effects of
technological change, at enterprise, industry, regional or country level. There are many reasons for
this. Some are theoretical, to do with the difficulties of building technological change into
economic theory and analysis.
The Oslo Manual regards design as an essential part of the innovation process. It covers plans and
drawings aimed at defining procedures, technical specifications and operational features necessary
to the conception, development and manufacturing and marketing of new products and processes.
However, consistent with the Linear model, the Oslo Manual classifies technology development
into product and process development and then continues as follows: A Product innovation is
the commercialization of a technologically changed product. A process innovation occurs when
there is significant change in the technology of the production of an item. This may involve new
equipment, new management and organization methods, or both
JOHN ZIMANS NEURAL NET MODEL
This model of innovation was developed by John Ziman from the Science Policy Research Unit
Group in London UK. Using the basic concepts of the Linear model, such as basic and applied
research and development, this model seeks to provide a metaphor based on drawing lines to
represent causal or interactive linkages between various points in a cognitive space16.

Non-Linear Models of Innovation

Section 5 Page 6

REFERENCES

Alic J, (et al) 1992, Beyond Spin-off: Military and Commercial Technologies in a Changing World, Harvard
College, Cambridge MA
2
Kline S, and Nathan Rosenberg, 1986, An Overview of Innovation. Appeared in: Ralph Landau and Nathan Rosenberg
(Editors), 1986, The Positive Sum Strategy: Harnessing Technology for Economic Growth, National Academy Press,
Washington, DC
3
Kline S, 1991, Models of Innovation and Their Policy Consequences. Appeared in Kingery, David (ed), 1991,
Japanese/American Technological Innovation, Elsevier, New York
4
Kline S, 1991, Styles of Innovation and Their Cultural Basis, Chemtech, August 1991
5
OECD, 1991, Oslo Manual: Proposed Guidelines for Collecting and Interpreting Technological Innovation Data,
OECD, Paris
6
OECD, 1992, Technology and Economy: The Key Relationships, OECD, Paris
7
de la Mothe J, 1994, Canada and the National Systems of Innovation. Appeared in Government of Canada, 1994,
Resource Book for Science and Technology Consultations, Volume II, Ottawa.
8
Kline S, 1991, Models of Innovation and Their Policy Consequences. Appeared in Kingery, David (ed), 1991,
Japanese/American Technological Innovation, Elsevier, New York
9
Kline S, 1991, Styles of Innovation and Their Cultural Basis, Chemtech, August 1991
10
Poncelet E, 1991, Japanese and American Approaches to Technological Innovation: Cultural Influences.
Appeared in Kingery, David (ed), 1991, Japanese/American Technological Innovation, Elsevier, New York
11
OECD, 1992, Technology and Economy: The Key Relationships, OECD, Paris
12
Gomory R,1989 From the Ladder of Science to the Product Development Cycle, Harvard Business Review,
December. 1989,
13
Alic J, (et al) 1992, Beyond Spin-off: Military and Commercial Technologies in a Changing World, Harvard
College, Cambridge MA
14
Branscomb L, 1993, Empowering Technology: Implementing a US Strategy, MIT Press, Cambridge MA
15
OECD, 1991, Oslo Manual: Proposed Guidelines for Collecting and Interpreting Technological Innovation Data,
OECD, Paris
16
Ziman H, 1991, A Neural Model of Innovation. Science and Public Policy, Volume 18, No 1, February 1991, pp.
65-75

Last update: February 10, 1997

Non-Linear Models of Innovation

Section 5 Page 7