You are on page 1of 12

Research Policy 40 (2011) 403–414

Contents lists available at ScienceDirect

Research Policy
journal homepage: www.elsevier.com/locate/respol

A framework for analysis of multi-mode interaction among technologies with


examples from the history of alternative transport fuels in Sweden
Björn A. Sandén a,∗ , Karl M. Hillman b
a
Environmental Systems Analysis, Department of Energy and Environment, Chalmers University of Technology, Göteborg 41296, Sweden
b
Institute for Management of Innovation and Technology (IMIT), University of Gävle, Sweden

a r t i c l e i n f o a b s t r a c t

Article history: The relationship between technologies is a salient feature of the literature on technical change and terms
Received 31 August 2009 like ‘dominant design’ and ‘technology lock-in’ are part of the standard vocabulary and put competition
Received in revised form among technologies in focus. The aim of this paper is to provide an account of the wide range of interaction
20 December 2010
modes beyond competition that is prevalent in transition processes and to develop a conceptual frame-
Accepted 22 December 2010
work to facilitate more detailed and nuanced descriptions of technology interaction. Besides competition,
Available online 26 January 2011
we identify five other basic modes of interaction: symbiosis, neutralism, parasitism, commensalism and
amensalism. Further, we describe interaction as overlapping value chains. Defining a technology as a
Keywords:
Technology selection
socio-technical system extending in material, organisational and conceptual dimensions allows for an
Competition even more detailed description of interaction. The conceptual framework is tested on and illustrated by
Symbiosis a case study of interaction among alternative transport fuels in Sweden 1974–2004.
Transition © 2011 Elsevier B.V. All rights reserved.
Lock-in
Technological innovation system

1. Introduction (1955) described the latter phenomenon as ‘technological inter-


relatedness’. While none of these concepts explicitly address
The literature on technology interaction has hitherto focussed complementarities between potential substitutes, Islas (1997, p.
on competition. A dominant idea in the literature on technical 64), in his story of gas and steam turbines, saw that “two technolo-
change is that following a crisis or a technical breakthrough there gies can be complementary and in competition at the same time,
is an ‘era of ferment’ when many novel technologies emerge and and in the same market sector”. Geels (2005) discussed hybridisa-
compete to be selected as the new dominant design (Abernathy tion, such as simultaneous use of steam and sails and also observed
and Utterback, 1978; Arthur, 1988; David, 1985; Tushman and that one technology may catalyse a development and open devel-
Anderson, 1986; Utterback, 1994). This model is attractive due opment pathways for others.2 For economic actors (firms) such
to its simplicity but could be too simple to effectively describe ‘spillovers’ are termed positive externalities, i.e. free utilities paid
change processes. There is evidence in the literature that besides for by someone else. Marshall (1890) underlined the importance of
competition also other modes of interaction could be of great economies that are external to the firm but internal to an industry,
importance.1 Rosenberg (1976) observed ‘technological conver- and Porter (1990) observed that positive external economies also
gence’ where one process developed in one industry is imitated often extend to related industries within a nation.
in others, and ‘technological complementarities’ where two tech- Taking these observations one step further Pistorius and
nologies are combined to fulfil or enhance a function. Frankel Utterback (1997) outlined a multi-mode framework for technology
interaction based on similar frameworks in organisational ecology
(Brittain and Wholey, 1988) originating from community ecology
in biology (Odum and Barrett, 2005).3 They pointed out that tech-
∗ Corresponding author. Tel.: +46 31 772 8612; fax: +46 31 772 2172.
E-mail address: bjorn.sanden@chalmers.se (B.A. Sandén).
1
Margulis and Sagan (2002) argue that the literature on biological evolution has
2
also paid too much attention to competition and selection of species and less to As an example of a catalysing technology, Geels (2005, p. 692) takes the bicycle
the emergence of species. They put forward a theory that put symbioses centre in the transition from horse-drawn carriages to automobiles: “the bicycle led to
stage (instead of accidental mutations). Symbiosis of already existing species creates processes of change in the socio-technical regime on which the automobile could
new genomes in nature, and is thus the motor in evolution. This has its parallel in later build.”
3
evolutionary economics and the idea of radical change as a recombination of existing ‘Community ecology’ instead of ‘population ecology’ since it is about the inter-
ideas and technologies. action of many populations (species).

0048-7333/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.respol.2010.12.005
404 B.A. Sandén, K.M. Hillman / Research Policy 40 (2011) 403–414

nologies not only compete but could also be in symbiosis or in Complementary chains Alternative chains
predator–prey relationships and provided some examples. a b
To inform public policy and firm strategy, we believe it is essen- Hn,11 Hn,1
tial to better understand the many possible modes of technology Pn,1 Pn,1
interaction. It could be of particular importance in policy areas, Upstream
chains
such as energy and climate policy, where attempts are being made
to govern long term change processes in large technical systems. Hn-1,1 Hn-1,2 Hn-1,1 Hn-1,3
It has been suggested that reaching carbon neutral energy and
transport systems is so radical that a stepwise change process is
required, where ‘hybrid technologies’ (Geels, 2002) or ‘two-world c d
technologies’ (Kemp and Rotmans, 2001) function as ‘bridging tech- H’n+1,1 H’n+1,2 H’n+1,1 H’n+1,3

nologies’ (Andersson and Jacobsson, 2000). New technologies need


to make use of the existing system and then gradually transform Downstream
chains
it (Clark, 1985; Freeman, 1996). The ‘bridging’ function indicates Pn,1 Pn,1

that some kind of technological complementarities, or spillovers, H’n,1 H’n1,


need to be involved. However, others warn that initial investments
in technologies with limited potential could lock out technolo-
gies with better long-term prospects (Andersson and Jacobsson, Fig. 1. Value chain hierarchies. All products and processes P are part of several value
chains forming hierarchies. Hierarchies made up of value chains constitute basic
2000; Arthur, 1988; Menanteau, 2000; Sandén, 2004). Given this
building blocks in our definition of technology. From any given point of reference
dilemma, policy and firm strategy cannot successfully address tech- (level n) a distinction can be made between upstream supply chains and down-
nologies in isolation, nor retreat to, so called, ‘technology neutral’ stream application chains. Further, some value chains are complementary while
policies (Sandén and Azar, 2005). The relationships between tech- others are alternatives (indicated by the dotted line). Here we use H as a short for
nologies need to be investigated and taken into account. To be hierarchies made up of upstream value chains and H for those made up of down-
stream value chains. The arrows indicate that a lower order product or process is
able to differentiate things like ‘bridging technologies’ and dead used in a higher order product or process (closer to end use).
ends, there is a need for a framework that in more detail describes
technology interaction.
The aim of this paper is twofold. First, we want to provide an chains. We conclude that interaction emanates from shared ele-
account of the wide range of interaction modes beyond competi- ments in different parts of the value chain. These overlaps result
tion that is prevalent in transition processes. Second, we want to in different modes of interaction ranging from pure competition to
take one step further and systematize the observations made in pure symbiosis. We end Section 2 by an outline of how our formal-
previous literature and develop a more elaborate model of multi- isation of technology interaction could enrich diffusion models.
mode interactions among technologies. While the framework,
described in Section 2, is broad enough to include any technology 2.1. Bundles of value chains and technology boundaries
interaction, we focus on interactions between different emerging
technologies that in principle fulfil the same function and that in Theories of design hierarchies (Baldwin and Clark, 2000; Clark,
a simpler model would be viewed as competing alternatives in an 1985) and complex systems (Murmann and Frenken, 2006; Simon,
‘era of ferment’.4 In Section 3, the conceptual framework is tested 1962) have dealt with the problem of describing similarities
and illustrated by a case study of interaction among alternative between technologies and technology delineation and provide us
transport fuels in Sweden 1974–2004.5 Finally, our results are with a useful starting point. They present technologies as systems
summarised in Section 4. built up by subsystems of lower order, while in turn being sub-
systems of higher order systems. A kind of pyramidal structure is
2. A model of technology interaction implied with many subsystems making up a technology, e.g. dif-
ferent materials are combined into components that are combined
It should be stated upfront that we in this article put technology into a car.
centre stage, rather than people and organisations. A result of this Instead of viewing these hierarchies as hierarchies of products,
is that technologies tend to get a life of their own in the text. This we would prefer to see them as hierarchies of products and pro-
does not imply that we allocate agency to technologies rather than cesses, since some differentiating aspects, e.g. the environmental
to people. We take this approach merely since the purpose of the and ethical characteristics of a technology, are often invisible in
paper is to analyse relations between technologies. the final product as such.6 As depicted in Fig. 1a, a technology
In a model of technology interaction we need to be able to sep- can then be viewed as a hierarchy Hn,1 made up by complemen-
arate technologies, i.e. establish technology boundaries and define tary upstream supply chains, or sub-hierarchies, Hn−1,1 and Hn−1,2 ,
‘technology’. We arrive at a definition of technologies as socio- brought together in the process or product Pn,1 . However, not all
technical systems made up of heterogeneous elements, such as supply chains are complementary; there may also be many alter-
physical objects, organisations, knowledge and regulation. Further, native supply chains that can fulfil the same function, as Hn−1,1 and
these elements are organised in value chains. Any specific technol- Hn−1,3 , denoted by the dotted line in Fig. 1b. What is meant by a
ogy referred to by a common word or phrase such as ‘car’ or ‘wind technology in everyday language, e.g. a ‘car’, normally allows for
power’ is defined by a set of complementary and alternative value many alternative supply chains.
Further, one can observe that the pyramid can be turned upside
down. One product or process can be used in many alternative
4
In the terminology of Raven (2005, p. 270) this would correspond to the case downstream applications (Fig. 1d). The car can be used as a taxi or
where several ‘niches’ (technologies) develop against the backdrop of one ‘regime’
(sector). The history of alternative fuels in Sweden described in the article also con-
tains elements of the case when multiple niches develop against the backdrop of
6
several regimes (Sandén and Jonasson, 2005). Such intangible characteristics can be made more tangible and visible in prod-
5
The full report on the history of alternative fuels is provided in Sandén and ucts by the use of branding, e.g. eco-labelling. Intangible differences may also result
Jonasson (2005). In Hillman and Sandén (2008), we explore future trajectories and from different uses of a product. Different brands of jeans may be physically indif-
interactions in the same system. ferent but associated with different user groups.
B.A. Sandén, K.M. Hillman / Research Policy 40 (2011) 403–414 405

H’n+1,1 H’n+1,3 H’n+1,1 H’n+1,3

H’n,1 H’n,1
Pn,1 Pn,1 a b c d
Ta Tb
Hn,1 Fig. 3. Alternative conceptual boundaries. The name given to a technology, or bun-
dle, tends to be associated with one or many waists on the bundle. A technology can
be defined by being delimited in terms of alternative processes towards upstream
Hn-1,1
n- Hn-1,3
n- Hn-1,1
n and downstream endpoints, e.g. ‘biofuel for transport’ (a), at some mid-point, e.g.
‘methanol’ (b), mainly far downstream, e.g. ‘transportation technology’ (c), or mainly
far upstream, e.g. ‘nanotechnology’ or ‘bioenergy’ (d). If (d) is ‘nanotechnology’ and
H’n+1,1 H’n+1,3 we add a qualifier as in ‘nanotechnology for energy applications’ we get a definition
H’n,1
H that more resembles graph (a) inscribed in graph (d).

Tc Td Te is true for ‘methanol’, which can be made from many resources


and used for other ends besides fuel combustion (Fig. 3b). Some-
Hn,1 Hn,1
Hn-1,3 times, a technology is only specified with regard to end use, such
as ‘transportation technology’ (Fig. 3c), and sometimes the delim-
itation lies mainly upstream such as in ‘nanotechnology’ (Fig. 3d).
Fig. 2. Bundles of value chains. A technology T can be defined by ‘a bundle of value
chains’ that includes many or a few alternative upstream supply chains (Hn−1 ) and If Fig. 3d illustrates ‘nanotechnology’ and we add a qualifier as in
downstream application chains (Hn+1 ). Technology Ta illustrates a wider definition, ‘nanotechnology for energy applications’ we get a definition that
while Tb , Tc , Td and Te are more narrowly defined. Tb and Tc overlap downstream more resembles Fig. 3a inscribed in Fig. 3d.

(Hn,1 ) while Td and Te overlap upstream (Hn,1 ). Complementary value chains are not
shown in the graphs. For Tc , Td and Te some lower level details are hidden.
2.2. Structural overlap and elements of socio-technical systems

as a police car. To obtain symmetry we note that also some appli-


Now we can make the observation that two technologies may
cations may be complementary (Fig. 1c) in the sense that they fulfil
overlap at different levels of the value chain (in different parts of
multiple purposes or use co-products or non-exclusive goods like
the bundle), that is, in supply chains (production processes) or in
non-patented knowledge.
applications. For example, Tb and Tc in Fig. 2 share the same appli-
Any technology is a combination of upstream and downstream  , and T and T share the same production processes
cations Hn,1 d e
hierarchies as in Fig. 2. We thus come to the more general defini-
Hn,1 . Two technologies overlap if they at some level in the hierar-
tion of a technology as a ‘bundle of value chains’ (or more precisely a
chy use or can use the same input or process, or fulfil or can fulfil the
“system of socio-technical elements organised in bundles of value
same function. This overlap is the basis for technology interaction
chains”, see Section 2.2), made of complementary and alternative
of different kinds.
upstream supply chains (subsystems) and downstream applica-
To develop a more elaborate classification of overlaps we make
tions (higher order systems). A technology can fully be defined by
use of the literature on ‘socio-technical systems’. The bundle of
how it can be made and what it can be used for.7
value chains defines in fact a socio-technical system. The literature
Technology boundaries may be more or less inclusive, i.e. the
on economics of innovation and science and technology studies
bundles may include many or a few alternative value chains. There
has identified a number of structural elements that constitute a
is not one correct boundary but different boundaries could be more
socio-technical system (see e.g. Bergek et al., 2008a,b; Carlsson
or less useful in a given situation. The usefulness of a particular
et al., 2002; Carlsson and Stankiewicz, 1991; Geels, 2004; Hughes,
delineation depends on the purpose of making it. In the concept ‘Ta ’
1987).8 Here, we group them into three main categories: artefacts,
we can for example include all possible ways of producing a device
actors and schemata. By ‘schemata’ we refer to virtual proper-
that are capable of converting solar energy directly to electricity,
ties, regularities that can be abstracted from artefacts and actors,
i.e. all types of solar cells regardless of semiconductor material and
such as knowledge and rules.9 Correspondingly we may say that
production process, and include all applications. Ta in Fig. 2 illus-
the technology, understood as a socio-technical system, extends in
trates such a wide definition including all alternative upstream and
 ]. Alternatively, we could be
a multidimensional space with material, organisational and con-
downstream chains, i.e. Ta = [Hn,1 , Hn,1
 ] or T that includes
ceptual dimensions. Some systems exist as full-fledged systems,
concerned with a technology Tb = [Hn−1,1 , Hn,1 c extending in all dimensions, being materially, organisationally and
a limited number of alternative upstream supply chains, such as
conceptually well-developed, while others exist as embryos (e.g. as
silicon solar cells only, or Td and Te that includes solar cells used in
a piece of knowledge scribbled on the back of an envelope or as an
a limited number of downstream applications, such as electricity
expectation held by a few individuals). Two systems may overlap
production in satellites or on roof-tops.
in one or many dimensions.
The name given to a bundle tends to be associated with one or
many waists on the bundle. For example, compare the denomina-
tions ‘biofuel for transport’ and ‘methanol’, the former bundle being 8
The terms ‘technological system’, ‘technological innovation system’ and ‘socio-
conceptually thinner (specific) towards the resource and applica- technical system’ are used by different authors that give them a slightly different
tion endpoints while allowing for many chemical compositions and content. In some versions, the physical artefacts are not included. As noted by
productions processes in the midpoint (Fig. 3a), while the opposite Markard and Truffer (2008) and Bergek et al. (2008b) there are two general func-
tions of a system that are given more or less attention. The first function is to enable
production and use of a product, the second is to enable growth of production and
use of a product, captured by the term innovation. (An analogy is the metabolic sys-
7
A similar, but somewhat less general, delineation is provided by Carlsson et al. tem of a child that enables playing as well as growing.) We will reserve the concept
(2002). They distinguish between three levels of analysis of technological systems: of technological innovation system for the theoretical model describing the growth
first technologies (a generic knowledge field) that are applied in many different aspect (Section 2.4).
9
products, second, products (or artefacts) that are combinations of many knowledge This use of ‘schemata’ may deviate somewhat from how the term is used in e.g.
fields and used in many applications, and third applications (or functions) that are psychology or computer science. Our somewhat broader meaning of the term is
combinations of many related products (compare Fig. 3). close to how it is used in Sewell (1992).
406 B.A. Sandén, K.M. Hillman / Research Policy 40 (2011) 403–414

In the material dimension a socio-technical system is constituted procedures and routines in organisations (Scott, 2001). A fraction of
by physical artefacts. Two systems overlap in this dimension when this technical knowledge is embedded in physical artefacts. Since
they use the same physical artefact, e.g. if the same production technical knowledge can be codified in symbolic systems (articles,
plant is used to produce two different fuels, or if the same physical patents or more general, text, drawings, etc.) it can also exist out-
infrastructure such as roads is used for different types of cars. side of the actors it informs and the artefacts it describes. Technical
In the organisational dimension the system is constituted by knowledge is constantly transformed from one type to another: in
actors. Actors could be individuals, or individuals closely linked into one direction, general codified knowledge is turned into practical
firms and other organisations, or more loosely linked networks of skills and particular artefact designs, and in the other direction,
individuals or organisations. Organisations include not only firms practical experience is generalised into codified knowledge.
but also universities, industry associations, NGOs and government Knowledge overlaps are dealt with in a rich literature on knowl-
bodies. There are many types of networks. Learning networks link edge spillovers. The focus of the literature appears to be spillover
suppliers to users, related firms or competitors, or university to of codified knowledge between sectors (e.g. Grupp, 1996), or
industry (Carlsson and Jacobsson, 1993). They constitute impor- spillover of skills between firms in regional innovation systems (e.g.
tant modes for the transfer of knowledge and perception of what is Audretsch and Feldman, 1996). While overlaps in the form of tacit
possible and desirable. Political networks are of equal importance. skills need to also overlap in the organisational dimension (shared
Policy making takes place in a context where advocacy coalitions, actors), codified knowledge developed for one technology can be
made up of a range of actors sharing a set of norms or beliefs, try applied in the development of other technologies without any actor
to influence policy in line with those beliefs (Sabatier, 1988). Obvi- overlap.
ously, one actor can support more than one technology, thereby Expectations have to do with beliefs about the future perfor-
creating a systemic overlap, and a constellation of actors (organi- mance of sometimes large and complex systems (van Lente and
sation or network) once developed to support one technology may Rip, 1998). There is not a clear demarcation between expectations
at a later stage fit another technology. and technical knowledge. Expectations can also be framed by the
In the conceptual dimension the system is constituted by schemata “if A then B” but cannot be tested and verified in the
schemata. These schemata define what actors and artefacts are same way as technical knowledge. Expectations of one technol-
able to do and what they ought to do. They are not only constrain- ogy may spill over to other technologies, in particular if they are
ing but also enabling. Schemata may be embedded in actors and conceptually lumped together.
physical artefacts, but they can also exist separately from these, Normative (prescriptive) schemata regulate interactions
codified in symbolic systems. Hence, schemata in themselves can between actors and define what actors and artefacts should or
be transferred between systems and two systems can overlap in a should not do. These rules include hard regulations (controlled
conceptual dimension while not sharing any physical artefacts or by juridical systems) and normative rules such as norms and
actors. We distinguish between three basic types of schemata: pos- attitudes (controlled by social systems). Like positive schemata,
itive schemata (technical knowledge and expectations), normative also normative schemata may be codified in symbolic systems, or
schemata (normative and regulative rules) and concepts.10,11 be embedded in people, organisational routines and artefacts (e.g.
Positive (or descriptive) schemata can be divided into technical as standards). Interaction in this dimension includes for example
knowledge and expectations, i.e. beliefs that are more or less verifi- regulation adapted to one technology that exclude or suit another
able. In much of the literature on socio-technical systems technical technology and attitudes towards one technology that spill over to
knowledge (or instrumental knowledge) is not given the status of another.
being a separate structural element in the same way as ‘institutions’ A fundamental type of schemata, that is not always given the
or ‘rules’ (see footnote 11). This strikes us as odd. While norma- attention it deserves, is the differentiation and meaning of concepts,
tive and regulative rules define what actors and artefacts should i.e. the very structure of symbolic systems. These are schemata in
or should not do, technical knowledge constrains what actors and the sense of conventions and they are the building blocks of all other
artefacts are able to do. This kind of schemata can be captured by kinds of schemata. Concepts go beyond empirical testing, but shape
correlations and cause–effect chains: if A then B. Such a schema what can be thought and understood (Polanyi, 1958). Concepts are a
makes a repeatable and verifiable prediction about the future of a prerequisite for codified technical knowledge and expectations, as
limited system that is stable over time. It does apply not only to the well as for codified normative and regulative rules. Different tech-
production and use of artefacts but also to knowledge related to e.g. nologies and products may have an overlap in terms of a shared
marketing and business models. Technical knowledge is found as label, such as ‘environmentally friendly products’, ‘renewables’ or
competence within actors: as explicit knowledge or more expe- ‘biofuels’, or depend on classifications needed for theories, techni-
rience based tacit skills within individuals (Polanyi, 1958) or as cal descriptions, testing procedures, regulation etc.
The different types of schemata are not always easily separated.
Taken jointly they influence decisions and actions in the form of
‘frames’ (Bijker, 1995; Geels, 2002) or ‘paradigms’ (Dosi, 1982).
10
The first two reflect the classical philosophical dichotomy between how things Legitimacy is a term used for how well a phenomenon (here tech-
‘are’ and how they ‘ought to be’ (Hume, 1740). nology) fits the dominant frames, e.g. how it appears in relation to
11
In the literature on technological innovation systems (e.g. Bergek et al., 2008a),
the term ‘institutions’ is normally used, while for example Geels (2004) prefer the
widely accepted performance criteria, legal frameworks and beliefs
term ‘rules’. Since we want to fully capture the virtual dimension we are not com- about the future. Legitimacy can be built up by a real presence
pletely satisfied with any of these terms. In the terminology used here, we believe and proved performance, but also by expectations of future pres-
‘institutions’ would include not only normative schemata but also concepts and ence and performance. From a practical standpoint, when collecting
expectations (‘cognitive’ institutions) but not technical knowledge. An alternative
empirical evidence of overlap, technical knowledge and regulation
subdivision is then technical knowledge (codified and tacit) and institutions (regu-
lation, attitudes, expectations and concepts). This also appears to be the definition is more easily distinguished, while attitudes and expectations are
of institutions used by Scott (2001) with the small difference that Scott also includes often difficult to separate.12
organisational routines as ‘institutions’. In the literature on technological transitions,
technical knowledge would normally not be included in the ‘rules’ category. For
some reason these literature strands tend to treat technical knowledge as embod-
12
ied in actors, while other schemata are given a more independent position. There are Attitudes and expectations can be tracked by studies of written and spoken
of course also those that highlight the role of technical knowledge, see for example language (‘discourse analysis’). Various internet applications are opening up new
the discussion on ‘design space’ by Stankiewicz (2000). possibilities for analysing huge amounts of statements and tracking opinions.
B.A. Sandén, K.M. Hillman / Research Policy 40 (2011) 403–414 407

Table 1
Two-technology (two-species) population interaction.

Mode of interaction Technology 1 Technology 2 General nature of interaction

Competition − − Inhibition when common resource or market is in short supply


Symbiosis + + Interaction favourable to both
Neutralism 0 0 Neither population affects the other
Parasitism (and predation) − + Technology 2 is benefited and 1 is inhibited
Commensalism 0 + Technology 2 is benefited, 1 not affected
Amensalism 0 − Technology 2 is inhibited, 1 not affected

Based on Odum and Barrett (2005).

2.3. Six modes of interaction be mutually dependent in an application, such as cars and petrol
(Fig. 1a). This kind of overlap in applications have been described
From the previous sections we conclude that interaction through the concepts ‘enabling technologies’ (Utterback, 1994) and
emanates from overlaps, i.e. shared elements in different parts of ‘technological complementarities’ (Rosenberg, 1976). Upstream,
the value chain. As stated in Section 1, interaction between tech- co-products are more or less dependent on each other, as cap-
nologies is commonly understood as competition only. However, tured by the term ‘economies of scope’ (Panzar and Willig, 1981).
there is more to it than that. As suggested by Pistorius and Utterback At one end of the spectrum, where one co-product is a by-product
(1997), a list of two-technology interaction modes can be borrowed of low economic value, the interaction borders on commensalism.
from biology, more precisely community ecology. In a recent text- Our interest here is focused on technologies that are potential sub-
book, Odum and Barrett (2005) distinguish between nine modes stitutes. In this case too, we may find examples of static symbiosis.
of interaction between species. In Table 1, we list six categories Upstream, substitutes may be produced in the same process (they
we find useful in the context of technology interaction.13 Besides are co-products), such as ethanol and biogas in some fermentation
competition, there are five other modes of interaction, including no plants (Fig. 1c). Downstream, potential substitutes may be used
interaction (neutralism). Since these six modes comprise all possi- in combination to increase performance, e.g. when electricity and
ble combinations of signs (+, −, 0) they cover all possible interaction petrol is used in plug-in-hybrid vehicles. This phenomenon has
modes and provide us with a useful terminology. been captured by the term ‘hybridisation’ (Geels, 2005).
Before providing examples of how overlap gives rise to the inter- Neutralism typically occurs when two technologies deliver dif-
action modes in Table 1, a distinction between quasi–static and ferent services and use different resources (no overlap) or when a
dynamic interaction needs to be made. In the short term, for mature common resource is a non-exclusive good, such as (non-patented)
technologies, many structural elements can be assumed to be fairly knowledge, or is in abundant supply. Another example is when two
static (Marshall, 1890). When the overlapping part of the bundles technologies are separated geographically, and thus may coexist
is assumed to be fixed, we use the term ‘quasi–static interaction’ while they in principle compete for the same market (Maréchal,
between technologies. For example, a resource flow used by two 2007). In biology, this kind of separation is called the Gause princi-
technologies is assumed to be constant, or the size of a market ple or competitive exclusion principle (Odum and Barrett, 2005).
shared by two technologies is assumed to be fixed. We use the Competition, symbiosis and neutralism are given a slightly dif-
term ‘dynamic interaction’ when technologies interact via struc- ferent meaning when we look at interaction dynamically and take
tural change in overlapping parts of the bundles. Such interaction structural change into account. This is in particular important for
takes into account effects of changing demand, production systems the interaction between two emerging technologies. Two emerg-
and knowledge pools. Since our starting point in this article is tech- ing technologies could compete to structure the downstream part
nical change and the relationship between emerging technologies, of value chains as well as upstream production chains. As a result
our main concern is dynamic interaction. However, quasi–static structures involving artefacts, actors and schemata could be created
interaction is in focus in much of the (neoclassical) economic liter- that fit one technology while the other is locked out. On the other
ature and can be also categorised with our terminology. hand, two technologies can together create a new market or new
Quasi–static interaction comes in the form of one of the first supply chains and share the burden of structural change – dynamic
three symmetric interaction types in Table 1: competition, sym- symbiosis. They may also evolve in different directions to avoid
biosis and neutralism. competition, creating new niches and differentiating resource sup-
Competition could be not only competition for markets, but also ply (Windrum and Birchenhall, 1998). They may thus develop a
competition for resources. Fig. 2 illustrates how, due to overlap, neutral relation.
technology Tb competes with Tc for the common market (or down- When we look at interaction dynamically we also find examples
stream system) Hn,1  , and how T and T compete for a common of the asymmetric interaction types in Table 1. We use the term par-
d e
input (or upstream system) Hn,1 . asitism for the situation when an emerging technology enters the
Symbiosis exists between technologies defined at different lev- market space developed by a more entrenched technology or can
els in the same value chain. The bundles ‘microprocessors’ and make use of the same upstream supply chains that were developed
‘computers’ clearly overlap in a symbiotic way. Such symbiosis by the older technology. If we for example view Tc in Fig. 2 as a new
could be necessary, or beneficial but not necessary. Also comple- technology, it can make use of the downstream structure Hn,1  that
mentary products (defined at the same level in a hierarchy) could was developed by Tb . Parasitism occurs when the new technology
gains from the existence of the old, while the old loses market share
or resource supply to the new. On the other hand, the old technol-
13
Pistorius and Utterback (1997) focus on competition, symbiosis and predation
ogy may also benefit from a resource or market niche developed
(here parasitism and predation). While we keep neutralism, commensalism and by the new technology or feed on more intangible assets such as
amensalism, we also leave out competition by direct interference, i.e. fighting, com- expectations. When the resource that is developed by one technol-
bine the two categories mutualism (interaction favourable to both and obligatory) ogy and made available for a second technology is a non-exclusive
and protocooperation (interaction favourable to both but not obligatory) into the
good such as non-patented knowledge, we instead have a situation
category ‘symbiosis’ and combine parasitism and predation into one category. In
the following we will use the term parasitism since we think it better reflects the of commensalism. In a situation when an emerging technology is
types of relationships we will describe. structurally locked out and does not fit into the system developed
408 B.A. Sandén, K.M. Hillman / Research Policy 40 (2011) 403–414

H’2,1
2 Told

1 P1,1 P1,2

0 P0,3 Tnew2
P0,1 P0,2 Tnew1

H0,3
-1 H-1,1 H-1,2 H-1,3

Fig. 4. An illustration of a bridging technology. A technology Tnew1 = [H−1,2 , P0,2 , P1,1 ,


 Fig. 5. Simultaneous competition and symbiosis. Two emerging technologies, Tnew1
H2,1 ] shares a first order downstream application (P1,1 ) with the established technol-
 and Tnew2 , compete for markets or resources, while at the same time expanding their
ogy Told = [H−1,1 , P0,1 , P1,1 , H2,1 ] and an upstream value chain (H−1,2 ) with Tnew2 = [H0,3 ,
 combined system (new markets or supply chains) at the expense of an established
P1,2 , H2,1 ] and can thus function as a bridging technology. For clarity, only alternative
alternative Told .
value chains are depicted.

around the old technology, we have amensalism where the new is on another technology in one part of the value chain and a negative
inhibited, while the old is not affected. Another case of amensalism effect in another part. The end result is not easy to foresee.
is when a bad reputation spills over from one technology to a less
known related technology or when one technology benefits from 2.4. Technological innovation systems – towards a
the decline of a second technology, which in turn is not affected by multidimensional model of diffusion
the first.
Bridging technologies can now be defined in terms of para- Our primary interest is the diffusion of novel technologies and
sitism (or commensalism). A bridging technology parasitizes on an how that diffusion is affected by technology interaction. This far we
established technology, while a third technology parasitizes on the have outlined that technologies can be described on the one hand as
bridging technology. In this way the third technology could grow bundles of value chains and on the other as socio-technical systems
even if it is in pure competition with the established technology extending in material, organisational and conceptual dimensions.
and should be structurally locked out. Taken together, these descriptions can be used to specify points
Fig. 4 provides a stylised example. An established technol- of overlap, i.e. points of interaction, e.g. a shared knowledge input
ogy, for example petroleum based transportation (Told ) is defined in a supply chain or a shared group of users. But to understand
by the value chain petroleum (H−1,1 ), gasoline (P0,1 ), combus- the dynamic processes during which interaction takes place, we
tion engine drive train (P1,1 ) and car transport (H2,1  ). The figure
also need a model of technology diffusion. We suggest here some
also show two emerging alternative technologies, a biofuel, basic building blocks of such a model, while leaving the detailed
 ], and renewable electricity propul-
Tnew1 = [H−1,2 , P0,2 , P1,1 , H2,1 elaboration (and illustration) for future studies. Again, our starting
 ]. The biofuel (T
sion, Tnew2 = [H0,3 , P1,2 , H2,1 new1 ) shares the first point is population ecology, complementing this with ideas from
order downstream application P1,1 (the internal combustion drive transition literature and an innovation system framework.
train) with gasoline (Told ), which is not the case for electricity The population ecology approach to diffusion observes that
(Tnew2 ) that shares only the second order downstream application the growth of technological and biological populations is gov-

H2,1 (car transport) and hence share fewer complementary supply erned by positive and negative feedback. The size of the population
chains. affects the growth rate. Technology diffusion in an early phase
It is difficult to develop a car propelled by renewable electricity stimulates further growth by increasing returns to adoption, i.e.
in a situation dominated by gasoline. However, if the biofuel par- positive feedback due to, for example, economies of scale and learn-
asitizes on gasoline benefiting from the shared drive train (P1,1 ), ing (Marshall, 1890, p. 265).14 When resources dwindle, market
renewable electricity can benefit from the development of H−1,2 , a potentials are exhausted or major drawbacks with the technology
common upstream process of biofuel and renewable electricity. In become apparent, negative feedback (decreasing returns to adop-
this example H−1,2 could be cultivation and gasification of biomass tion) starts to outweigh positive feedback and growth comes to a
that can generate a biofuel or renewable electricity. In this case, the halt. Combined positive and negative feedback generates the clas-
biofuel acts as a bridging technology between gasoline and renew- sical s-shaped curve of diffusion (Fisher and Pry, 1971; Geroski,
able electricity. Tnew1 parasitizes Told and Tnew2 parasitizes Tnew1 . In 2000). A more general dynamics is formalised in the Lotka–Volterra
this way, two emerging technologies can compete internally while functions, which also can take into account that growth rates may
at the same time helping one another (symbiosis), or one helping depend on other species (Porter, 1991, pp. 187–199). Variants of
the other (commensalism or parasitism), to grow at the expense of the Lotka–Volterra functions have been used to model technology
an established technology (Fig. 5). Taking the example in Fig. 4 one competition and substitution (Marchetti and Nakicenovic, 1979).
step further, renewable electricity use ([P0,3 , P1,2 ] or broader H0,3  ,
Pistorius and Utterback (1997) point out that the same framework
not visualised in the figure) could form a bridge between bioelec- can also be used to model symbiosis and parasitism/predation by
tricity production (H−1,2 ) and solar electricity production (H−1,3 ) changing the signs in the formulas (see signs in Table 1).15
completing a shift from a gasoline powered car [H−1,1 , P0,1 , P1,1 ,
 ] to a solar electric car [H
H2,1 
−1,3 , P0,3 , P1,2 , H2,1 ].
There is also the risk (from the perspective of the electric car)
14
that biofuels further strengthens the position of the internal com- See for example Arthur (1988) and Sandén and Azar (2005) for lists of feedback
mechanisms.
bustion drive train P1,1 at the expense of the electric drive train P1,2 15
Ahmadian (2008) models three-technology interaction in a Lotka–Volterra
and thus contributes to locking out renewable electricity propul- framework. He also explores the possibility of developing a formal model in a system
sion. Hence, one emerging technology may have a positive effect dynamics framework based on a conceptual innovation systems model.
B.A. Sandén, K.M. Hillman / Research Policy 40 (2011) 403–414 409

The simple population (community) ecology approach treats tion, e.g. entry or exit of actors or changed regulation, depends not
technologies as a one dimensional entity described by population only on this spillover effect but also on the state of other parts of
size. Taking the socio-technical systems approach we note that Ty and on exogenous forces.19
not only artefacts multiply, but also dedicated actors, networks, This type of multidimensional diffusion model is obviously com-
terminology, technical knowledge, expectations, positive attitudes plex. However, we believe it can work as a qualitative model that
and adapted regulations accumulate and expand. Observation of guides our thinking about the growth of socio-technical systems.
such multidimensional diffusion is of particular importance in the Together with empirical research, it will possibly enable identifi-
formative phase of socio-technical systems when the diffusion of cation of some general patterns of interaction and a limited set of
products has barely started. This phase can extend over many typical development paths that reduce complexity.
decades and decisive interaction between emerging alternatives
is bound to take place at this stage. The literature on technologi- 3. Alternative transport fuels in Sweden 1974–2004
cal innovation systems has identified several feedback loops that
do not directly involve production and use of products. One exam- The history of alternative transport fuels in Sweden displays
ple relates to knowledge formation: an early finding may lead to richness in terms of multimode and multidimensional interaction.
actor entry, further knowledge accumulation, increased expecta- This case will here be used to illustrate the actual complexity of
tions and a mobilisation of resources enabling further actor entry a process that often is viewed simply as a competition between
and knowledge production (Kamp, 2008; Sandén et al., 2008). A sec- a few alternative technologies. We also use the case to demon-
ond example involves legitimacy and regulation: expectations and strate how the terminology developed in this paper can be used
positive attitudes can lead to actor entry, which lead to increased to describe multimode and multidimensional interaction. A more
legitimacy and lobbying for changed regulation, which enable more complete description of the history is provided in Sandén and
actors to entry (Bergek et al., 2008b; Jacobsson and Bergek, 2004; Jonasson (2005).
Sandén, 2005). Suurs (2009) identifies four generic types of such Boundaries between technologies in this case is mainly based
feedback loops that he calls ‘motors of innovation’. on praxis, i.e. those technologies which are commonly viewed as
Not only endogenous forces (governed by feedback), but also separate alternatives in the current discourse and hence are given
exogenous forces, i.e. structures and events that do not depend on different names, are here considered to be different technologies.
the socio-technical system in focus may affect the development. In This delineation is found to be based on the chemical composition
the Lotka–Volterra model, exogenous factors are treated as con- of the fuel and upstream processes, or more correctly, the type of
stants. However, they may change over time. For example, the primary resource used, e.g. ‘ethanol from wood’ or ‘biogas’.
emergence of an international debate on climate change may affect The case contains interactions taking place between 1974 and
attitudes and expectations around a renewable energy technology 2004. During these years, several alternative technologies were
such as tidal power. It is reasonable to treat it as an external vari- present with varying intensity in different dimensions. The history
able since the development of tidal power has little influence on can be subdivided into three periods, each dominated by differ-
the climate change debate.16 Similarly, Rip and Kemp (1998) and ent external forces that influenced the course of events. In the
Geels (2002) describe system innovation or technological transi- first period, roughly from 1974 to 1985, large-scale oil substitu-
tions as a multilevel reconfiguration process, involving the three tion was the primary concern and methanol from gasified coal and
levels niche, regime and landscape. Novel technologies grow at the wood completely dominated the alternative fuel stage. The second
niche level. Their growth is governed by internal feedback but also period (1986–1997) saw the rise of ethanol, natural gas and bio-
by exogenous influences from the entrenched technological system gas in some geographical areas and a short visit of electric vehicles.
residing at the ‘regime level’ and from the even broader societal or The dominant external driving force was a focus on air pollution in
‘landscape level’.17 cities. From around 1998, climate change became an increasingly
Systemic growth can now be described by some ‘innovation sys- important issue, in later years complemented by rising oil prices.
tem functions’ that relate the growth (or decline) of the elements of A renewed interest in fuels from gasified biomass emerged, while
the system (and emergent system properties) to the system itself ethanol and biogas took steps towards large scale diffusion. Hydro-
and to external forces (Bergek et al., 2008b; Hillman et al., in press; gen has been present in all periods as an option for the distant
Hillman and Sandén, 2008; Sandén and Jonasson, 2005). This sys- future.20 All these alternatives to the entrenched fuels, petrol and
tem model of change has been termed a technological innovation diesel, are seen by many as competitors, but as we will see below,
system.18 The overlaps between different socio-technical systems other modes of interaction are also prevalent.
as described in previous sections can be inserted in this model.
When actors within socio-technical system Tx develop a system
3.1. Methanol and ethanol – parasitism and succession to the
element that is shared with socio-technical system Ty , or consumes
throne
an element (e.g. a limited resource) that is also required in Ty , this
will affect other elements of Ty . The final outcome of the interac-
During the 1980s, ethanol gradually overtook the role as the
prime alternative fuel from methanol. This came partly as a result

16
In a closed and interlinked world the only causes of change that are perfectly
exogenous are related to independent natural events such as radioactive decay and
19
meteorites from outer space. However, there is a scale from the more endogenous In a mathematical model of the innovation system the functions would be dif-
forces created within the studied system and exogenous forces that give an impulse ferential equations: dpy,i /dt = fy,i (py,1 , . . ., py,n , z1 , . . ., zm , ε), where py,i represent
from the outside. internal structural elements of technology Ty , zi represent external influences and
17
It is very difficult for niche actors, and even regime actors, to affect the land- ␧ is a stochastic perturbation. When a structural element is shared with technology
scape level. On shorter time scales the impact only goes in one direction. We may Tx (py,i = px,i ), the development of Tx influences the development of Ty . This inter-
observe that there is an even less mouldable level that has an effect on the evo- pretation opens for system dynamics modeling of innovation systems. Due to the
lution of socio-technical systems: the laws of nature. By this we do not mean our instability of innovation systems and difficulty with parameterization, such models
current formulations of the laws of nature, which can be considered to be part of are unlikely to add much to the predictive capacity of innovation system studies.
the landscape. However, formal models are useful for developing conceptual clarity and as tools
18
This interpretation differs slightly from earlier conceptualisations of technolog- for illustration and education (Sterman, 2000).
20
ical innovation systems (e.g. Bergek et al., 2008a; Hekkert et al., 2007). See also Electricity and battery electric vehicles did not re-enter the scene until after
Markard and Truffer (2008) for a review. 2004.
410 B.A. Sandén, K.M. Hillman / Research Policy 40 (2011) 403–414

of the shift in the main exogenous force of change from oil depletion became a bridge to wheat ethanol by the advancement of the idea
to air pollution, but the ethanol–methanol interaction in the 1980s that wheat ethanol was a bridge to wood ethanol! Interestingly,
also displays a case of commensalism developing into parasitism the interaction mode in the material dimension, in real applica-
where ethanol made use of downstream overlaps with methanol. tions, developed into neutralism. Wood ethanol was used for bus
Ethanol could benefit from being physically and conceptually fuel and E85, while wheat ethanol was used for low concentration
similar to methanol in multiple ways. The choice between the blends (E5).
words ‘methanol’, ‘ethanol’ or ‘alcohol’ seems to have been carefully As the groups around wood ethanol gained strength, it became a
made in many texts in the early half of the 1980s, not only for tech- political force of its own. In addition, the need for support from the
nical reasons. The term alcohol was used to borrow legitimacy in wheat producers decreased since ethanol from imported sugar cane
one way or the other, or to hide controversies, which in one period became the short-term option favoured by wood ethanol interests.
benefited both and illustrates a case of symbiosis based on con- The arguments for wheat ethanol as a necessary short-term option
ceptual overlap. The fight to change taxes (regulation) to become were thus weakened; wheat ethanol was increasingly seen as a
fair or even beneficial to alcohols had been fought by the methanol burden in the environmental debate. Possibly, this shifted the inter-
advocates but eventually came to benefit ethanol (commensalism). action mode from symbiosis to parasitism. If, on the other hand,
The standard blends of methanol and petrol (M100, M85, M15, M5 wheat ethanol had reached a point where it no longer needed wood
referring to the percentage of methanol) had become well-known ethanol, we instead of parasitism, have amensalism. The bad rep-
concepts. These concepts were taken over by ethanol (E100, E85, utation of wheat ethanol spills over to wood ethanol, while wheat
E5). Furthermore, technical knowledge and experience with alco- ethanol due to its organisational and material strength can continue
hol fuel and flexifuel vehicles (that can run on mixes of methanol as before!
and/or ethanol and petrol) had been gained and was kept within
the Swedish car manufacturers Volvo and Saab.21 The diffusion of 3.3. Biogas and natural gas – from commensalism to symbiosis
ethanol buses in the 1990s was made possible by the overlaps with
the preceding methanol system on the one hand and the differences Even for biogas and natural gas there is an influential overlap in
related to upstream resources and actor constellations on the other. that they share the same chemical configuration downstream, i.e.
Ethanol was considered inherently renewable while methanol was they both primarily consist of methane.22 This means that drivers
associated with coal and natural gas. can fill up their vehicles with either gas depending on what is avail-
At the end of the 1990s, when the need for large scale options able. From the start in the early 1990s, biogas was able to benefit
was emphasised, resulting in lowered expectations for ethanol and from the availability of technology and experience first developed
raised expectations for fuels from gasified biomass (e.g. methanol), for natural gas vehicles and filling stations. For technical knowl-
the Foundation for Swedish Ethanol Development (SSEU) was edge this interaction is a typical example of commensalism. Later
renamed to BAFF – BioAlcohol Fuel Foundation. This indicates the on, biogas and natural gas helped one another in the competition
renewed credibility for alcohols other than ethanol, i.e. methanol, with petrol and diesel through symbiosis regarding expectations
even if methanol was not really lobbied for by the organisation. The and physical artefacts. In addition, we find a touch of neutrality
existence of methanol as a concept once more benefited ethanol here; the choice between the two fuels was usually made in relation
and as long as ethanol is strong in material and organisational to local conditions regarding distance to the natural gas grid and
dimensions the conceptual overlap can be used without risk for availability of biogas (or of raw materials for biogas production).
real competition. Biogas was first introduced for vehicles in Linköping because it
was expected that the natural gas pipeline was going to be extended
3.2. Ethanol of various origins – symbiotic in mind and physically to that region. The first drivers and vehicles running on biogas were
neutral then supposed to shift to natural gas, i.e. biogas was considered a
bridge to natural gas regarding knowledge, actors and physical arte-
The interaction between different kinds of ethanol, mainly from facts. When it was decided at the national level that the natural gas
wheat and from wood, originates in the obvious downstream over- expansion was to be stopped, the use of biogas in Linköping was
lap. The fuels have the same properties in pure form (bus fuel) expanded. Thus, biogas had been introduced due to the physical
and when blended into petrol (E5 and E85). In addition, the same overlap with natural gas, and it was further adopted due to the lack-
physical artefacts can be used, such as buses, cars and filling sta- ing upstream overlap in terms of raw material and regulations. In
tions. Following from this downstream overlap, upstream actors retrospect, the relation of biogas to natural gas in Linköping can be
representing different kinds of ethanol contributed to the forma- categorised as commensalism in the realm of expectations (biogas
tion of the interest organisation SSEU (the Foundation for Swedish benefited from growing expectations for natural gas), followed by
Ethanol Development). The farmers’ organisation, which supported amensalism in the local setting but commensalism at the national
the construction of a wheat ethanol plant, recognised a common level (biogas benefited from that natural gas did not materialise
interest in ethanol with forest regions, which regarded supplying in Linköping at the same time as natural gas materialised in other
the raw material for wood ethanol as an opportunity. cities).
The resulting interaction mode was symbiosis in terms of chang- The modes of interaction in Linköping were commensalism
ing regulation and creating expectations. Already at an early stage, and amensalism, rather than parasitism and competition, since
the expectations of wheat ethanol as a long-term solution were the development of natural gas was not affected by the biogas in
low. Due to the abundance of forests in Sweden, the expectations Linköping at this stage. However, when the focus slowly shifted
for wood ethanol were higher. On the other hand, wood ethanol ini- from local air quality to climate change at the end of the 1990s, fossil
tially needed the support from the advocates around the short-term natural gas needed the connection to its renewable sister. Natural
option. It was argued that wheat ethanol could be used as a bridg- gas proponents argued both that natural gas was a bridge to biogas
ing technology awaiting the long-term option. Hence wood ethanol (see e.g. Sandebring, 2004), and that biogas could not survive eco-
nomically without teaming up with natural gas actors. Conversely,

21
Alcohol–petrol blends as well as flexifuel vehicles represent examples of physi-
22
cal downstream overlaps with the entrenched technology that made it possible for To acquire the quality needed for use in vehicles, so-called ‘natural gas quality’,
both methanol and ethanol to parasitize petrol. biogas needs extra upgrading (not necessary for use in stationary installations).
B.A. Sandén, K.M. Hillman / Research Policy 40 (2011) 403–414 411

to increase legitimacy for biogas as being part of a large scale solu- the development of a group of technologies sharing gasification and
tion, connections to the larger scale and economically stronger methanol use as downstream processes, in a later period, benefited
natural gas system were stressed. In addition, biogas could benefit a group of technologies sharing biomass supply and gasification as
from using natural gas as a backup in case of a shortage of biogas. upstream processes. The latter group gained from the existence
Hence, the relationship between biogas and natural gas was sym- of technical knowledge and a group of dedicated people that had
biotic in that they complemented each other in terms of attitudes ‘overwintered’ in the electricity domain. Due to the direction of
and expectations and shared downstream artefacts, knowledge and time, the former group was not affected by the latter, and hence we
regulation. However, as the climate change debate was intensified regard it as commensalism instead of parasitism. That the former
the connection to fossil natural gas became more of a burden for group did not return in a later period can be explained by exogenous
biogas, while natural gas still benefited from biogas (a shift towards forces (focus on climate change and carbon neutrality).
parasitism).
3.5. Renewable fuels and clean vehicles – concepts leading to
3.4. Syngas fuels – lagged commensalism parasitism, symbiosis and neutrality

Another group of interacting technologies are those sharing the From the 1990s, there have been several examples of inter-
process termed thermal gasification, collecting a large number of actions stemming from the treatment of alternative fuels as a
possible value chains at different stages of development.23 In prin- group, sometimes including ‘clean vehicles’,26 i.e. vehicles that can
ciple, the gasification process can convert carbon containing raw run on any alternative fuel or electricity, including hybrid elec-
materials, such as oil, coal, peat or biomass into a ‘synthesis gas’ (or tric, flexi-fuel and bi-fuel vehicles. These interactions highlight the
‘syngas’) composed of hydrogen and carbon monoxide.24 The syn- importance of concepts and point to overlaps regarding attitudes
gas can be used in several ways. The hydrogen can be separated and and regulations.
used directly, heat and electricity can be produced through com- In 1991, a political agreement on energy policy created a space
bustion of the syngas or the syngas gas can be used to synthesize for entrepreneurial experimentation in the form of a national
a range of different fuels, e.g. methanol, dimethyl ether (DME) and demonstration programme for biofuels in heavy vehicles. It was
Fischer–Tropsch (FT) diesel. Methane and ethanol are increasingly initially launched to investigate (and stimulate) ethanol, but an
put forward as other possible products from gasification. increasing share of the funding was used for biogas. The common
The first activities in Sweden related to gasification and trans- biological origin represented an important physical and concep-
port fuels appeared in the mid-1970s. Then a strong network of tual upstream overlap between biogas and ethanol that increased
actors was built up around R&D on gasification of various raw the resources available for biogas (parasitism).27 The unexpected
materials and the demonstration of methanol fuel blends used in emergence of biogas as an alternative transport fuel thus depended
vehicles. The gasification process as well as the use of methanol on the perfect timing of the commensalism–amensalism relation to
represented a downstream overlap for the different value chains. natural gas (outlined above) and the parasitic relation to ethanol.
In the 1980s and 1990s gasification received scarce attention in The national demonstration programme resulted in learning
the transport sector, and the overlap with the electricity system and changed the legitimacy for renewable fuels as a group among a
became important. The group of actors around gasification could large number of actors, including bus transit companies, municipal
maintain and increase their competence in the domain of elec- administrations and vehicle manufacturers (symbiosis). In addi-
tricity when the fuel domain was closed. At a later stage, when tion, consultants were hived off in the process and pushed for and
the large scale gasification of biomass gained increasing atten- facilitated continued diffusion of renewable fuels and clean vehi-
tion in the transport sector as a mitigator of climate change, cles.
these actors returned to promote not only methanol but also The introduction of cars that can run on alternative fuels was
other biofuels produced from synthesis gas.25 For these tech- made possible not only by the relative success of ethanol and
nologies, gasification together with the supply of biomass was methane in the bus niche, but also by the parallel development of
part of an upstream overlap. Physical artefacts from the electric- electric vehicle demonstrations. Electric vehicle tests in the early
ity system could also be used; a gasification demonstration plant to mid 1990s in the larger Swedish cities were considered total
developed for electricity production which had been taken out of failures. However, they created organisations that could host large
service was rebuilt to produce synthesis gas more suitable for fuel vehicle tests and that were interested in trying new things. This
synthesis. definitely paved the way for the new organisations clean vehicles
Regarding fuels and electricity, this would appear to be compe- in Stockholm and its sister organisation in Göteborg. Thus, electric
tition for actors and artefacts. However, considering the dynamics vehicles were a bridge to other technologies in terms of actors and
over time this is rather a case of alternating combinations of organisational routines. These city organisations were large enough
commensalism and amensalism. The decline in expectations for to test many types of vehicles and fuels, and lump them all together
gasification for fuel production in the 1980s and for electricity pro- under the hat of ‘clean vehicles’.
duction after 2000 was mainly due to exogenous forces. After the Various actor groups and networks that had been formed around
respective declines the group of experts oriented itself towards ethanol and biogas, as well as the farmers’ organisation lobbied for
the other application. In both shifts, the ‘new’ application was not more favourable policies, such as general tax exemptions that were
responsible for the decline of the ‘old’, while the ‘new’ was depen- to benefit several renewable fuels, mainly ethanol, biogas and RME
dent on the combined rise and fall of the ‘old’ (compare biogas and (rapeseed methyl ester). According to some, ethanol and methane
natural gas above). vehicles helped one another to stimulate the build-up of a mar-
Regarding the interaction between fuels, looking at the period ket for ‘clean vehicles’. They were still small compared to petrol
as a whole, gasification represents a case of commensalism where and diesel and were often lumped together as alternative trans-

23 26
Gasification of biomass is still at a stage of pilot and demonstration projects. The Swedish term for ‘clean vehicles’ is ‘miljöfordon’, which can be translated
24
Steam reforming of natural gas is another process used to produce synthesis gas. to ‘environmental vehicles’.
25 27
One of those was dimethyl ether (DME), which had been rediscovered as a diesel Within the programme, there were also a few projects and studies concerning
substitute at the end of the 1990s. DME and methanol.
412 B.A. Sandén, K.M. Hillman / Research Policy 40 (2011) 403–414

port fuels. In a small market for clean vehicles, the growth of one of ferment’ is clearly a too simple model of the real process. Also
type would raise the general awareness and legitimacy and thus the interaction modes symbiosis, neutrality, parasitism, commen-
benefit all other types. Thus, the conceptual overlap between all salism and amensalism appear in different technology relations
alternative fuels was important among general car users. How- from time to time. Two technologies can show different interac-
ever, they did compete for markets and political attention at the tion modes in different dimensions at the same time. For example,
municipal level, as most cities focused on one alternative fuel. Fur- they could be symbiotic in terms of knowledge development while
thermore, the vehicle manufacturers in Sweden looked for one best at the same time compete for favourable regulation. It is not an easy
alternative to include in their product portfolio. task to foresee the weighted outcome, and disclose the dominant
The division into ethanol and gas municipalities also created interaction mode. We further observe that the dominant interac-
a kind of neutrality where the parallel development of ethanol tion mode between two technologies tends to change over time.
and methane (biogas and natural gas) in the 1990s was facilitated These shifts are related not only to the growing maturity of the
by the existence of different geographical niches. After the turn socio-technical systems themselves but also to changing exoge-
of the century, another kind of neutrality emerged. A vocabulary nous forces, such as the shift from concern for oil scarcity to air
was developed that grouped renewable fuels into three categories: pollution and from air pollution to climate change. The emergence
first, second and third generations of renewable fuels. This can be and growth of one socio-technical system can make use of different
interpreted as an attempt to tame the competition for legitimacy, elements developed in several parallel systems. Hence, technolo-
and create neutrality based on separation of ‘time-niches’. “They all gies can act as bridging technologies in many ways and the exact
have a value, but on different time scales”. Another interpretation is timing is sometimes critical. The example of how biogas emerged is
that the conceptual grouping is part of a competitive relationship. a case in point. We also find that the apparent mode of interaction
It could for example be used to argue that the second generation change with the resolution of the observation. An interaction that
is better than the first while the third generation is too distant, can be classified as one mode when taking a longer time perspective
and that resources should therefore be focused on the second can be broken down into sequences of many different short-term
generation. interactions in different dimensions. What we observe, instead of
pure competition and selection between distinct alternatives, is a
multidimensional transformation process where ‘symbiogenesis’,
4. Conclusions the emergence of new species through the merger of ‘genetic mate-
rial’ of existing species (Margulis and Sagan, 2002), is as prevalent
The relationship between technologies has been a salient fea- as selection.
ture of the literature on technical change and terms like ‘dominant In conclusion, compared to more simplistic single mode interac-
design’ and ‘technology lock-in’ are part of the standard vocabu- tion models we have now a richer model of technology interaction
lary. In the growing literature on sustainable innovation, transition that opens for new empirical observations. Such observations can
management and climate policy the question of how to escape inform decision making. Even if it will remain difficult to anticipate
from dysfunctional locked-in systems while avoiding new dead- which overlaps that will be decisive in specific situations, anticipat-
ends are gaining increased attention, and hence, the understanding ing that spillovers will occur is a safe bet. Therefore, we believe that
of technology interaction becomes critical. The aim of this paper the model itself can inform policy and strategy making at a general
is to provide an account of the wide range of interaction modes level. The understanding that technologies commonly viewed as
beyond competition that is prevalent in transition processes and competitors not only compete but also strengthen one another calls
to develop a conceptual framework to facilitate more detailed and for a policy that attempts to foster many technologies in parallel
nuanced descriptions of technology interaction. and facilitates spillovers in multiple dimensions, rather than policy
The list of interaction modes between species borrowed from that only focuses on creating arenas for competition. The complex-
community ecology provides a helpful starting point. In this paper ity of technology development also calls for policy and strategy
we make use of six basic forms of interaction ranging from pure that makes use of the specific dynamics in different socio-technical
competition to pure symbiosis. In between, we find neutrality as systems and takes advantage of ongoing development in related
well as three forms of asymmetric relationships. As a refinement socio-technical systems. Policy, as well as firm strategy, needs to
compared to earlier work (Pistorius and Utterback, 1997), we make be technologically informed.
a couple of distinctions to identify the locus of interaction more pre- From these initial findings we see three pathways for further
cisely. First, the interaction (or overlap) can be localised in the value work. First, there is an empirical path. In this study we have cho-
chain. Technologies can share upstream production processes or sen to use only one case to illustrate how the framework can
downstream applications, giving rise to symbiotic as well as com- be used to describe technology interaction. If the framework is
petitive relationships. Depending on the purpose of the study, a applied systematically on many empirical cases it might be pos-
technology may be defined by a more or less limited bundle of sible to find repeated patterns. From these, more specific policy
value chains. Second, defining a technology as a socio-technical and strategy recommendations could be derived. Second, quan-
system allows for an even more detailed description of interac- titative modelling of technological innovation systems that takes
tion. There are material, organisational and conceptual dimensions interaction into account could be explored. Such an approach would
of socio-technical systems. The socio-technical systems can over- result in a system dynamics model that potentially could be used
lap in terms of physical artefacts, actors, technical knowledge, to play around with parameters to illustrate possible pathways and
expectations, attitudes, regulation and concepts. We further find general dynamics. Third, based on new empirical work, a refined
that a technological innovation systems framework could provide theoretical framework could be developed. Following from our dis-
a richer model for technology diffusion and technology interac- cussion on the arbitrariness of technology demarcation and the
tion than for example simple Lotka–Volterra equations. Such a multitude of overlaps between systems, one interpretation of this
model could in principle take into account multidimensional inter- text is that technologies separated by names are merely tempo-
action between socio-technical systems as well as influence of rary outgrowths on a socio-technical web. Similarly, organisations
exogenous forces. are outgrowths on the same web. Possibly, there are other units of
The case of alternative transport fuels in Sweden illustrates analysis that could be used to describe symbiogenesis and selec-
a great variety of interaction modes, involving all dimensions of tion in new ways to shed light on the multi-dimensional process of
socio-technical systems. The idea of competing designs in an ‘era technical change and its governance.
B.A. Sandén, K.M. Hillman / Research Policy 40 (2011) 403–414 413

Acknowledgements Hekkert, M.P., Suurs, R.A.A., Negro, S.O., Kuhlmann, S., Smits, R.E.H.M., 2007. Func-
tions of innovation systems: a new approach for analysing technological change.
Technological Forecasting and Social Change 74, 413–432.
This paper has a long history. We would like to thank all the Hillman, K.M., Nilsson, M., Rickne, A., Magnusson, T. Fostering sustainable
people interviewed during the collection of the empirical mate- technologies—a framework for analysing the governance of innovation systems.
rial in 2004 and 2005. Earlier versions have been presented at Science and Public Policy, accepted for publication.
Hillman, K.M., Sandén, B.A., 2008. Exploring technology paths: the development of
conferences and in a report. We thank all those who have com- alternative transport fuels in Sweden 2007–2020. Technological Forecasting and
mented on various versions and presentations. In particular, we Social Change 75, 1279–1302.
would like to thank Staffan Jacobsson for intense discussions and Hughes, T.P., 1987. In: Bijker, W.E., Hughes, T.P., Pinch, T.J. (Eds.), The Evolution of
Large Technological Systems. MIT Press, Cambridge, MA.
encouragement, Duncan Kushnir and Rickard Arvidsson for valu- Hume, D., 1740. A treatise of human nature, vol. 3.
able comments on late drafts, and two anonymous referees for Islas, J., 1997. Getting round the lock-in in electricity generating systems: the exam-
insightful and critical comments that substantially improved the ple of the gas turbine. Research Policy 26, 49–66.
Jacobsson, S., Bergek, A., 2004. Transforming the energy sector: the evolution of
text. We are grateful to the Swedish Energy Agency, Göteborg
technological systems in renewable energy technology. Industrial and Corporate
Energy Ltd. Research Foundation, CPM – The Competence Centre Change 13, 815–849.
for Environmental Assessment of Product and Material Systems, Kamp, L., 2008. Analyzing the introduction of renewable energy technologies in
Chalmers Environmental Initiative, Volvo Research and Educational The Netherlands with the FIS approach–possibilities, limitations and additions.
DIME International Conference “Innovation, sustainability and policy”, Bor-
Foundations (VREF) and Chalmers Energy and Transport Areas of deaux, France, September, pp. 11–13.
Advance which all at different stages contributed with financial Kemp, R., Rotmans, J., 2001. The Management of the Co-Evolution of Technical,
support to the project. Environmental and Social Systems, Towards Environmental Innovation Systems.
Garmisch Partenkirchen, Germany.
Marchetti, C., Nakicenovic, N., 1979. The dynamics of energy systems and the logis-
tics substitution model. International Institute for Applied Systems Analysis,
References Research Report 79-13, 71.
Maréchal, K., 2007. The economics of climate change and the change of climate in
Abernathy, W.J., Utterback, J.M., 1978. Patterns of industrial innovation. Technology economics. Energy Policy 35, 5181–5194.
Review 80, 40–47. Margulis, L., Sagan, D., 2002. Aquiring Genomes. Basic Books, New York.
Ahmadian, A., 2008. Exploring system dynamics modeling of technological inno- Markard, J., Truffer, B., 2008. Technological innovation systems and the multi-level
vation systems. Master thesis, Environmental Systems Analysis, Energy and perspective: towards an integrated framework. Research Policy 37, 596–615.
Environment. Chalmers University of Technology, Göteborg, Sweden. Marshall, A., 1890. Principles of Economics, 8th ed. Macmillan and Company Ltd.,
Andersson, B.A., Jacobsson, S., 2000. Monitoring and assessing technology choice: London.
the case of solar cells. Energy Policy 28, 1037–1049. Menanteau, P., 2000. Learning from variety and competition between technological
Arthur, B.W., 1988. Competing technologies: an overview. In: Dosi, G., Freeman, C., options for generating photovoltaic electricity. Technological Forecasting and
Nelson, R., Silverberg, G., Soete, L. (Eds.), Technical Change And Economic Theory. Social Change 63, 63–80.
Pinter Publishers, London, pp. 590–607. Murmann, J.P., Frenken, K., 2006. Toward a systematic framework for research on
Audretsch, D.B., Feldman, M.P., 1996. R&D spillovers and the geography of innovation dominant designs, technological innovations, and industrial change. Research
and production. American Economic Review 86, 630–640. Policy 35, 925–952.
Baldwin, C.Y., Clark, K.B., 2000. Design rules. The Power of Modularity, vol. 1. MIT Odum, E., Barrett, G.W., 2005. Fundamentals of Ecology. Thomson Brooks/Cole, Bel-
Press, Cambridge, MA. mont, CA.
Bergek, A., Jacobsson, S., Carlsson, B., Lindmark, S., Rickne, A., 2008a. Analyzing the Panzar, J.C., Willig, R.D., 1981. Economies of Scope. The American Economic Review
functional dynamics of technological innovation systems: a scheme of analysis. 71, 268–272.
Research Policy 37, 407–429. Pistorius, C.W.I., Utterback, J.M., 1997. Multi-mode interaction among technologies.
Bergek, A., Jacobsson, S., Sandén, B.A., 2008b. ‘Legitimation’ and ‘development of Research Policy 26, 67–84.
positive externalities’: two key processes in the formation phase of techno- Polanyi, M., 1958. Personal Knowledge: Towards a Post-critical Philosophy. Rout-
logical innovation systems. Technology Analysis and Strategic Management 20, ledge & K. Paul, London.
575–592. Porter, A.L., 1991. Forecasting and Management of Technology. Wiley, New York.
Bijker, W.E., 1995. Of bicycles, bakelites, and bulbs: Toward a theory of sociotechnical Porter, M., 1990. The Competitive Advantage of Nations. McMillan, London.
change. MIT Press, Cambridge, Massachusetts. Raven, R., 2005. Strategic Niche Management for Biomass. Technical University Eind-
Brittain, J.W., Wholey, D.R., 1988. Competition and coexistence in organizational hoven, Eindhoven.
communities, population dynamics in electronic component manufacturing. In: Rip, A., Kemp, R., 1998. Technological change. In: Malone, S.R.A.E.L. (Ed.), Human
Carroll, G.R. (Ed.), Ecological Models of Organizations. Ballinger, Cambridge, MA. Choice and Climate Change. Battelle Press, Columbus, US, pp. 327–399.
Carlsson, B., Jacobsson, S., 1993. Technological systems and economic performance: Rosenberg, N., 1976. Perspectives on Technology. Cambridge U.P., Cambridge.
the diffusion of factory automation in sweden. In: Freeman, D.F.C. (Ed.), Tech- Sabatier, P.A., 1988. An advocacy coalition framework of policy change and the role
nology and the Wealth of Nations. Pinter Publishers, London/New York, pp. of policy-oriented learning therein. Policy Sciences 21, 129–168.
77–94. Sandebring, H., 2004. Utredningen om förnybara fordonsbränslen: Slutbetänkande.
Carlsson, B., Jacobsson, S., Holmén, M., Rickne, A., 2002. Innovation systems: analyt- Fritzes, Stockholm.
ical and methodological issues. Research Policy 31, 233–245. Sandén, B.A., 2004. Technology path assessment for sustainable technology devel-
Carlsson, B., Stankiewicz, R., 1991. On the nature, function, and composition of tech- opment. Innovation: Management, Policy and Practice 6, 316–330.
nological systems. Journal of Evolutionary Economics 1, 93–118. Sandén, B.A., 2005. The economic and institutional rationale of PV subsidies. Solar
Clark, K.B., 1985. The interaction of design hierarchies and market concepts in tech- Energy 78, 137–146.
nological evolution. Research Policy 14, 235–251. Sandén, B.A., Azar, C., 2005. Near-term technology policies for long-term climate
David, P., 1985. Clio and the economics of QWERTY. Economic History 75, targets: economy wide versus technology specific approaches. Energy Policy
332–337. 33, 1557–1576.
Dosi, G., 1982. Technological paradigms and technological trajectories. Research Sandén, B.A., Jacobsson, S., Palmblad, L., Porsö, J., 2008. Assessment of the impact
Policy 11, 147–162. of a market formation programme on the Swedish PV innovation system.
Fisher, J.C., Pry, R.H., 1971. A simple substitution model of technological change. DIME International Conference “Innovation, sustainability and policy”, Bor-
Technological Forecasting and Social Change 3, 75–88. deaux, France, September, pp. 11–13.
Frankel, M., 1955. Obsolescence and technological change in a maturing economy. Sandén, B.A., Jonasson, K.M., 2005. Variety creation, growth and selection dynamics
American Economic Review 45, 296–319. in the early phases of a technological transition: the development of alterna-
Freeman, C., 1996. The greening of technology and models of innovation. Techno- tive transport fuels in Sweden 1974–2004. Environmental Systems Analysis.
logical Forecasting and Social Change 53, 27–39. Chalmers University of Technology, Göteborg, Sweden.
Geels, F.W., 2002. Technological transitions as evolutionary reconfiguration pro- Scott, W.R., 2001. Institutions and Organizations. Sage, Thousand Oaks, CA.
cesses: a multi-level perspective and a case-study. Research Policy 31, Sewell, W.H., 1992. A theory of structure: duality, agency, and transformation. Amer-
1257–1274. ican Journal of Sociology 98, 1–29.
Geels, F.W., 2004. From sectoral systems of innovation to socio-technical systems: Simon, H., 1962. The architecture of complexity. Proceedings of the American Philo-
insights about dynamics and change from sociology and institutional theory. sophical Society 106, 467–482.
Research Policy 33, 897. Stankiewicz, R., 2000. The concept of ‘design space’. In: Ziman, J.M. (Ed.), Tech-
Geels, F.W., 2005. Processes and patterns in transitions and system innovations: nological Innovation as an Evolutionary Process. Cambridge University Press,
refining the co-evolutionary multi-level perspective. Technological Forecasting Cambridge, pp. 234–247.
and Social Change 72, 681. Sterman, J., 2000. Business Dynamics: System Thinking and Modeling for Complex
Geroski, P.A., 2000. Models of technology diffusion. Research Policy 29, 603–625. World. McGraw-Hill, New York.
Grupp, H., 1996. Spillover effects and the science base of innovations reconsidered: Suurs, R.A.A., 2009. Motors of Sustainable Innovation – Towards a Theory on the
an empirical approach. Journal of Evolutionary Economics 6, 175–197. Dynamics of Technological Innovation Systems. Department of Innovation and
414 B.A. Sandén, K.M. Hillman / Research Policy 40 (2011) 403–414

Environmental Sciences, Copernicus Institute for Sustainable Development and van Lente, H., Rip, A., 1998. The rise of membrane technology: from rhetorics to
Innovation. Utrecht University, Utrecht, The Netherlands. social reality. Social Studies of Science 28, 221–254.
Tushman, M.L., Anderson, P., 1986. Technological dicontinuities and organizational Windrum, P., Birchenhall, C., 1998. Is product cycle theory a special case? Dom-
environments. Administrative Science Quarterly 31, 439–465. inant designs and the emergence of niches through coevolutionary-learning.
Utterback, J.M., 1994. Mastering the Dynamics of Innovation. Harward Business Structural Change and Economic Dynamics 9, 109–134.
School Press, Boston.

You might also like