New Visions of Technology

Technology Studies as a Field
“Technology studies” is the name for a loosely knit multidisciplinary
field with a wide variety of contributing disciplines, such as sociology,
history, cultural studies, anthropology, policy studies, urban studies,
and economics. Technology studies are concerned with the empirical
study of the development of technical artifacts, systems, and techniques
and their relation to society. Technology studies are part of science and
technology studies, or STS, a larger field that emerged in the 1970s and
that is based on studies of science and technology and their relation to
society that are both empirically informed and on sound theoretical
footing. STS is today an established discipline, with departments
and programs around the world, as well as specialized conferences and
journals.7
A full review of theories and approaches in technology studies is well
beyond the scope of this paper and is complicated because of the relative
youth of the field and the diversity of its topics and approaches. In what
follows, I focus on two subfields of technology studies that are at the
core of many STS departments and programs. They are social studies of
technology, which look at social and cultural aspects of technology, and
the history of technology, which studies the historical development of
technologies and their relation to society.8 In discussing the history
of technology, moreover, I focus on contextual approaches, which
are dominant in STS, and which look at the historical development oftechnologies in relation to their social context, instead of taking an
internalist approach that focuses on purely scientific and technological
contexts only.9 This selective choice means that I ignore, among other
studies, the important work that has been done in two distinct fields:
economics of technology and philosophy of technology.10
Contemporary technology studies, with their focus on social, cultural,
and historical dimensions, cover a wide variety of topics. Scholars rarely
consider “Technology-with-a-capital-T.” Instead, they examine specific
technologies, such as genetic engineering or nuclear technology; specific
engineering fields and approaches, such as mechanical engineering or
cold fusion research; specific techniques, such as rapid prototyping or
cerebral angiography; and technical artifacts, machines, materials, and
built structures, such as ceramic vases, Van de Graaff generators, polystyrene,
and the Eiffel Tower. In addition, many scholars study large
technological systems, such as railroad systems or early warning systems
in missile defense, and the processes of technological change, such as the
development of the bicycle in the nineteenth century or the invention
and development of electric lighting.
Technology studies analyze these technological entities in relation to
their social context. Roughly, this is done in one of three ways. In one
set of studies, the focus is on the shaping of the technology itself and the
role of societal processes. How did the technology come into existence?
What (social) factors played a role in this process? What modifications
has it undergone since it first came into being, and why did these occur?
In other studies, the focus is on how a technology has shaped society,
or, alternatively, on the social changes that accompanied the introduction
and use of the technology. In yet other studies, these processes are
considered together, emphasizing how a technology and its social context
co-evolve, or co-construct each other. A significant proportion of
work that takes up this co-construction theme even denies that there is a
meaningful distinction between technology and society, and attempts to
study “sociotechnology,” which consists of dynamic seamless webs of
entities that are only labeled as technological or social after they have
fully evolved (Bijker and Law 1992; Latour 1987; Callon 1987). There
is also a fourth category of studies in technology, which historian John
Staudenmaier (1985: p. 17) calls “externalist,” that do not focus ontechnology per se but only on contextual aspects, such as engineer-ing
communities, technological support networks, or public images of technology.
The core of contemporary technology studies consists of social studies
of technology and the history of technology, both of which have been
influenced by New Left critiques of science and technology. I discuss
these two subfields in order. In social studies of technology, the research
focus is on the social contexts in which technologies are developed and
used, such as engineering labs, factories, and homes. The research examines
how elements in these contexts interact with each other and with
the technology in question. Such elements include individual agents and
social groups, along with their behaviors, interactions, identities, and
statuses (gender, race, class), as well as organizational structures, institutional
settings, and cultural contexts.
Contemporary social studies of technology are in large part an outgrowth
of social studies of science. The specific tradition of which it is
an outgrowth is sometimes called sociology of scientific knowledge
(SSK). The SSK approach to the sociology of science, which is the dominant
approach today, holds that scientific knowledge itself, and not just
the social and institutional context of scientific inquiry, ought to be the
key focus of the sociology of science. SSK holds that scientific knowledge
is not a rational process exempt from social influences, but a social
process, and that scientific truth is not objectively given but socially constructed.
This SSK approach deviates from what was the dominant approach
in the sociology of science until the late 1970s: the Mertonian
approach, named after Robert K. Merton, which focused only on the institutional
context of scientific inquiry while assuming that scientific inquiry
itself is by and large rational and objective. SSK also distinguishes
itself from traditional (positivist) philosophy of science and epistemology,
which also holds scientific inquiry and truth to be rational and
objective. Instead, it takes its inspiration from philosopher of science
Thomas Kuhn’s work on the structure of scientific revolutions, which is
critical of images of science as a rational and cumulative process (Kuhn
1962).11
It was a founding principle of SSK that “nature” and “rationality”
and “truth” in science do not explain the process of scientific inquiry,but are themselves contingent social constructs that must be explained.
This central principle was extended in the early 1980s, when some
SSKers began to publish work in social studies of technology. The principle
is modified to read: the working of machines does not provide an
explanation of technological and social change, but is itself something
that must be explained, at least in part by investigating social agents,
their interactions, and their beliefs about technology.12 Technology is regarded,
in part or wholly, as a social construction that must be explained
by reference to social processes, and within which no appeal
can be made to objective standards of truth, efficiency or technological
rationality.
Although some contemporary work in (contextual) history of technology
finds inspiration in social studies of technology, the history of technology
is itself a much older field (Cutcliffe and Post 1989; Westrum
1991; Staudenmaier 1985; Fox 1999). Yet, although there has always
been an interest in the social context of technology in the history of
technology, approaches that put this social context at center stage have
only recently come to dominate. A typical study in a contextual history
of technology considers how a particular technology, such as electric
power transmission, the internal combustion engine, or the personal
computer, evolved historically and how the technology came to
reflect the contexts in which it has been developed and used. The investigation
is often bounded in time (a particular historical era or development
stage of the technology) and space (a particular geographical area
or setting). Contextual elements that such historians consider may include
organizational, policy, and legal settings, including relevant individual
actors, social groups, and organizations (engineers, firms,
industries, government bodies, activist groups) and their discourses and
behaviors. In sociohistorical studies of technology, in which social studies
of technology intersect with the history of technology, the development
of technologies is studied with special reference to their social
contexts and uses (see Bijker 1995b for a review).
Most studies in social studies and the history of technology are case
studies that consider particular settings or events in which technologies
are developed and used.13 Others are what John Staudenmaier (1985:
p. 206) calls “expanded studies,” which look more broadly at severaltypes of technologies or several types of settings or historical episodes.
Yet other studies are primarily theoretical or methodological, focusing
on such issues as technological determinism or the interpretive flexibility
of technological artifacts, or on methodological issues within technology
studies. Most studies operate at a micro or mesolevel of analysis, focusing
on individual actors, social groups and organizations, and their interactions,
rather than on the macrolevel of institutions and cultural
frameworks. The research methods are diverse and include textual
analysis, discourse analysis, participant observation, ethnomethodology,
and quantitative analysis.
Theoretical Claims of Technology Studies
The strong empirical orientation of most work in social studies and the
history of technology is visible, not only in its case analyses, but also in
its theoretical and methodological assumptions, which have often been
inspired by, or modified as a result of, these case studies. As a consequence
of this, there has been a fair amount of agreement on a number
of theoretical assumptions. I will try to characterize some of these assumptions,
along with some others that are also salient but more
controversial.
One of the most central theoretical assumptions in technology studies
is the assumption that technology is socially shaped. Technological
change is conditioned by social factors, and technological designs and
functions are the outcome of social processes rather than of internal
standards of scientific-technological rationality; technology is society
made durable.14 The social shaping thesis denies the technological determinist
idea that technological change follows a fixed, linear path, which
can be explained by reference to some inner technological “logic,” or
perhaps through economic laws. Instead, technological change is radically
underdetermined by such constraint and involves technological controversies,
disagreements, and difficulties that engage different actors or
relevant social groups in strategies to shape technology according to their
own insights.
Some scholars may discern technological or scientific constraints
on technological change, but others point out that such constraints, if
they exist at all, are themselves also socially shaped—for example,expectations of growth within the business, engineering, or user communities.
Also, while some scholars recognize separate stages in the development
of technology (e.g., invention, development, innovation),
others, particularly in social studies of technology, analyze technological
change as an entirely contingent and messy process, in which heterogeneous
factors affect technological outcomes, and in which the process of
invention continues after technologies leave the laboratory or factory.
These scholars emphasize that users, regulators, and others also affect
the design and operation of technologies and the way in which technologies
are interpreted and used (Bijker 1992; Lie and Sørenson 1996; Oudshoorn
and Pinch forthcoming). In contrast to a linear-path model of
technological change, proposals have been made for a variation and selection
model, according to which technological change is multidirectional:
there are always multiple varieties of particular design concepts,
of which some die, and others, which have a good fit with social context,
survive (e.g., Pinch and Bijker 1987; Ziman 2000).
The social-shaping thesis implies a weak constructivist claim that
technological configurations are variable and strongly conditioned by
social factors. Social constructivist approaches go beyond this claim to
arrive at the strong constructivist claim that technological change can be
entirely analyzed as the result of processes of social negotiation and interpretation,
and that the properties of technologies are not objective,
but are effectively read into the technologies by social groups. Social
constructivism is hence a contemporary form of idealism, denying the
possibility or desirability of a reference to any “real” structures or forces
beyond the representations of social groups. Whether a certain technology
works or is efficient or user-friendly, and the nature of its functions,
powers, and effects is not a pregiven, but the outcome of social processes
or negotiation and interpretation.15
Those social-shaping theorists who do not embrace social constructivism
also recognize that the meaning or use of technologies is not pregiven.
Most theorists agree that technology has interpretive flexibility,
meaning that technologies can be interpreted and used in different ways
(Pinch and Bijker 1987). When social negotiations surrounding technological
change come to a close, interpretive flexibility is held to diminish
because the technology stabilizes, along with concomitant (co-produced)meanings and social relations. Stabilization implies the embedding of
the technology in a stable network consisting of humans and other technologies,
and the acceptance of a dominant view on how to interpret
and use the technology. Stabilization of a technology implies that its
contents are “black-boxed” and are no longer a subject of controversy.
Its stabilized properties come to determine the way that the technology
functions in society. Yet, black boxes can be reopened. The history of
technology shows how technologies such as the telephone, the Internet,
or the automobile take on particular functions or societal roles that may
vary from time to time and place to place.
The flip side of the claim that technology is socially shaped is the
claim that society is technologically shaped, meaning that technologies
shape their social contexts. This goes considerably beyond the claim that
new technologies may open up new possibilities that change society, or
that technologies may have side effects. Obviously, the steam engine
changed society by making new types of industrial production possible,
and the printing press effected change by making written information
more available and easier to distribute. Obviously, also, technologies
may have side effects such as environmental pollution or unemployment.
The technological-shaping thesis refers not just to such recognized
functions and side effects of technologies, but to the multiplicity of functions,
meanings, and effects that always, often quite subtly, accompany
the use of a technology. Technologies become part of the fabric of society,
part of its social structure and culture, transforming it in the process.
The idea of society as a network of social relations is false, because society
is made up of sociotechnical networks, consisting of arrangements of
linked human and nonhuman actors.
The notion of a sociotechnical network is a central notion in actornetwork
theory (ANT), which is a third influential approach to technology
studies, next to the social-shaping and social-construction approaches.
It studies the stabilization processes of technical and scientific objects as
these result from the building of actor networks, which are networks of
human actors and natural and technical phenomena. Actor-network
theorists employ a principle of generalized symmetry, according to
which any element (social, natural, or technical) in a heterogeneous
network of entities that participate in the stabilization of a technologyhas a similar explanatory role (Callon 1987; Latour 1987; Callon and
Latour 1992). Social constructivism is criticized by ANT for giving special
preference to social elements, such as social groups and interpretation
processes, on which its explanations are based, whereas natural or
technical elements, such as natural forces and technical devices, are prohibited
from being explanatory elements. Actor-network theory allows
technical devices and natural forces to be actors (or “actants”) in networks
through which technical or scientific objects are stabilized. By an
analysis of actor networks, any entity can be shown to be a post hoc
construction, but entities are not normally socially constructed because
stabilization is not the result of social factors alone.
The notion that society is technologically shaped means, according to
most scholars in technology studies, that technology seriously affects
social roles and relations; political arrangements; organizational structures;
and cultural beliefs, symbols, and experiences. Technology scholars
have claimed that technical artifacts sometimes have built-in political consequences
(Winner 1980), that they may contain gender biases (Wajcman
1991; Bray 1997), that they may subtly guide the behavior of their users
(Sclove 1995; Latour 1992), that they may presuppose certain types of
users and may fail to accommodate nonstandard users (Akrich 1992) and
that they may modify fundamental cultural categories used in human
thought (Turkle 1984, 1995).
Latour (1992), for example, discusses how mundane artifacts, such as
seat belts and hotel keys, may direct their users toward certain behaviors.
Hotel keys in Europe often have heavy weights attached to compel
hotel guests to bring their key to the reception desk upon leaving their
room. Winner (1980) argues that nuclear power plants require centralized,
hierarchical managerial control for their proper operation. They
cannot be safely operated in an egalitarian manner, unlike, for example,
solar energy technology. In this way, nuclear plants shape society by
requiring a particular mode of social organization for their operation.
Sclove (1995) points out that modern sofas with two or three separate
seat cushions define distinct personal spaces, and thus work to both respect
and perpetuate the emphasis of modern western culture on individuality
and privacy, in contrast to, for example, Japanese futon sofa-beds.
Finally, Turkle (1984) discusses how computers and computer-operatedtoys affect conceptions of life. Because computer toys are capable of behaviors
that inanimate objects are not normally capable of, they lead children
to reassess the traditional dividing lines between “alive” and “not
alive” and hence to develop a different concept of “alive.” Most authors
would not want to claim that technologies have inherent power to effect
such changes. Rather, it is technologies in use, technologies that are already
embedded in a social context and that have been assigned an interpretation,
that may generate such consequences.
To conclude, the major insights of technology studies have been that
technologies are socially shaped and at the same time society is shaped
by technology, or, alternatively, that society and technology co-construct
each other. They are not separate structures or forces, but are deeply interwoven.
Moreover, technological change is not a linear process but
proceeds by variation and selection, and technologies have interpretive
flexibility, implying that their meanings and functions and even (according
to social constructivists) their contents are continually open to renegotiation
by users and others.
Technology Studies and Modernity Theory: Mutual Criticism
The Treatment of Technology in Modernity Theory
It is difficult to overlook the pervasive role of technology in the making
of modernity. As argued earlier, technology is a central means by which
modernity is made possible. It is a catalyst for change and a necessary
condition for the functioning of modern institutions. However, it is
more than that. What can be learned from technology studies is that the
institutions and culture of modernity are not just shaped or influenced
by technology, they are also formed by it. The social systems of modernity
are sociotechnical systems, with technology an integral part of the
workings of social institutions. Social institutions are societal structures
that regulate and coordinate behavior and in this way determine how
certain societal needs are met. In the modern age, however, their regulative
functions are no longer a direct outcome of collective actions, since
most collective actions have become thoroughly mediated and shaped by
modern technologies, which function as co-actors. For example, collective
acts of voting are now thoroughly mediated by voting technologiesthat help determine whether people get to vote at all, how votes are defined,
and whether votes are counted. Modern culture is, likewise, a
technological culture, in which technologies are not just material substrates
of existing cultural patterns, but also have a major role in defining,
shaping, and transforming cultural forms. Information technology,
for example, is transforming basic cultural concepts and experiences
such as those of time, space, reality, privacy, and community and is also
effecting fundamental shifts in cultural practices.
If this analysis of the role of technology in modernity is anywhere
near correct, then it is surprising, to say the least, to find that technology
is not a central topic in the vast literature in modernity theory. Indeed,
of the many hundreds of books that bear the word “modernity” in the
title, fewer than a handful also refer to technology or one of its major
synonyms or metonyms (e.g., technological, computers, biotechnology,
industrial).16 Many of the major works in modernity theory make only
passing reference to technology. For example, technology is referenced
only once in the recent edited volume, Theories of Modernity and Postmodernity;
it is not mentioned at all in Zygmunt Bauman’s Intimations
of Postmodernity; and there are only four or five brief references to it
in Alain Touraine’s Critique of Modernity (see Turner 1990; Bauman
1992; Touraine 1995).
What can explain this apparent neglect of technology in modernity
theory? It is not denial that technology has an important role in the constitution
of modernity, for most authors would agree that its role is pivotal.
A better explanation is that the dominant dimensions along which
modernity has traditionally been analyzed (institutional, cultural, and
epistemological) have not allowed technology to play a major identifiable
role, but have instead assigned it the status of a background condition.
Technology is often analyzed as a mere catalyst of institutional,
cultural, and epistemological change, or as a mere means through which
institutions, cultural forms, and knowledge structures are realized.
In institutional analyses, modernity is analyzed as being constituted
by institutions and their transformations. Technology is not usually recognized
as an institution itself; it is not seen as a separate regulative
framework such as capitalism, government, or the family, but rather as
one of the means through which these frameworks operate. More oftenthan not, institutions such as capitalism, industrialism, or military
power are discussed without specific reference to the technologies that
sustain them. The role of technology in transforming these institutions
(e.g., in the transition to an information society) is more difficult to ignore.
However, here one often finds technology subsumed as part of a
broader phenomenon, such as rationalization (Weber), productive
forces (Marx), or disembedding mechanisms (Giddens), of which it is
only a part. Even in Marxist theory, which assigns an important role to
production technology in the making of modernity, this technology still
only serves as an external constraint on economic structure, which ultimately
determines the social forms of society.
In most cultural and epistemological theories of modernity, technology
is either analyzed as a mere catalyst of cultural and epistemological
changes, or it is robbed of its materiality and reduced to knowledge, language,
or ideas. In Heidegger’s critique of modernity, in which technology
“enframes” us and turns the world into “standing reserves,” technology
turns out not to be defined as a material process or as a mode of action,
but as a particular mode of thinking (Heidegger 1977). The same idealism
is also visible in much of critical theory, in spite of its greater emphasis
on social institutions. There, technology is often identified with
technological or formal rationality, which is a mode of thinking that
characterizes not only modern technology but also modern thought and
economic and social processes. Habermas, moreover, has defined technology
as “technological knowledge and ideas about technology”
(Habermas 1987: p. 228). Finally, in postmodern theory, technology is
often reduced to language, signs, or modes of knowledge, along with
everything else.
When technology is referred to in modernity theory without being reduced
to something else, still other problems emerge. One is the level of
abstraction at which technology is discussed: technology is usually
treated as a monolith, as a macroscopic entity, Technology-with-acapital-
T, about which broad generalizations are made that are supposed
to apply equally to nuclear technology and dental technology, to
vacuum cleaners and gene splicers. This abstract, undifferentiated treatment
leads to vagueness, obscures differences between technologies, and
fails to distinguish the varied ingredients that make up technology(knowledge, artifacts, systems, actions) and the way these relate to their
context.
Giddens, for example, employs the notion of an “expert system,”
which is a key mechanism for decontextualizing social relations. He defines
expert systems as “systems of technical accomplishment or professional
expertise that organize large areas of the material and social
environments in which we live today” (Giddens 1990: p. 27). He discusses
few examples of expert systems, but makes it clear that virtually
any system in which the knowledge of experts is integrated and that
contains relevant safety measures qualifies as an expert system, including
automobiles, intersections, buildings, and railroad systems. Moreover,
Giddens goes into hardly any detail on the way in which expert
systems decontextualize social relations.
A monolithic treatment of technology easily leads to essentialism and
reification. In an essentialist conception, technology has fixed, contextindependent
properties that apply to all technologies. As Andrew Feenberg
(1999a: pp. viii–ix) has argued, technological essentialism usually
construes technology’s essence as its instrumental rationality and its
functionalism, which reduces everything to functions and raw materials.
This essentialism often correlates with a reified conception of technology,
according to which it is a “thing,” with static properties, that interacts
with other “things,” such as the economy and the state. Essentialism
and reification, in turn, have a tendency to promote technological determinism,
in which technology develops according to an internal logic,
uninfluenced by social factors, and operates as an autonomous force in
society, generating social consequences that are unavoidable.17 Technological
determinism is evident in dystopian critiques of modernity, such
as those of Heidegger, Marcuse, and Ellul, in which technology engulfs
humanity and rationalizes society and culture. In many other theories of
modernity, it is also present, albeit in a more subtle way. Marx’s thesis
that the productive forces determine or constrain production relations
has often been interpreted as a form of technological determinism.
Daniel Bell (1976) presents a similar view in characterizing the transition
to a postindustrial society as the result of economic changes that
are due to increased productivity, which is conditioned by information
technology. Baudrillard (1995) construes the transition from modernityto postmodernity in technological determinist terms by analyzing it as
the result of information technology and media, whose models and
codes yield a new social order. James Beniger’s (1989) detailed historical
study of the making of the information society is also built on technological
determinist principles, with technological change being a cause
of social change, while itself remaining relatively independent of social
influences.
In conclusion, the treatment of technology in modernity theory is
problematic in several respects. Often, technology is not assigned a
major role in modernity; it is subsumed under broader or narrower phenomena
or one-dimensional phenomena; its treatment is often abstract,
leading to vagueness, overgeneralization, detachment from context, and
a failure to discern detailed mechanisms of change. In addition, technology
is often reified and essentialized, and the conceptions of it are often
deterministic. There is also the problem that modernity theory’s sweeping
generalizations about technologies do not normally rest on microlevel
elaborations of the macro theory or on case studies. Modernity
theory’s generalizations, it will be clear by now, tend to go against many
key ideas of technology studies (the social character of technology and
its interpretive flexibility, the path dependence of technological change,
etc.). Moreover, when theories of modernity provide inadequate accounts
of technology and its role in modernity, their accounts of social
institutions, culture, and the dynamics of modernity suffer as well.
There are theories that avoid many of the problems listed (e.g., Castells
1996), but they are exceptions to the rule.
The Treatment of Modernity in Technology Studies
Modernity theory must provide an account of technology because of its
major role in the shaping of modernity. Technology studies, on the contrary,
do not seem to require a consideration of modernity in their
analyses of technology. It is not obvious that a historical study of the
telephone or an analysis of the development and advertisement of fluorescent
lighting must refer to macroscopic structures and events such as
disembedding mechanisms and changes in capitalist production modes.
And in fact, most work in technology studies does not refer to such
macro structures but instead remains at the micro (or meso) level. Itstudies actors (individuals, social groups, organizational units); their
values, beliefs, and interests; their relations and (inter)actions; and the
way in which these shape or are shaped by specific technologies. Case
studies and extended studies based on this approach contain rich descriptions
of complex dynamics that lead to social and technological
outcomes. However, the aim of many of these studies is not just to describe
what happens, but also to explain why it happened. For example,
in analyzing the history of the Penny Farthing bicycle, Pinch and Bijker
(1987: p. 24) do not want to only describe various bicycle models and
the social groups involved in their manufacture and use; they want to
understand the factors that determine what models are successful and
the reasons social groups assign certain meanings to a model. I argue
that microlevel accounts cannot fully explain technological and social
change unless they are linked with macrolevel accounts.
The main reason for this is that a sufficiently rich account of actors
and their relationships, beliefs, and behaviors requires an analysis of the
wider sociocultural and economic context in which these actors are
operating. This broader analysis is needed to explain why actors have
certain attitudes, values, beliefs, or relationships, and it may even be
necessary to infer their very existence. For example, an understanding of
why certain types of men were attracted to high-wheeled bicycles in late
nineteenth-century England, and perhaps also the identification of social
groups with this attraction, is likely to require an account of masculine
culture in late nineteenth-century England. The failure to look at this
cultural context would result in superficial and possibly also unreliable
descriptions of actors. More generally, to base explanations of technological
and social change merely on observations of actors and
their behaviors would be to subscribe to a form of methodological individualism,
a questionable form of reductionism that holds that social
explanations can be reduced to facts about individuals and hence that
no reference to supraindividual social structures is required (Lukes
1994).
Granted, the actors in technology studies also include more complex
entities, such as social groups and organizations, and nonhuman actors
such as machines, but these are still particular actors to which agency is
attributed, frequently along with beliefs and attitudes. If the actions,beliefs, and attitudes of these actors are not related to wider sociocultural
contexts, then any explanation is likely to fall short. This is a recurring
problem in most approaches in technology studies that emphasize
an actor perspective, including social-shaping and social-constructivist
approaches and the actor-network approach of Bruno Latour, Michel
Callon, John Law, and their associates. This latter approach does relate
the properties of individual actors to a wider context, which is the network
of actors in which they are operating, and holds that this network
defines these properties. However, the networks are limited in scope, usually
containing only the actors thought to have a direct role in the development
or functioning of a particular technology. Actor-network studies
rarely provide sufficiently complete accounts of the networks that shape
the behaviors or attitudes of the other actors in the network (e.g., engineers,
corporations, or politicians), who therefore tend to be analyzed in a
methodological individualist way.18
There is also another reason microlevel approaches have only limited
explanatory power. As Paul Edwards points out (chapter 7, this volume),
a major distinguishing feature of modern societies is their reliance
on infrastructures, large sociotechnical systems such as information and
communications networks, energy infrastructures, and banking and
finance institutions. As Edwards argues, these infrastructures mediate
among the actors that are studied in microlevel analysis. In this sense
they function as disembedding mechanisms, defining social relations and
guiding social interactions over large distances of time and space (Brey
1998). However, these infrastructures themselves are best studied at the
macrolevel. Microlevel approaches that ignore infrastructures run the
risk of providing an insufficient account of the relations between actors
in modernity (whereas accounts of social relations in premodern societies
can more easily remain at the microlevel because they are not usually
mediated by infrastructures). The recent transition to a post-Fordist,
global economy has heightened the inadequacy of microlevel analyses
by fragmenting industrial production and marketing and reorganizing it
on a global scale (Rosen 1993).
Social constructivists, while acknowledging the need to consider
the societal context in which actors operate, have sometimes objected
to an appeal to social theory because of its “realism,” which would beincompatible with (strong) social constructivism (see, e.g., Pickering
1995; Elam 1994) However, there is no inconsistency in invoking categories
of social theory in social constructivist analyses. Social constructivist
explanations proceed by deconstructing entities in terms of the
activity of other entities, specifically social groups. These entities are
often not deconstructed themselves for pragmatic reasons because deconstruction
has to stop somewhere. For instance, Bijker’s (1992) social
constructivist analysis of fluorescent lamps refers to the involvement of
General Electric as a “real” entity. As Bijker (1993) later explained, his
primary interest had been the social construction of fluorescent lamps,
and not the social construction of General Electric. Because of this specific
interest, it was excusable to present some parts of the sociotechnical
world as fixed and as undeconstructed entities that function in the explanation
of the development of fluorescent lamps, even though these
entities are social constructions as well. But if reference can be made to
General Electric in social explanation, then surely reference can be make
to Fordism, disembedding mechanisms, and other socially constructed
entities of social theory.19
Another criticism of modernity theories, and a reason cited for avoiding
them, is their alleged tendency to totalization, universalization, functionalism,
rationalism, panopticism, and determinism, not just in their
treatment of technology, but in their treatment of society as a whole.
This mirrors the criticism by postmodernists of macrolevel metanarratives.
Tom Misa has argued, for instance, that macrolevel theories tend
to “impute rationality on actors’ behalfs or posit functionality for their
actions, and to be order-driven,” and that these tendencies quickly lead
to “technological, economic or ecological determinism.” Microlevel
studies, instead, focus on “historical contingency and variety of experience”
and are “disorder-respecting” (Misa 1994: p. 119). While the
former tendencies are clearly visible in the majority of theories of
modernity, I hold that they are not inherent to macrotheorizing. The
macrostructures postulated in macrotheories inevitably impose constraints
on the actions of individuals, but this does not mean that they
must also determine these actions. Moreover, macrostructures can be
defined as contingent, heterogeneous, and context dependent, such as
Castells’ networks.A final objection to macrotheories is that they are often speculative
and not elaborated or tested empirically. While there are good exceptions
(again, Castells), these virtuoso performances confirm the general
rule. My point is that this category of theorizing should not be rejected,
but instead that these theories should be developed, tested, and refined. I
conclude, tentatively, that there are no good reasons for scholars in
technology studies to avoid macrotheories of modernity and that there
are good reasons to employ them. Working toward integrated studies of
modernity and technology involves, then, developing and testing
macrotheories and working to bridge the micro-macro gap that now
often separates modernity theory from technology studies. These two
tasks are the topic of the next section.
Modernity, Technology, and Micro-Macro Linkages
The Problem of Micro and Macro
In large part, the problem of connecting the topics of modernity and
technology, and of connecting modernity theory with technology studies,
is the problem of connecting the macro with the micro. Modernity
theory typically employs a macrolevel of analysis, analyzing macrolevel
phenomena, such as late modernity and globalization, in terms of other
macrolevel phenomena, such as time-space disembedding and the gradual
decline in Western global hegemony. Much work in technology
studies operates at the micro level, analyzing microlevel entities such as
fluorescent lighting or the advertising of a new daylight fluorescent lamp
by reference to other microlevel entities such as “the influence of Ward
Harrison of the incandescent lamp department of General Electric” or
“the writing of a report on daylight lighting by the Electrical Testing
Committees.” In addition, one could claim that modernity theory typically
employs a structure perspective, focusing on social structures
and their properties, whereas technology studies often employ an actor
perspective.
I assume that there is a mutual need in technology studies and modernity
theory to bridge the gap between the micro and the macro, and between
structure and actor perspectives. Nevertheless, the problem of
micro and macro (not to mention the problem of structure and agency)remains one of the great unsolved problems in social science. In spite of
the attention this problem has generated, there is still no recipe, no
method, and few inspiring exemplars on how to connect macrolevel and
microlevel analyses. In the discussion that follows, I try to advance this
general issue by looking more analytically at the problem. I argue that
progress on the micro-macro problem has been hampered by a failure to
recognize the multiplicity of levels of analysis (micro and macro being
coarse distinctions only) and a failure to distinguish two distinct dimensions
within the micro-macro distinction: size and level of abstraction. I
then outline four principal ways in which levels of analysis may map
onto each other, and conclude by drawing implications for an integration
of modernity theory with technology studies.
Size and Level of Abstraction
What makes a phenomenon studied in the social sciences or humanities
a macro phenomenon? And what makes a concept a microlevel concept?
Considerable confusion exists over this matter. Sometimes it is held that
macroanalysis is distinct from microanalysis because it focuses on larger
things. Social systems are large and individuals and their actions are
small; therefore social systems are the subject of macroanalysis and individuals
the subject of microanalysis. Another claim sometimes made
about the micro-macro distinction is that macrolevel phenomena and
the concepts that refer to them are abstract and general, whereas microlevel
phenomena tend to be concrete and specific.
Thus there are at least two parameters along which macroanalyses are
distinguished from microanalyses: the size of the units of analysis, and
their level of abstraction.20 Very few attempts exist in the literature to
further define or operationalize these parameters, or to study their interrelationships.
It is usually assumed that they tend to interrelate; that the
units of macrolevel analysis are typically, if not invariably, large, abstract,
and general, whereas things in microlevel analysis tend to be
small, concrete, and specific. Yet there are many exceptions to this rule.
For example, the modern self is both smaller and more abstract than
protest marches during the inauguration of George W. Bush. The
modern self is a smaller unit of analysis because protest marches involve
many modern selves. It is more abstract because it refers to a generaltype of phenomenon, whereas the protest marches denote a specific type
of phenomenon. Other units of analysis do not seem to have a definite
size at all. For example, reflexivity is a property that can apply to both
large things (e.g., social systems) and small things (e.g., the knowledge
processes of organizational units).
To understand the connections between the micro and the macro, we
must therefore first better understand the parameters by which these notions
are defined—the notions of level of abstraction and size. I discuss
these in order.
Level of Abstraction What does it mean to say that one phenomenon
is more abstract than another? Principally, I want to argue, this means
that the phenomenon is more general. For example, rationalization is a
more general process than the standardization of testing in aviation
schools (a form of rationalization), and that is why it is more abstract.
Starting from an abstract phenomenon, one can arrive at more concrete
phenomena by introducing additional properties that bound it. Starting
with the abstract phenomenon of industrial society, one can arrive at the
more specific and therefore more concrete phenomenon of late nineteenth-
century British industrial society by adding properties that specify
time period and nationality. Likewise, starting from the notion of a
parent, one can arrive at the somewhat more concrete notion of a
mother (a female parent) by adding a gender property (female). Conversely,
when one starts with a concrete phenomenon, one can arrive at
a more abstract one by removing properties from it. One can go from
late nineteenth-century British industrial society to industrial society and
from mother to parent by subtracting properties, that is, by generalizing.
In this way it is possible to construct hierarchies of entities that range
from abstract to concrete, with the more concrete entities being species
(subtypes or instances) of the more abstract entities. For example, one
can construct a hierarchy going from transportation vehicle to bicycle
to Penny Farthing bicycle to the Bayliss-Thomson Ordinary Penny Farthing
bicycle to the specific Bayliss-Thomson Ordinary Penny Farthing
bicycle of which I have a picture. Notice, however, that concretization is
not just a matter of adding adjectives (or abstraction a matter of subtracting
them). The relation between more abstract and more concretephenomena is not always linguistically transparent, and conceptual
analysis, if not empirical investigation, may be needed to observe the relation
(e.g., that a mother is a type of parent, or that a standard-setting
body in health care is a type of bureaucratic organization).
Size Units of analysis can often be ordered according to their size. For
example, a social system is obviously larger than a social group in that
system, and a social group is larger than an individual in that group. The
reference to size here does not imply a reference to absolute metric or numerical
properties. Rather, size is used here in a relative or comparative
sense. Phenomenon a is larger than phenomenon b if a can contain b, or b
is a part of a. For example, there are part-whole relations between the
economy and the individuals participating in it because an analysis of
economic processes ultimately reveals individuals engaged in economic
behavior. This is why economic systems are larger than individuals. Large
units of analysis are larger than small units of analysis because they are
able to stand in a part-whole relation or a relation of (partial) containment
to these smaller units (Castells 1996: pp. 174–179). Because parts
may have parts themselves, hierarchies can be constructed of units of
analysis that range from large to small. For example, a social system may
include a market system that includes organizations that include organizational
units that include divisions that include employees. Likewise, the
British railway system includes train stations that include platforms and
station staff. Also, items may be parts of multiple wholes. For example,
pay-per-view virtual museums may be part of the post-Fordist economy,
but they are also part of postmodern culture. Notice that part-whole relations
between units of analysis, which refer to their size, are clearly different
from the types of genus-species relations discussed earlier, which refer
to level of abstraction. For instance, Internet advertising is a species of advertising,
but a part of the post-Fordist economy.
Levels of Analysis and Their Interrelationships
What the distinction between size and level of abstraction shows is that
the micro-macro distinction encompasses at least two hierarchies: a
hierarchy from abstract to concrete and one from large to small. Things
can be simultaneously small and abstract (the modern self) or large andconcrete (the locations of capital cities across the globe in the year
2001). In practice, however, there are correlations between these two hierarchies.
What should also be clear from the discussion is that the distinction
between two levels of analysis (macro and micro) or even three
levels (macro, meso, and micro) is a gross oversimplification. Going
from abstract to concrete or from large to small, many levels may be encountered
in between. So what is commonly called the “macrolevel” in
fact relates to multiple levels of analysis that may range from very large
or abstract phenomena such as modernity, western culture, and industrial
society to significantly smaller or more concrete entities such as the
Internet economy, gender in late nineteenth-century France, and the
Kansai region in Japan. Similarly, microlevel phenomena may range
from larger and more abstract entities such as advertising agencies,
hackers and local area networks, to smaller or more concrete entities
such as Bill Gates, Mary’s filing of a petition, and the software error in
Fred’s computer.
The terminology of micro and macro is therefore too coarse because it
does not distinguish between size and level of abstraction, and it does
not discriminate the different levels and hierarchies that exist within
macro- and microlevel analyses. The consequence of this is that it becomes
difficult to see how various kinds of micro- and macrolevel analyses
may be integrated with each other. Yet, arriving at an adequate
integration of levels of analysis is the major problem faced by theories of
modernity and technology. How do you get from a discussion of late
modernity, rationalization, and the state to a consideration of the development
and use of specific technologies? Conversely, how do you get
from talk about Pentium computers, hacker culture, and virtual communities
to talk about globalization, the modern self, and post-Fordist
economies? Any adequate study of modernity and technology requires
such a bridging of the micro and the macro, of the abstract and general
and the concrete and empirical, of the large and diffuse and the small
and singular.
The major question, then, for theories of modernity and technology,
is how the gaps that exist between different levels of analysis can be
bridged. My distinction between size and level of abstraction indicates
that gaps between levels occur in two ways: because the higher levelrefers to more abstract phenomena than the lower level (e.g., bureaucratic
organizations versus standard-setting bodies in health care), or because
it refers to larger phenomena (e.g., social systems versus markets).
But my discussion also suggests how these gaps may be bridged: by
identifying genus-species relationships (for phenomena at different levels
of abstraction) and part-whole relationships (for phenomena of different
sizes). For instance, an analysis of standard-setting bodies in health care
may be linked to an analysis of bureaucratic organizations by identifying
standard-setting bodies as species or instances of bureaucratic organizations.
Similarly, an analysis of markets may be linked to an analysis
of social systems by identifying markets as subunits within social systems.
Such matches provide the conceptual links that are necessary to
connect discourses that would otherwise remain disconnected.
However, in most studies in the social sciences and humanities that
involve the linking of levels of analysis, the aim of such linking is not
merely to connect disparate discourses. Most studies have a more specific
aim; for instance, explaining events at the micro level, or analyzing
the structure of macrolevel phenomena. Most studies center on a specific
macro- or microlevel phenomenon that the study aims to analyze (e.g.,
late industrial society, or changes in the design of the bicycle in the late
twentieth century). Links created to levels of analyses that are either
higher or lower than that of the analysandum are hence asymmetrical:
the higher- or lower-level entities are invoked to explain or analyze the
analysandum.
Four Types of Interlevel Analysis
When something is analyzed in terms of phenomena at another level,
these phenomena may be from a lower or a higher level, and may differ
in their level of abstraction and their size. This implies that there are
four ways in which analysis may bridge levels. I call these “decomposition”
(the analysis of a larger unit in terms of smaller units), “subsumption”
(the analysis of a smaller unit by reference to larger units),
“deduction” (the analysis of a more concrete unit by analyzing it as a
subclass of a more general phenomenon) and “specification” (concretization;
analyzing a more abstract unit by analyzing one or more of
its more concrete forms). I discuss these in order.• In decompositional analysis (or reductive analysis), a large phenomenon
is analyzed in terms of (much) smaller phenomena. For example,
the behavior of markets (at the macro level) is analyzed as the product
of the behavior of individuals (at the micro level).
• Subsumptive analysis is the opposite of decompositional analysis.
With it, one tries to account for smaller phenomena by (partially) subsuming
them under a larger (structural, functional, or causal) pattern of
which they are a part. For example, given the macroevent of a transition
from Fordism to post-Fordism, in which the bicycle firm Raleigh (a microentity)
is one of the players, there is a modest expectation that
Raleigh will invest in product differentiation, since product differentiation
is part of the transition to post-Fordism (see Rosen 1993).
• In deductive analysis, a phenomenon is identified as a species or token
of a more general phenomenon, and knowledge about this more general
phenomenon is subsequently applied to the more specific phenomenon.
That is, one deduces features from the general to the specific. For example,
a regulatory agency in health care is identified as a bureaucratic organization,
and one’s theory of bureaucratic organizations is subsequently
applied to it.
• In specificatory analysis, finally, a phenomenon is studied by identifying
and studying one or more subtypes or tokens of it. Case analysis,
when used to elaborate a more abstract analysis, is one type of specificatory
analysis, one that makes reference to tokens. An example of specificatory
analysis is Castells’ analysis of East Asian business networks (a
meso unit). Castells analyzes them by distinguishing various kinds (at
meso- and microlevels of analysis) and studying their similarities and
differences (Castells 1996: pp. 174–179).
Implications for Studies of Technology and Modernity
This perspective on levels of analysis can be used to show how modernity
theory can incorporate lower-level analyses in technology studies
and how technology studies can make better use of higher-level analyses
in modernity theory. To begin with the former, macrolevel modernity
theory can benefit from microlevel work in technology studies by using
such work to elaborate its macrolevel descriptions in a way that makesthe overall account more concrete and empirical. Such elaboration can
proceed through a process of decomposition and specification in which
macro units are decomposed into smaller parts and concretized through
the identification of species or subtypes. For example, in an elaboration
of the notion of the bureaucratic organization, decomposition would
specify the components of bureaucratic organizations, and specification
would aim to distinguish various sorts of bureaucratic organizations.
Both types of analysis may be repeated to arrive at levels of analysis that
refer to ever smaller and more concrete phenomena. Such elaboration
makes macrotheories both more easily testable and more capable of informing
microlevel analyses. Such elaboration ultimately makes it easy
to link up with the lower-level analyses of technology studies.
The incorporation of modernity theory into studies of technology can
be similarly clarified. Here, the required types of analysis are subsumption
and deduction. To illustrate, Paul Rosen (1993) has attempted to use
David Harvey’s theory of the shift from a Fordist mode of production to
flexible accumulation in the late 1960s and early 1970s in an explanation
of the constant shifts in the design of mountain bikes. Connecting this
latter fact to Harvey’s theory requires deduction (e.g., identifying it as a
species of product differentiation, a process mentioned in Harvey’s
theory) and subsumption (e.g., identifying accompanying advertisements
as part of the dialectic of fashion and function in post-Fordist economies).
To make an adequate connection, Rosen has to do a good deal of level
building, analyzing the cycle industry and advertising at various levels.
This not only involves bottom-up construction (building up levels from his
microlevel analyses of mountain bike design, firms, and advertisements)
but also top-down construction (elaborating Harvey’s theory). This makes
it possible for him to have the two analyses meet halfway.
I conclude that integrated analyses of technology and modernity,
which build on macrotheories of modernity and microtheories of technology,
are possible, although they require hard work. Analysts have to
work at level building, engaging often in decomposition, subsumption,
deduction, and specification. This, I believe, is the responsibility of both
modernity theorists and scholars in technology studies. It is a joint project
that can begin to abolish the boundaries between two now all-tooseparate
fields.