Tech Affordance Mate Agenc

RESEARCH ARTICLE
WHEN FLEXIBLE ROUTINES MEET FLEXIBLE TECHNOLOGIES:

AFFORDANCE, CONSTRAINT, AND THE IMBRICATION OF
HUMAN AND MATERIAL AGENCIES1
Paul M. Leonardi
Department of Communication Studies and Department of Industrial Engineering and Management Sciences,
Northwestern University, 2240 Campus Drive, Evanston, IL 60208 U.S.A. {Leonardi@northwestern.edu}
Employees in many contemporary organizations work with flexible routines and flexible technologies. When
those employees find that they are unable to achieve their goals in the current environment, how do they decide
whether they should change the composition of their routines or the materiality of the technologies with which
they work? The perspective advanced in this paper suggests that the answer to this question depends on how
human and material agencies—the basic building blocks common to both routines and technologies—are
imbricated. Imbrication of human and material agencies creates infrastructure in the form of routines and
technologies that people use to carry out their work. Routine or technological infrastructure used at any given
moment is the result of previous imbrications of human and material agencies. People draw on this infra-
structure to construct a perception that a technology either constrains their ability to achieve their goals, or
that the technology affords the possibility of achieving new goals. The case of a computer simulation tech-
nology for automotive design used to illustrate this framework suggests that perceptions of constraint lead
people to change their technologies while perceptions of affordance lead people to change their routines. This
imbrication metaphor is used to suggest how a human agency approach to technology can usefully incorporate
notions of material agency into its explanations of organizational change.
Keywords: Affordances, agency, materiality, routines, organizational change, technological change, percep-
tion, imbrication
Introduction1 geographical positioning systems to accurately identify the

correct depth for foundation footings; city planners use 3D
Today, few people complete their work without the use of animation software to model traffic patterns and identify
advanced information technologies. Employees in knowl- bottlenecks before choosing the locations of roadways and
edge-intensive firms are inundated with information tech- pedestrian paths.
nologies, from productivity tools, to simulation software, to
decision systems. The use of information technologies The recognition that advanced software-based information
outside of knowledge-intensive occupations is also growing. technologies (hereafter, just technologies) are permeating
Professionals such as plumbers use digital imaging and every aspect of people’s work has prompted researchers to
recording devices to find clogs in sewer lines; excavators use question the extent to which these technologies determine our
practices and, in so doing, dominate our lives. A burgeoning
line of research adopting a human agency perspective has
1
Carol Saunders was the accepting senior editor for this paper. Mike
provided a fairly optimistic answer to this question. Human
Chiasson served as the associate editor. agency is typically defined as the ability to form and realize
MIS Quarterly Vol. 35 No. 1 pp. 147-167/March 2011 147

Leonardi/When Flexible Routines Meet Flexible Technologies
one’s goals (Emirbayer and Mische 1998; Giddens 1984). For change some other aspect of their environment. Research
example, a person may ask a question because she wants a suggests that organizational routines—sequential patterns of
response, or use a word processing program because she social action (Pentland and Rueter 1994)—are the aspects of
wants to produce a report. Both are empirically observable work most often changed. For example, Poole and DeSanctis
examples of enacted human agency. A human agency per- (1992) showed that when users of a GDSS experienced
spective suggests that people’s work is not determined by the constraints on their ability to reach group consensus, they
technologies they employ. As Orlikowski (2000, p. 412) changed decision-making routines so they could come to
notes, people “have the option, at any moment and within agreement. Zack and McKenney (1995) showed that when
existing conditions and materials, to ‘choose to do otherwise’ e-mail constrained newspaper editors’ ability to follow the
with the technology at hand.” Studies show that even in the norms of their functional structure, they changed their consul-
face of the most apparently constraining technologies, human tation routines so they could preserve occupational roles and
agents can exercise their discretion to shape the effects those responsibilities. Leonardi (2007) showed that when a knowl-
technologies have on their work (Azad and King 2008; edge management technology constrained computer tech-
Boudreau and Robey 2005; Vaast and Walsham 2005). nicians’ ability to learn from coworkers, they changed their
documentation routines. In each case, people had goals that
Within this approach, scholars are beginning to recognize that the technology made possible, but difficult to achieve, so they
people often enact their human agency in response to tech- exercised their human agency to change their routines so they
nology’s material agency (Jones 1998; Pickering 2001; could still achieve their goals in spite of the constraints they
Volkoff et al. 2007). Material agency is defined as the perceived material agency created for them. Human agency
capacity for nonhuman entities to act on their own, apart from is realized by both using the capabilities provided by tech-
human intervention. As nonhuman entities, technologies nology and resisting the limitations those capabilities impose.
exercise agency through their “performativity” (Barad 2003; In none of these cases did the authors document users
Pickering 1995); in other words, through the things they do physically changing the technology directly to achieve their
that users cannot completely or directly control. For example, goal by writing a script, developing a new module, or asking
a compiler translates text from a source computer language developers to modify the functionality of an application.
into a target language without input from its user and a finite Indeed, most studies depict the relationship between tech-
element solver calculates nodal displacements in a mathe- nology and organizing as a process in which human agency
matical model and renders the results of this analysis into a reacts to material agency by producing changes in routines
three-dimensional animation without human intervention. (sometimes even by using the technology’s existing material
agency in unanticipated ways), but leaving the technology’s
Most authors have treated human and material agencies as features intact (e.g., Barley 1990; Chu and Robey 2008;
having a unidirectional relationship. People who have goals Schultze and Orlikowski 2004).
and the capacity to achieve them (human agency) confront a
technology that does specific things that are not completely in The resultant theoretical image of flexible routines and
their control (material agency). In the enactment of their inflexible technologies that current research proposes does not
goals, then, people must contend with the material agency of adequately represent the empirical reality of many contem-
the technology. That is, people must figure out how to porary workplaces. Today, workers have many opportunities
maneuver around it. Studies show that rejecting the tech- to make material changes to the technologies with which they
nology (Constantinides and Barrett 2005; Lapointe and work. A good deal of research within the field of information
Rivard 2005; Markus 2004) or using its features for purposes systems has shown new technologies to be quite flexible.
other than designers or implementers intended (Boudreau and From data repositories (Alavi and Leidner 2001) to networked
Robey 2005; DeSanctis and Poole 1994; Schultze and Boland collaboration tools (Majchrzak et al. 2000) to computer
2000), are two common strategies by which such maneu- simulation software (Dodgson et al. 2007), technologies are
vering occurs. This unidirectional characterization of the increasingly designed to be customizable and adaptable to the
relationship between them casts people as dynamic (they can needs of developers and users. Although not all people have
form new goals and imagine new methods to achieve them) the skills to change the features of a technology themselves,
and technologies as static (the technology has a fixed set of research shows an increasing number of individuals employed
material parameters that do not change across contexts or within organizations as in-house developers, IT staff, or on-
groups of users). site consultants that do have them (Pollock et al. 2007;
Tabrizi 2005). Thus, a flexible technology is not necessarily
To achieve their goals in the face of constraints imposed by flexible because of any inherent properties of the artifact.
an obdurate piece of hardware or software, people must Rather, it is flexible because it is embedded in a context
148 MIS Quarterly Vol. 35 No. 1/March 2011

where people can have it modified to fit their needs in rela- that humans are relatively free to enact technologies in
tively short order. In many modern organizations it may be as multiple ways… .Technology is implicated in social change
easy for people to change the material makeup of a tech- at the discretion of human agents.”
nology, and hence its material agency, as it is for them to
change existing routines. Agency is, of course, given a central role in structuration
theory. Giddens (1984) defines agency as the “capacity for
In this paper, the relationship between human and material action.” At first glance it may seem that such a definition
agencies in contexts where people work with flexible routines extends agency to humans and technologies alike. But
and flexible technologies is explored. The specific argument Giddens makes an important qualification. He suggests that
is that studying contexts in which people can choose whether all action involves motivation, rationalization, and reflexive
they will change routines or technologies puts into relief that monitoring (p. 5). These cognitive processes are linked to
human and material agencies are the shared building blocks human intention.2 People have goals that motivate them.
of routines and technologies. This suggests that, although They can rationalize their goals as acceptable given a set of
they interact directly with one another, human and material circumstances and they can continuously monitor their
agencies are distinct phenomena. By themselves, neither environment to determine whether or not the goal is being
human nor material agencies are empirically important. But achieved. For this reason, Giddens notes that “agency con-
when they become imbricated—interlocked in particular cerns events of which an individual is the perpetrator” (p. 9).
sequences—they together produce, sustain, or change either Given Giddens’ explicit bestowal of agency upon humans,
routines or technologies. To explain what changes when Rose et al. (2005) observe that the idea that technologies
depends on an understanding of the sequence of imbrications could be seen to have material agency is problematic within
between human and material agencies. In what follows, this structuration theory because it “sees agency as a uniquely
perspective is developed theoretically and illustrated empi- human property” (p. 133) and “technology as having no
rically with data from a longitudinal study of the organizing agency of its own” (p. 137). Orlikowski (2005, p. 184) con-
process around a new simulation technology in a large curs that “structurational treatments…privilege human agency
automotive engineering firm. and (inappropriately) discount technological agency” and has
suggested that structuration theory may be unable to fully
account for the fluid and flexible interchange between the
material agency of technologies and the human agency of
Agencies, Affordances, those who produce and use them because “structurational
perspectives reflect the humanist tradition of making the
and Imbrications human subject the center of the action.”3
Differences between Human and Although Giddens limits agency to being a property held by
Material Agencies humans, his theory of structuration does respect the role that
materiality plays as a prop for human action. He argues (in
Over the past two decades, proponents of a human agency Giddens and Pierson 1998, p. 821), for example, that
approach have adopted several theoretical perspectives from
the social sciences that foreground the role that humans play [people] do what they do in lots of different con-
in shaping their own outcomes. Among these perspectives, texts, including physical contexts, which are highly
structuration theory has been most often employed (see Jones strongly relevant to the possibilities and constraints
and Karsten 2008; Poole and DeSanctis 2004) to show that facing any individual or group....We live in a physi-
people draw on norms and communication processes to shape
their interaction with a technology. Because people approach 2
Giddens suggests that not all actions achieve one’s goals for them. Many
technologies from varying vantage points, they often enact actions produce outcomes that were unanticipated, meaning that they are
distinct technologies-in-practice (i.e., ways of using the different than the outcomes people envisioned when forming their goals. As
technology, Orlikowski 2000). IS-specific variants of struc- we shall see, this conception is compatible with the theory of imbrication
turation theory foreground people’s interpretations, which are developed in this paper. As argued below, imbrications of human and
guided by their goals, in shaping their interactions with a material agencies produce technologies or routines, which (intended or not)
may often guide future action.
technology (DeSanctis and Poole 1994; Orlikowski 1992;
Walsham 2002). Building on the groundwork laid by struc- 3
For a more extended review of authors who claim that structuration theory
turational treatments of technology, Boudreau and Robey overlooks or denies the material agency of technologies, see Markus and
(2005, pp. 3-4) note that “a human agency position suggests Silver (2008, pp. 615-616).
MIS Quarterly Vol. 35 No. 1/March 2011 149

cal world that has causal effects in the sense that you a way that thoughtfully builds at the intersection of structura-
just can't walk straight through a wall. tion and actor–network approaches. They argue that although
human and material agencies both influence people’s actions,
Giddens’ view of agency is developed and deployed in his their influence is disproportionate because human agency
specific discussions about social structure, which he argues always has a “head status” while material agency has a
“doesn’t have the same kind of existence as a physical struc- “complement status”:
ture, nor do its causal effects” (Giddens and Pierson 1998, p.
821). That technologies are not within Giddens’ sphere of It is not that Agent1 (some human) had purposes and
interest, and hence not treated directly in a theory of struc- Agent2 (tool) does not (they both incorporate pur-
turation, does not mean that they do not exert some form of poses, and both have a history that is grounded in
influence on the social. Their influence is of a different order previous mediations), but that we generally attribute
to that of the social agency/structure relationship with which head status to the human agent and complement
Giddens is concerned in his work. The difficulties IS scholars status to the tool. In this way, the human subjec-
have with structuration theory may well arise because tivity is given recognition even as the subject’s
Giddens does not attempt to address their concerns, not agency (the capability of acting) is constituted objec-
because his theory is necessarily inimical to them. tively, as an actor in mediated communication with
others. In this sense, Giddens is right: Intentionality
Recognizing that a structurational approach provides a useful is a product of reflexive interpretation. However, he
framework for exploring how people actively structure their fails to see the implications of his own principle.
environments, but that it lacks a specific capacity for theo- That to which we accord intentionality becomes,
rizing the role of technological artifacts, some proponents of ipso facto, an agent—individual or not.
the human agency approach have looked to augment the
structurational approach with the concept of material agency, By treating the relationship between human and material
which they borrow from actor–network theory. The con- agencies in this way, Taylor et al. are able to successfully
ceptualization of nonhuman entities, such as technologies, incorporate into the human agency approach the recognition
doing things that cannot be reduced to human intentionality is that technologies have a material agency that transcends
a core tenet of actor–network theory. Yet owing to their roots changes in context while still giving primacy to the people
in structurational approaches, proponents of a human agency who design and use them.4 The ability to do this rests on the
approach have been uncertain about how to respond to the use of a metaphor of imbrication.
provocative claim in actor–network theory that material
agency is equivalent, in semiotic terms, to the agency of
humans (Callon 1991; Latour 1992). Because of this semiotic Imbrication of Human and Material Agencies
equivalence, actor–network theorists argue that neither human
nor material agency should be given priority in the explana- Authors such as Taylor (2001), Ciborra (2006), and Sassen
tion of people’s construction of outcomes. Rather, each con- (2006) have recently begun to characterize the interweaving
tributes equally to shaping the other, and people and tech- of human and material agencies as a process of imbrication.
nologies are recursively implicated as they use each other to To imbricate means to arrange distinct elements in over-
build associations (Latour 2005). On the one hand, propo- lapping patterns so that they function interdependently. The
nents of the human agency perspective very much resonate verb imbricate is derived from names of roof tiles used in
with the assertion in actor-network theory that technologies ancient Roman and Greek architecture. The tegula and
have material properties that confront humans as external imbrex were interlocking tiles used to waterproof a roof. The
objects. Authors discuss how human agents need to grapple tegula was a plain flat tile laid on the roof and the imbrex was
with the material agency of technology as they work to a semi-cylindrical tile laid over the joints between the tegulae.
achieve their goals (Boudreau and Robey 2005; Walsham
2005). On the other hand, proponents of the human agency
approach are uncomfortable with actor–network theorists’ 4
This observation is similar to Collins and Kusch’s (1998) claim that
belief in the equivalence and interchangeability of human and researchers (and lay observers) often attribute agency to machines when the
material agency (Chae and Poole 2005; Leonardi and Barley “vantage point is low in the action tree” (p. 124). In other words, the analyst
2008). looks only at the technology without recognizing that it is embedded in a web
of human agency. Collins and Kusch don’t discount that technologies can act
on their own, but they caution researchers to keep in mind, like Taylor and
Taylor and colleagues (2001, p. 71) provide a useful language his colleagues do, that it is always humans that are configuring material
for reconciling the notions of human and material agency in agency and deciding how it will become interwoven with their goals.

The interlocking pattern of tegulae and imbrices divided the If you mention an agency you have to provide the
roof into an equal number of channels. The imagery of tiling account of its action, and to do so you need to make
suggests that different types of tiles are arranged in an more or less explicit…its observable traces.…If
interlocking sequence that produces a visible pattern. A roof agency is one thing, its figuration is another. What
could not be composed solely of tegulae nor of imbrices—the is doing the acting is always provided in the account
differences between the tiles in terms of shape, weight, and with some flesh and features that make them have
position prove essential for providing the conditions for some form or shape.
interdependence that form a solid structure. Human and
material agencies, though both capabilities for action, differ Thus, sometimes, human and material agencies interweave in
phenomenologically with respect to intention. Thus, like the ways that create or change routines; other times, they weave
tegula and the imbrex, they have distinct contours yet they together in ways that produce or alter technologies.
form an integrated structure through their imbrication.5
Taylor and his colleagues (2007, p. 399) suggest that this The metaphor of imbrication is in several ways useful for
integrated structure is an organizational structure: “Applied explaining the interweaving of human and material agencies.
to organizational analysis we consider [imbrication] to be the First, imbrication suggests that human and material agencies
way that interagency relationships are interweaved to form… are effectual at producing outcomes (e.g., routines or tech-
infrastructure.” nologies) only when they are joined together, but that their
interdependence does not belie their distinct characters. As
The argument in this article is that the routines and tech- Sassen (2006, p. 345) suggests,
nologies are the infrastructure that the imbrication of human
and material agencies produce. Put another way, if we were I use the term imbrication to capture the simul-
to examine routines and technologies under a microscope, we taneous interdependence and specificity of each the
would find that each is made up of the same basic building digital and the nondigital. They work on each other
blocks: human and material agencies.6 Although we may but they do not produce hybridity. Each maintains
make the ontological claim that routines and technologies are its distinct irreducible character.
indistinguishable phenomena because they are both consti-
tuted by human and material agencies, we must be mindful Thus, the notion of imbrication allows for maintaining the
that the ways in which those agencies are weaved together distinction between human and material agencies with respect
produce empirically distinct figurations.7 Latour (2005, p. 53) to intentionality while still recognizing their synergistic
defines figuration as the process by which agencies take on interaction. The metaphor of imbrication is distinct from
observable properties: Latour’s (1993, 1999) notion of the hybridicity between the
human and the material. Latour argues that human and
material agencies are indistinguishable (they are hybrids) such
5
Clearly, using the imagery of imbrices and tegulae has its problems. Both that action has no point of origin. In other words, either
tiles are made of “material” in the sense that they are physical creations of people or technologies can begin changes in sequences of
clay. The struggle to find a suitable image with which to describe the imbri- action.8 By keeping the distinction between human and
cation of human and material agencies points to the conceptual difficulty of material agencies, the imbrication metaphor asserts a slightly
integrating these phenomena. Thus, the analogy is meant to be illustrative
different relationship: people have agency and technologies
rather than to be read literally.
have agency, but ultimately, people decide how they will
6
This view coincides with the recent tendency to call into question the respond to a technology. As Cooren (2004, p. 377) suggests,
distinction between routines and technologies. This tendency is based on “To say that nonhmans do things does not mean that human
ontological claims such as (1) routines are patterns of social action that are contributions are passed over…humans can appropriate what
often mediated by technology (Pentland and Feldman 2007) and (2) a nonhumans do.”
technology’s material form is shaped to a significant degree by the routines
of product development (Dougherty 1992; Garud and Rappa 1994), and a
technology can only have effects on the way people work if it becomes a
Second, because the metaphor of imbrication sensitizes us to
technology-in-practice; that is, if it is incorporated in people’s existing work the production of durable patterns, it reminds us that all inter-
routines (Orlikowski 2000; Vaast and Walsham 2005). actions between human and material agencies produce an
organizational residue. When human and material agencies
7
This is akin to saying (as chemists do) that our theories of physical
chemistry should treat diamond and graphite as equivalent because they are
both made of carbon, but empirically we must distinguish between them
8
because they have different crystal structures by virtue of the ways in which For a more detailed discussion of this point, see Latour’s (1994, pp. 30-33)
the carbon atoms have interlocked. example of the relationship between a person and a gun.

imbricate to produce routines or technologies, those figura- (on which it acts) not just as the result of the other,
tions have staying power. Routines persist absent their but also the ensuing execution of that result.
creators (Cyert and March 1963), as do technologies (Mackay
and Gillespie 1992). As the people within the organization Imbrications at one point in time create the possibility for
continue to use routines and technologies in practice they (and set certain restrictions on) future imbrications, although
become, in Star and Ruhleder’s (1996) terms, infrastructure. in a nondeterministic way. By recognizing that accumulated
That is, they provide the context and the means for organizing nature of past human–material imbrications, an imbrication
to happen, but they are taken-for-granted as natural relations. perspective must provide a language to explain how activities
The imbrications that produce routines and technologies in the past condition (as opposed to cause) future human–
become “black-boxed” such that we no longer actively ques- material sequencing.
tion why we have certain routines or technologies or what
they are good for. As organizational infrastructure, Taylor et
al. (2001) suggest that imbrications are only ever, to use
Construction of Affordances and Constraints
Heidegger’s (1959) distinction, “ready-to-hand” (zuhanden)
as Catalysts for Imbrication
as opposed to “present-at-hand” (vorhanden). That is, the
capabilities that human and material agencies create as they
interlock with one another become proceduralized and for- The preceding discussion uses the metaphor of imbrication as
gotten over time. As long as human and material agencies are a way of recognizing that human and material agencies are
imbricated in ways that allow people to get their work done, distinct phenomena but that they are fundamentally inter-
the structures they create are transparent and always ready-to- dependent; that past imbrications accumulate to help explain,
hand as opposed to actively and reflexively drawn upon in although certainly not predict, how human and material
every day action (present-at-hand). Thus the products of prior agencies will become conjoined in the future; and that people
imbrications (e.g., a routine or a technology) lay the ground- actively work, within the framework established by previous
work for continued organizing in that they provide the rou- imbrications, to reconcile their goals (human agency) with the
tines and technologies that people can use to structure their things that a technology can or cannot do (material agency).
actions. Because the interweaving of human and material
agencies produces routines and technologies that are regularly As outlined above, human and material agencies are the basic
used by organizational members, we can say that past human– common building blocks of routines and technologies. To the
material imbrications influence how human and material extent that routines and technologies are both flexible; people
agencies will be imbricated in the here-and-now. can change or have them changed as they are developing or
using them. Changes in routines or technologies, then, re-
With this recognition, the imbrication metaphor provides a quire new imbrications of human and material agencies. Yet
third benefit to theory in that it enables a way to appreciate if a person has the option of changing a routine or changing
accumulation over time without resorting to deterministic a technology, how does she decide which one to change? To
language. As several scholars have argued, in order to bridge answer this question, we must consider the differential ways
the gap between the extreme poles of determinism and volun- in which human and material agencies can become imbri-
tarism, researchers must better explain how the accumulation cated. To do so, we turn to a theory of affordances, which
of past changes bears on present changes (Leonardi and provides a vocabulary useful for theorizing the imbrication of
Barley 2008; Schultze and Boland 2000). Imbrication implies human and material agencies.
accumulation in the sense that the overlap of human and
material agencies is not replicated in the same way over time In an effort to explain how animals perceive their environ-
and does not necessarily have inertial tendencies, but that the ments, James Gibson (1986) a perceptual psychologist, sug-
way imbrication occurs at Time 1 will influence the way it gested that surfaces and objects offered certain affordances
occurs at Time 2. As Ciborra (2006, p. 1345) explains, for action:
Imbrication is…more subtle than a mere overlapping If a terrestrial surface is nearly horizontal…nearly
or mutual reinforcement….It is more “active” than flat…sufficiently extended…and if its substance is
that. Its sense possibly can be best captured by the rigid…then the surface affords support….It is stand-
technical meaning of the term imbrication in the on-able, permitting an upright posture for quadru-
(French version) of the Unix operating system: peds and bipeds….Note that the four properties
imbrication is the relationship between two lines of listed—horizontal, flat, extended, and rigid—would
code, or instructions, where one has as its argument be physical properties of a surface if they were mea-

sured with scales and standard units used in physics. For Norman, affordances are “designed-in” properties of arti-
As an affordance of support for a species of animal, facts. The goal of an affordance is to signal to the user what
however, they have to be measured relative to the the technology can do and how it is to do that thing. To do
animal. They are unique for that animal. They are this, designers must make affordances easy to perceive: “The
not just abstract physical properties (p. 127). designer cares more about what actions the user perceives to
be possible than what is true” (Norman 1999, p. 39). Users
In Gibson’s formulation, people do not interact with an object are important to Norman inasmuch as they can identify a
prior to or without perceiving what the object is good for. As technology’s affordances; however, they play little role in
he suggests, the physical (or material) properties of an artifact creating affordances. Instead, affordances are created strate-
exist apart from the people who use them, but they are infused gically (if she is good at her job) by the designer. In this
with meaning “relative to the posture and behavior of the formulation, Norman’s argument is different than Gibson’s in
animal being considered” (pp. 127-128). The concept of that he claims affordances do not change across different
affordance is useful in explaining why human and material contexts of use; rather, they are always there waiting to be
agencies become imbricated: Technologies have material perceived.
properties, but those material properties afford different
possibilities for action based on the contexts in which they are Hutchby (2001) seeks a middle ground between these prior
used. Although the material properties of a technology are conceptualizations by emphasizing the relational character of
common to each person who encounters them, the affordances affordances. In his view, affordances are not exclusively
of that artifact are not. Affordances are unique to the parti- properties of people or of artifacts; they are constituted in
cular ways in which an actor perceives materiality. To this relationships between people and the materiality of the things
end, Gibson offers a perceptual explanation of the relationship with which they come in contact. In this formulation,
between materiality and affordances: materiality exists independent of people, but affordances and
constraints do not. Because people come to materiality with
diverse goals, they perceive a technology as affording distinct
The psychologists assume that objects are composed
possibilities for action. For Hutchby, the affordances of an
of their qualities…color, texture, composition, size
artifact can change across different contexts even though its
shape and features of shape, mass, elasticity,
materiality does not. Similarly, people may perceive that a
rigidity, and mobility….But I now suggest that what
technology offers no affordances for action, perceiving
we perceive when we look at objects are their
instead that it constraints their ability to carry out their goals.
affordances, not their qualities. We can discriminate
the dimensions of difference if required to do so in
Markus and Silver (2008, p. 620) suggest that
an experiment, but what the object affords us is what
we normally pay attention to (p. 134).
in terms of the relational concept of affordances…
properties of objects are seen as necessary but not
Because materiality can provide multiple affordances, it is sufficient conditions [for changes in action]….
possible that one artifact can produce multiple outcomes. Because action is goal-oriented, it is neither required
nor appropriate to describe objects and affordances
Gibson’s work has been most notably applied to discussions in a reductionist fashion.
of technology by Norman (1990, 1999), who argues that good
designers purposefully build affordances into a technology to To emphasize that affordances arise when a person interprets
suggest how its features should be used. Norman (1990) a technology through his or her goals for action, Markus and
seems to suggest that affordances are intrinsic properties of Silver define affordances as “the possibilities for goal-
artifacts and that the role of design is to make affordances oriented action afforded to specific user groups by technical
easily perceptible to would-be users: objects” (p. 622). Because affordances are relational—
existing between people and an artifact’s materiality—
Affordances provide strong clues for the use of their artifacts can be used in myriad ways and have multiple effects
materials. Plates are for pushing. Knobs are for on the organization of work (Fayard and Weeks 2007;
turning. Slots are for inserting things into. Balls are Zamutto et al. 2007).
for throwing or bouncing. When affordances are
taken advantage of, the user knows what to do just According to the relational view, we might argue that afford-
by looking: no picture, label, or instruction is ances and constraints are constructed in the space between
required (p. 9). human and material agencies. People’s goals are formulated,

M1 H1 M2 H2
Note: Circles represent figuration of agencies as routines, square represents figuration of agencies as technology. Routines and
technologies are constituted by different imbrications of the same basic agencies.
Figure 1. Imbrication of Human and Material Agencies Produces Routines and Technologies
to an important degree, by their perceptions of what a routine able to turn their boring text-document into a visually
or a technology can or cannot do, just as those perceptions are appealing, well-formatted newsletter. They attempt to use
shaped by people’s goals. For this reason, the argument is their current word processing technology to do so, but they
that as people attempt to reconcile their own goals with the discover no easy capabilities that will allow them to format
materiality of a technology, they actively construct perceptual the text in unique ways and to draw diagrams. In the space
affordances and constraints. Depending on whether they between the technology’s existing material agency (its ability
perceive that a technology affords or constrains their goals, to render text, but not manipulate its placement or draw
they make choices about how they will imbricate human and diagrams around it) and their goal to produce a visually
material agencies. Acting on the perceived affordances of a appealing newsletter, the coordinators construct a perception
technology can then lead users to realize new intentions that that the technology they currently use constrains their human
could be achieved through these material features. The dif- agency.9 What do they do? They could decide not to create
ferent ways in which human and material agencies are a newsletter. They could also change their work routines,
imbricated results in distinct outputs—either a new routine, or perhaps altering patterns of information dissemination so that
a new technology. they could communicate the information verbally and, thus,
not need a newsletter. But to the extent that the word pro-
To illustrate this point, consider the following example, which cessing program is flexible (e.g., it is designed to be easily
is extremely basic for descriptive purposes. Coordinators in redesigned, the coordinators have the skills to change the code
a not-for-profit community organization have recently begun that underlies its functionality, or they have access to people
to disseminate reports to community members that summarize who can make changes to the code for them) they may decide
the organization’s service activities. This news dissemination to change the materiality of the technology to meet their
routine is heavily reliant on the use of a word processing goals. Thus, the construction of the technology’s constraint
technology that coordinators can use to create the summary. arose between the material agency of the existing word
The word processing technology is configured to act in parti- processing program (M1) and the consultants’ collective goal
cular ways (material agency) and the capabilities that it (H1) to create a newsletter. To overcome this constraint, the
provides are partially what led the coordinators to envision coordinators changed the technology by giving it features that
creating the newsletter in the first place. The news dissemi- allow people to make digitized drawings. In so doing, they
nation routine is enabled by the functionality of the word gave the technology new material agency (M2); the program
processing technology (material agency) that allows coor- now contains a feature that allows people to make digitized
dinators to create a summary document and the goal of drawings. A human agency approach would treat the goal
disseminating information to community members (human
agency), which was partially formulated by the acquisition of
the technology. In Figure 1, this routine is depicted as a circle 9
Note that constraint, as it is used here, is not the property of the technology.
made up of material agency (M1) and human agency (H1). Rather, a technology that was once everything users wanted or needed is now
perceived by them as a constraint to achieving their new goal. This percep-
As the coordinators use the word processing technology and tion of constraint arises because (1) their goal has shifted and (2) they can’t
figure out how to achieve their goal with the features of the existing
reflect on its output, they decide that they would like to be technology.

that created the technology’s new features (H1) and the changes in technologies because they contain the same
material agency that the technology now has by virtue of building blocks, some of which are shared in common. Of
those new features (M2) as constitutive features of the course, where one begins reading the chain of imbrications
technology (Boudreau and Robey 2005; Orlikowski 2000). (from a material or a human agency starting point) is some-
Thus, the technology is represented by a square box around what arbitrary, and the chain of imbrications certainly
H1 and M2. stretches out in either direction. In other words, it is arbitrary
to look for beginning or end points in an imbricated system.
What we begin to see when examining Figure 1 is that the Instead, the analyst should be more interested in explaining
imbrication of an existing material agency with a new human how imbrication occurs and the effects that prior sequences of
agency (material → human) constitutes a routine. This imbri- imbrication have on future actions.
cation produces the perception of constraint. To overcome
that constraint, the consultants change the functionality of the The examples provided above, although entirely hypothetical,
technology thereby giving it a new material agency. begin to illustrate what imbrication looks like, how it might
Consequently, the imbrication of an existing human agency occur, and why different types of imbrications of human and
with a new material agency (human → material) brings material agencies may produce reciprocal changes in routines
changes to a technology at some level. The coordinators may or technologies. To move this emerging framework of imbri-
begin to use the newly changed features of the word cation from the hypothetical to the empirical, we turn to data
processing application (M2) to produce digitized drawings collected through an ethnographic investigation into the work
and, in so doing, begin to construct a perception that these of crashworthiness engineers at a major automobile firm.
capabilities could afford them the possibility of expanding the
readership of their newsletter because of the entertainment
value provided by catchy graphics. Consequently, they begin
to form a new goal that the readership of their newsletter An Empirical Illustration: Development
should expand (H2). To achieve this goal, however, the group and Implementaton of “CrashLab”
must reorganize itself by changing routines. Some members Simulation Technology
must specialize in drawing, others in writing copy, and others at Autoworks
in laying-out the newsletter to include both text and image.
The imbrication of this existing material agency with a new This section illustrates how the imbrication metaphor can be
human agency (material → human) results in changes in the useful for interpreting the findings of an ethnographic
news dissemination routine (e.g., specialization occurs and research study, which investigated activities occurring around
people increase or decrease their consultations with one a technology built to automate computer simulations for
another based on their newfound specialties). crashworthiness engineering work at Autoworks (a pseudo-
nym), a major automaker located in the United States.
As Figure 1 illustrates, the perception of constraints produces Crashworthiness (a vehicle’s ability to absorb energy through
a sequence of imbrication that changes technologies while the structural deformation) is assessed by performance engineers
perception of affordances produces a sequence of imbrication (hereafter just engineers) who conduct physical crash tests
that changes routines. Further, a new agency (human or and computer simulations. Because the cost of administering
material) does not just imbricate with an existing agency; and recording the data for physical crash tests (building a
rather, it is interwoven with an entire history of imbrications prototype vehicle, loading it with instrumentation, crashing it
that came before it. Here one might consider that the tiler of into an object, and interpreting the findings) takes so long,
a roof does not just imbricate a new imbrex with one existing Autoworks encourages its engineers to use finite element
tegula; she imbricates an imbrex with the entire history of simulation models to predict a vehicle’s crashworthiness and
imbrex–tegula relations that came before it, and that history provide recommendations for how a vehicle structure can be
influences, to an important degree, where and how she can changed to increase performance in an impact.10 The vast
place the newest tile. Thus although people can make choices majority of an engineer’s time is spent in iterative cycles of
about how they imbricate human and material agencies, the building finite element simulation models, analyzing their
accumulation of choices that were made prior to their decision
affect the types of imbrications they can make insomuch as
10
they shape their perceptions of affordances and constraints. Finite element analysis is a computational technique that divides the actual
When we look at an imbricated system of human and material geometry of a part into a bounded (hence finite) collection of discrete
elements. The elements are joined together by shared nodes. Nodes are the
agencies like the one idealized in Figure 1, we can begin to locations where values of unknowns (usually displacements) are approxi-
see how fundamentally related changes in routines are to mated. This collection of nodes and elements is commonly called a mesh.

performance, and making suggestions for changes in design. I also collected data on the work of crashworthiness engineers
To reduce the time and effort it took engineers to set up and before CrashLab was implemented, the activities of devel-
analyze simulation models, engineers in Autoworks’ R&D opers, trainers, and managers during implementation, and the
Organization developed a new software tool called CrashLab. work of engineers after CrashLab was implemented. During
CrashLab is a tool that is used for pre-processing (setting up each of these activities I utilized four primary data sources:
mesh models in ways that can be analyzed by a solver to observations made of informants at work, artifacts that infor-
produce desired results11) the finite element models that safety mants produced or used in their work (e.g., screen shots of
and crashworthiness engineers use to make predictions about computer simulations, test request forms, best-practice
how a vehicle will respond in a crash and post-processing guidelines), interviews conducted with informants about their
(how the results obtained from the computational analysis can work, and logs kept by informants tracking their use of
be extracted in such a way so as to have predictive power) CrashLab. These data collection methods resulted in 134
those results. The idea for CrashLab emerged out of observations (each of which was three hours or more in
Autoworks’ R&D Organization in 1995. After nearly 10 duration) with crashworthiness engineers, 51 interviews with
years of developmental work, CrashLab was deployed into engineers who used CrashLab, 17 additional interviews of
the user community in September 2005. people such as managers and implementers who were also in
some way involved with CrashLab, and more than 500 arti-
Data on CrashLab and the work system in which it was facts used by engineers. A summary of all procedures used to
embedded were obtained through the use of ethnographic collect the data for this project appears in Table 1. The
techniques. I spent nine full months over a two-year period observational data presented herein are drawn from the work
(2004-2006) conducting interviews and observations with of engineers on a vehicle program I called the Strut Group.
people who were involved in all aspects of CrashLab’s
lifecycle. I began by identifying key informants whom I These varied data sources are used to illustrate how the
knew were involved in CrashLab’s development, and through human and material agencies prevalent in the work of devel-
these interviews acquired the names of others who were also opers and users of CrashLab became imbricated and, in so
involved. Fortunately, Autoworks maintained a very detailed doing, produced subsequent changes in its features and in the
and accurate database of employee contact information and I routines conducted by engineers at Autoworks. I present
was able to use this database to track down informants even these data not to test the framework of imbrication, but to
if they had moved into new positions within the company. I illustrate its value in thinking about how the relationships
conducted interviews with members of five different organi- between human and material agencies are interwoven and
zations within Autoworks: Safety, Research and Develop- how this imbrication leads people to change their routines and
ment (R&D), Information Services (Infoserv), Technology technologies. To make this illustration, I intentionally gloss
Production (Techpro), and Best Practice Evaluation (Bestpra). over many of the details relevant to informants’ decision-
Additionally, I conducted interviews with informants in three making processes surrounding the development and use of
technology supplier organizations who helped, at various CrashLab. I also push the political dimensions of these deci-
points in time, to build CrashLab. I also conducted several sions to the background in order to foreground the processual
interviews with other individuals who were involved with and path-dependent nature of the imbrication metaphor. For
CrashLab but who were not members of any of these organi- further details about the development or initial implementa-
zations, such as Autoworks senior management, consultants tion of CrashLab please consult Leonardi (2010).
from other supplier firms, and professors at nearby univer-
sities. In total, I conducted 58 interviews with people With these limitations in mind, Figure 2 guides the presenta-
involved in changing CrashLab’s features.12 tion of data in this paper. Of course, this figure, and the data
used to elaborate it, illustrates idealized cycles of imbrication
11
with the goal of establishing a perspective on imbrication, and
A solver is program run on a supercomputer that applies equilibrium
it should be read as such. I discuss five imbrications through
equations to each element in a finite element model and constructs a system
of simultaneous equations that is solved for unknown values to produce
which the human and material agencies of Autoworks’ work
measures of a vehicle’s performance given the parameters specified in the system became interwoven. Three human → material imbri-
set-up of a simulation. cations (enclosed by perforated boxes in Figure 2) produced
12
Interviews about CrashLab’s development rely on retrospective accounts
given by informants. Thus, I am limited in my ability to discuss the actual
decisions that informants confronted in the practice of their work. I must, of the study, but one that should not diminish the overall value of the case in
therefore, rely on their accounts of their decisions and interpretations of illustrating the conceptual framework made available by the imbrication
events, even though they are constructed after the fact. This is a limitation metaphor.

Table 1. Summary of Data Collection

Interviews with Developers (By Organization)
R&D 12
Infoserv 6
Techpro 11
Bestpra 14
Safety 5
Other 10
Work of Engineers Before CrashLab
Observations 10
Artifacts 78
Interviews 49
Implementation Activities
Observation 14
Artifacts 4
Interviews 9
Work of Engineers After Crash Lab
Observations 124
Artifacts 571
Interviews 42
General Interviews
Managers 11
Designers 5
Vice President 1
changes in CrashLab’s features as people around the company. In 1994, the newly formed Engineering Mechanics
technology constructed the perception that it constrained their Department (EMD) within R&D was charged with this new
goals. The two human → material imbrications (enclosed by goal. To more fully understand how crashworthiness engi-
solid boxes in Figure 2) produced changes in the way neering work was being conducted and, by association, to
engineers at Autoworks interacted with one another as they develop ideas about what sorts of tools could possibly
began to construct the perception that CrashLab could afford improve it, developers in the EMD spent the first two months
them the possibility of changing the way they worked. As of 1995 working with crashworthiness engineers.
mentioned above, these starting and stopping points are
arbitrary. I intend to show why one type of imbrication led to EMD developers noticed that the steps engineers took to pre-
another and to use this understanding to explain how it is that process their models (preparing to submit them to the solver)
the human and material agencies at Autoworks were inter- were highly idiosyncratic. Unlike most of the engineers in
woven and the consequent effects that these imbrications had Safety, these developers worked in an R&D department that
on the process of organizing. focused on scientific research. The timelines for most pro-
jects in R&D were much longer than for projects in Safety
because R&D was not tied to the stage-gates outlined in
Imbrication 1 (Human → Material) Autoworks’ formal vehicle development process. Similarly,
because engineers in R&D published their work in journals
For many years, Autoworks’ R&D had placed a dominant and presented their findings at academic conferences, they
focus on the development of new technologies that could be were bound to the traditions of academic research, including
used in production vehicles (e.g., airbag sensors, anti-lock the need to make their methods transparent and their findings
brakes, etc.), but devoted few resources to developing tech- reproducible. Overall, EMD developers came to Safety with
nologies that would improve engineering processes within the an ideological orientation toward crashworthiness work as a

Constraint:
Engineers form Simulation Current finite
EMD group work should element tools do
be credible not help produce
consistent results
Affordance:
CrashLab built Aggregates CrashLab provides
to automate data in standard capability to
work reports automate standards
Imbrication 1 Constraint:
(Human → Material) Standard
Engineers form CrashLab does not
work should make engineering
Bestpra group increase speed
work comparable
Affordance:
Imbrication 2 Automatically CrashLab allows
(Material → Human) Auto-placement engineers to
generates input
features added coordinate analyses
deck
of models
Key
Imbrication 3
(Human → Material) Analyses Constraint:
= Change in Technology Users talk with should be tested CrashLab does not
others about systematically allow comparison
analysis of multiple
= Change in Routine iterations
Imbrication 4
= Human Agency (Material → Human) Iteration Generates
comparison cross-plots and
features added graphs
= Material Agency
Note: Perforated boxes represent process of Imbrication 5

Human → Material imbrication and solid (Human → Material)
boxes represent process of Material → Engineers form
Human imbrication RSA group
Figure 2. Imbrications of Human and Material Agencies Produce Changes in Routines and
Technologies
scientific activity. By contrast, engineers felt that it was far post-processing tools generated analysis results, engineers
more important to make sure a simulation model correlated only extracted data from them on certain parameters like
with the results of a physical test than to spend time making intrusion or energy dissipation. The results were not pre-
sure the methods they used to correlate their models could be sented in any systematic way, nor could they be stored or
easily discerned by others. Thus, the institutional background printed in a standard form. This constraint enforced by
of R&D along with the creation of the EMD, which brought existing finite element tools was problematic for EMD
together engineers who interacted around process improve- developers because, without standard results, the practices an
ment methodologies, shaped the goals of developers to make engineer used to build and analyze a simulation model could
crashworthiness simulation work credible and reliable. not be verified to determine their robustness, nor could they
be validated by an external reviewer. As one EMD developer
With these goals in mind, EMD developers began to notice explained,
that the current suite of finite element tools used by engineers
to build and analyze simulation models constrained their Crashworthiness work needs to be logically done.
ability to produce consistent results. When current pre and Everyone has to do it the same so you know you can

trust the answers….Using methods people can trust Imbrication 2 (Material → Human)
is important. So if you do something different than
what is normal, you should be telling this to your As EMD developers worked to refine CrashLab and add more
people so they can consider it all when they evaluate functionality to the tool, they began to circulate a fully func-
your results. Most tools don’t let this happen. tioning prototype throughout various engineering groups at
Autoworks. One of these groups, the Crashworthiness Focus
Thus, EMD developers perceived constraints with the current Group, became very interested in CrashLab. Each perform-
finite element technologies that had to be overcome in order ance engineering function at Autoworks maintained a focus
for them to realize their goal to make crashworthiness group, which was responsible, among other things, for
simulation work more credible. endorsing and sanctioning the use of all new technologies
used by the engineers in the particular function. The Crash-
EMD developers decided that they could develop a new worthiness Focus Group was no different, and according to its
technology that would work in conjunction with existing pre- formal mission statement aimed to “commonize, improve and
and post-processing applications already used by crash- manage vehicle crashworthiness simulation tools and method-
worthiness engineers to increase the credibility of simulation ologies necessary to synthesize and validate Autoworks
work. EMD developers knew that they did not have the product designs to meet crash safety requirements.” Put more
structural authority within Autoworks to mandate that engi- plainly, the focus group actively worked with members of
neers had to create standard reports, but they did have the R&D to develop new tools and methods that would aid the
means to impose their goals technically. They recognized work of computer simulation. Importantly, the focus group
had final approval of all tools implemented in the user
implicitly that their goal to make crashworthiness simulations
environment by crashworthiness engineers. If the focus group
more credible (human agency) could be fulfilled if they
did not deem a new technology sufficiently useful or appro-
created code that would automatically aggregate simulation
priate for the work of its engineers, it would refuse to endorse
results into standard reports (material agency) by building a
it and ultimately Autoworks’ IT department could not install
new technology. Over the next three years, EMD developers
the software on engineers’ machines.
worked with an external vendor to build a new technology
called CrashLab. The interface for CrashLab was relatively
By the winter of 1998, the focus group had thoroughly tested
straightforward. Along one side of the screen, a flow chart
CrashLab. Whereas the EMD developers wanted to make
indicating the procedures an engineer had to take to prepare
crashworthiness simulations more credible, focus group
a model for analysis would guide him through a set of ordered members had a different goal. They were interested in stan-
steps. As the user moved from one step to the next, a three- dardizing the work of all engineers. Their desire to stan-
dimensional model of the vehicle would appear along the dardize was buttressed by the hope that standardization would
other side of the screen and provide suggestions for how to remove human error from the vehicle design process. By
prepare the model for analysis. To allow engineers discretion interpreting CrashLab’s material features through their own
without sacrificing the credibility of their models, EMD goals, Focus Group members began to agree that CrashLab
developers designed a report format that would record any could afford the automation of engineering standards. The
deviations engineers took from the best practices embedded problem for the focus group, however, was that although the
in CrashLab. In its visual layout, the report contained a table current material features of CrashLab could afford automation
entitled “Best Practice Violation.” Any parameter in the pre- of engineering standards, no standards currently existed to
or post-processing of the model that diverged from the best automate. As the focus group’s chair commented,
practices appeared in this table. The second column to the left
contained a description of the best practice embedded in The problem that I and some others had with Crash
CrashLab and the column immediately to its right documented Lab was that, I mean, at that time we didn’t even
the parameter that was changed. A third column calculated have anything standardized.…If we are ever going
the percentage that the best practice was violated. The pur- to have math [simulations] lead the design and
pose of this third column was to give the reader of the report reduce our reliance on physical testing, we had to
confidence in how close the model used to generate these standardize the way we do the simulation work….
results was to an ideal test case. Larger percentages of devia- And the whole question of quality and accuracy of
tion might then mean that the results would not be repeatable the results is what drove us to know that the engi-
or widely generalizable. What EMD developers were most neers had to start doing the work the same way.
concerned about, however, was not the inherent generali-
zablilty of the findings, but that the reviewer of the report Before CrashLab could automate standard work procedures,
would know exactly how he could use the results to make Autoworks needed to define what those standard work
predictions for a vehicle’s crashworthiness. procedures were.

In early 1999, with the recommendation of the focus group, Imbrication 3 (Human → Material)
Autoworks’ vice president for global engineering approved
the creation of a new engineering organization called the Best Crashworthiness managers, who took it as their new goal to
Practice Evaluation (Bestpra) group. Bestpra’s goal was to speed up the way that engineers worked, began to search for
create and update standard work procedures for crashworthi- clear and feasible ways to make simulation building and
ness analysis. Structurally, Bestpra sat outside of regular analysis faster without compromising the quality of results.
engineering production work at Autoworks. The six engi- For many managers, CrashLab was an ideal IS application
neers who formed Bestpra were given the title of Subject with which to carry out their goal because it was easy to
Matter Experts (SME). Their role was to collect information envision how the technology could provide capabilities for
from engineers in Safety to determine what the standards automation. Given the history of automation in industrial
would be for a particular type of analysis. Based on this settings and in the automotive industry in particular, it is no
information, the SMEs began to create a series of documents surprise that crashworthiness managers looked to a tool like
outlining standards that all engineers should follow when CrashLab to make work faster. The history of manufacturing
setting up and analyzing models for specific crash loadcases.13 teaches us that that an essential precursor to automation is the
A typical standard work document was at least 20 pages of rationalization of tasks. At the turn of the 20th century, effi-
instruction on how to best execute practices, from routine ciency experts such as Taylor (1911/1998) and Gilbreth
mesh generation to more detailed and loadcase specific (1911/1993) pioneered methods to specify and rationalize
procedures, such as how to write the appropriate contact cards tasks, breaking a “job” down into its component parts. The
to specify how the solver should interpret relationships insight that tasks could be rationalized heavily influenced
between parts in the model. In March 1999, Bestpra pub- entrepreneurs like Henry Ford, who reasoned, following
lished its first standard work guidelines for frontal loadcases. Adam Smith, that if tasks could be broken into component
EMD developers used the best practices identified in this parts they could also be separated and performed by different
document to determine how to automate the process of setting workers (to increase efficiency) and then reintegrated later.
up and analyzing a simulation model to test FMVSS 208 Rationalization and fractionalization of work then led to auto-
specifications with CrashLab.14 mation, whereby machines (which could perform component
actions more quickly and reliably than humans) performed
With the SMEs working to create guidelines that would instrumental tasks and human workers served merely to
standardize the work of engineers, managers in Safety began reintegrate the parts at the end of the line, and eventually to
to develop a new set of goals. They were convinced that the the creation of flexible technologies that could do both
automation of standardized work should lead to faster instrumental and assembly tasks without human intervention.
simulation model building and analysis because engineers no In a similar way, CrashLab took the standardized (or rationa-
longer had to take time to discern which practices were most lized) work practices specified by Bestpra and used
appropriate or efficacious for the job at hand. Standard work FORTRAN code to automate their execution with minimal
guidelines would effectively replace this individual judgment user input.
and would lead to an increase in the speed with which engi-
neers could do crashworthiness work. Faster crashworthiness Although CrashLab automated output of simulation results, it
work was preferable because it meant that engineers could did not automate the practices whereby engineers set up
conduct more simulations in a given time period than they simulation models to be submitted to the solver. Even though
could before. By doing so, Safety could move from the use the SMEs rationalized crashworthiness tasks by writing
of simulations to retrospectively validate physical tests to the standard work guidelines, managers believed that standard-
use of simulations to prospectively predict the outcomes of ization could only increase the speed of work when paired
physical crashes. Thus, the material agency of CrashLab was with automation capabilities. As one manager commented,
imbricated with the human agency of Safety management during the late 1990s when simulations of vehicle impacts
through the creation of Bestpra, which began a new standard were just at the point of becoming faster and cheaper to run
work writing routine. than physical tests, speed was the goal exemplar:
Everything here is about speed. Simulations are

13
Loadcases are specific conditions through which loads, in the form of only helpful if they can be done quickly. The
energy from an impact, are applied to the vehicle’s frame. bottom line is that if a simulation takes longer to run
than a hardware test, even though it costs more in
14
Federal Motor Vehicle Safety Standard (FMVSS) 208 is a standard set by the short-term, management will always prefer the
the federal government’s National Highway Traffic Safety Administration
test. That’s because where you really start losing
(NHTSA) indicating minimum tolerances that a vehicle must meet in a frontal
impact to be saleable in the United States. your money is when you don’t get the vehicle to

market soon enough. If you didn’t pay say $300,000 Imbrication 4 (Material → Human)
for another crash test and instead you waited an
extra month for the simulation to be done and so you By the fall of 2005, all engineers who were involved with
delayed product launch for 30 days you would lose regular production related work at Autoworks felt tremendous
say something like $5,000 in profit per vehicle on time pressure. Management constantly complained about the
like 10,000 vehicles. So that’s, what, like number of months in the company’s vehicle development
$50,000,000 you just lost? No one would hesitate to process. Every year, senior vehicle architects set more and
do a $300,000 test to save $50,000,000. So the more ambitious goals for vehicle development stage-gates.
bottom line is it’s all about how fast you can do the These shortened engineering cycles directly affected engi-
math. neers who were now tasked with producing evaluations of
vehicle design that guaranteed increases in crashworthiness
When bringing their goals to their encounters with CrashLab, performance in less time than they had been accustomed to
crashworthiness managers began to perceive that the tech- previously. Despite these shorter design cycles, engineers
nology’s features constrained engineers’ ability to work faster were not anxious to speed-up all of their tasks. Engineers
than they could without the use of CrashLab. In many cases, consistently voiced a valued delineation between activities
engineers who were experienced at simulation work could such as model building or drafting, which required technical
build and analyze models faster by hand than they could if skill, but not detailed engineering intuition and judgment, and
they used CrashLab. Thus, the constraints that managers analysis activities, which required in-depth domain knowl-
perceived that CrashLab placed on engineers’ abilities to edge (i.e., physics, thermal dynamics). They hoped that by
work faster had to be overcome in order to realize the goal of reducing the amount of time they spent on routine or tedious
reducing the time it took to build and analyze simulation model building tasks they could devote the bulk of their
models. limited temporal resources to analyzing the results of their
simulations. Such a desire was understandable given that
Crashworthiness managers told EMD developers that participation in analysis activities placed engineers at the
CrashLab needed to include features that automated model center of decisions about product architecture and design.
set-up tasks in particular. To respond to these concerns, EMD Being at the center meant that engineers could play a primary
developers built auto-placement features into the tool. These role in the vehicle development process, which, in turn, would
features included scripts that would indicate to the engineer increase their status within Autoworks.
using them exactly where an accelerometer or section cut, for
example, should be placed per the standard work guidelines.15 It was in this environment that engineers in the Strut Group
Engineers would then have to simply execute a command began using CrashLab in regular production work. Conse-
indicating that they approved of these locations and the soft- quently, engineers interpreted CrashLab’s ability to auto-
ware would automatically write the coordinates into the input matically generate an input deck (material agency) as
deck—a text-based file that could be read by the solver. This affording them the opportunity to speed-up model building
automated procedure would eliminate the need for engineers activities, providing them with more time to analyze their
to perform these tasks manually and thus reduce the amount simulation results. Most engineers would open a CAD file in
of time it would take to set-up a model for analysis. a pre-processing tool, convert the CAD geometry to a finite
element data, and then export the model to CrashLab. Within
By 2003 CrashLab had a new set of features. To use Crash CrashLab’s user environment, engineers would follow the
Lab, engineers now had to follow the automated flow of steps automated routines for setting up a model (with instrumen-
that the tool presented to them and also abdicate to the tech- tation and load assessment points) and then allow CrashLab
nology the autonomy that they once had in creating their own to automatically generate an input deck indicating the
input decks. CrashLab now automated these steps and engi- parameters upon which the model would be solved. After
neers who used the tool had no choice but to allow it to do so. four months of using CrashLab in this way, engineers in the
Strut Group were setting up their models in similar ways to
one another and nearly all members of the group were using
CrashLab. Because their working practices were now
15
An accelerometer is an electromechanical device used to measure forces in coordinated, engineers could realistically talk with one
a vehicle impact. In physical crash tests, accelerometers are placed at various another about their analyses. Consequently, engineers bagan
areas on the vehicle to measure forces. In a simulation, nodes are selected as to talk to colleagues about design solutions that had worked
accelerometers at which the solver will determine the accelerative forces in
the model. A section cut is a plane placed through a vehicle member that is
in the past and to work together to troubleshoot problems in
used to measure the force in that particular member (section). a given simulation. Consequently, engineers began to

increase the frequency with which they consulted each other lations increased, engineers began to notice that it was very
about design solutions. Consider the following interchange difficult to compare results among them. CrashLab generated
between two engineers. Engineer 1 went to Engineer 2’s a separate standard report for each iteration of a simulation
cubicle to ask if changing the gauge (thickness) of a part test. By 2006, most engineers ran more than 20 simulations
would increase performance on a side impact test: to test a given change made to the geometry of a part or the
arrangement of parts in the vehicle, which meant that
E1: Could I change the thickness of the #2 bar?16 engineers had to deal with 20 reports generated by CrashLab.
E2: That can’t go down anymore because of seats. CrashLab did not provide capabilities to systematically com-
E1: Okay… pare the results of multiple iterations. To make such com-
E2: But also if you down-gauge it, it’s probably going to parisons, engineers had to use CrashLab to generate 20
deform more and pull. I mean you’ll start to see a separate reports (one for each iteration). Because the reports
little bit of a kink between the two front seat CrashLab produced were in HTML format, engineers had to
attachments for the driver. manually copy the important data from each report into an
E1: Really? Excel spreadsheet. Engineers could then use Excel to
E2: I mean the brackets are 0.8 mm right now. I think generate graphs and plots comparing the simulation results
the #2 bar is like 1.2 mm, right? You need like a two and take this information to their consultations with col-
mil stack. leagues to indicate what types of changes (e.g., to geometry
E1: Yeah, but to beef up those two pieces it would or material properties) produced the “best” results.
probably take away whatever you gained on the #2
bar, right? It became clear to engineers that CrashLab’s current material
E2: Maybe, but I’d at least start there with the test. configuration constrained their ability to compare results from
multiple iterations in a systematic manner. Specifically,
Such new consultation routines occurred commonly across CrashLab did not afford engineers the ability to plot results
analysis activities in the Strut Group, signaling an important from multiple iterations in a single chart for the sake of
change in the focus of crashworthiness engineering effort. comparison—what engineer’s called “cross-plots.” Engineers
thus began to complain to their managers that CrashLab
As engineers consulted one another about vehicle design, needed to have these new features. Crashworthiness
their goals began to shift. Because CrashLab’s material managers, who were keen on having engineers produce
agency provided them with the ability to conduct coordinated results faster, were eager to champion new features that could
analyses, engineers needed to be sure that they were armed speed up work while simultaneously increasing the robustness
with sufficient data points to engage in helpful discussions. of analyses. EMD developers were receptive to the demands
To generate more data points, engineers had to run more of crashworthiness managers and began to develop new
iterations of their simulation models. Running multiple itera- features in CrashLab that would allow engineers to effort-
tions was easy to do. After engineers used CrashLab to set-up lessly compare the results of multiple iterations. The tech-
a model for analysis, they had only to go into the input deck nical innovation made by EMD developers was to create code
and change the parameters they wished to test in the new that clustered performance metrics from various simulations
iteration. By following this practice, engineers took as their into one common report and, further, to produce cross-plots
emergent goal that analyses should be tested systematically to and graphs of these data points. Consequently, when engi-
provide as much information as possible for use in ongoing neers began to use the newest version of CrashLab, they
consultations with colleagues. CrashLab’s material agency found that the tool automatically performed the comparison
was imbricated with the agency of the engineers through functions they desired without extra effort on their part. The
changes in their routines around analysis activities. goal for more systematic analyses characterizing engineers’
human agency was imbricated with the material agency of
Imbrication 5 (Human → Material) automatic cross-plot formation through the addition of
iteration comparison features to CrashLab.
To achieve the goal of producing as many systematic analyses
as possible, engineers increased the number of iterations they Engineers normally brought the comparisons of their simula-
ran for a given crash test simulation. As the number of simu- tions to consultations with colleagues. Slowly, engineers
began to notice that the types of design changes that produced
good performance for one vehicle were sometimes similar to
16
The #2 bar is a lateral support (structural member) on the vehicle’s
the types of design changes that produced good performance
underbody. for another vehicle. Although engineers could provide anec-

dotal examples of certain changes that commonly produced The perspective advanced in this paper suggests that the
good performance outcomes, they had no way of system- answer to this question depends on how human and material
atically comparing the findings generated by engineers across agencies—the basic building blocks common to both routines
multiple vehicle programs. As one engineer commented, and technologies—are imbricated. As illustrated in the case
of CrashLab’s development and use, when an existing
I think it would be great if somebody took all of our material agency is imbricated with a new human agency
results, the ones that work and the ones that didn’t, (material → human) people may be likely to change their
and then do some analyses of them. If they did stuff routine, and when an existing human agency is imbricated
like that, they could tell us, “Hey, this sort of bracket with a new material agency (human → material) a technology
when it’s angled in this way improved performance changes. Thus routines and technologies, although distinct
on 9 out of 10 programs.” If they did that, we could empirical phenomena, are ontologically related in the sense
incorporate those learnings into our designs. But that they are both constituted by imbrications of human and
nobody does that. material agencies. As illustrated in Figure 1, a routine shares
its building blocks with a technology just as the technology
As crashworthiness managers came to realize that all of this
shares its building blocks with a routine. The result is that a
new data generated through the use of CrashLab could be
change in a technology at any given time is linked to the
mined systematically to suggest general design trends for
routines that came before it and will be linked to the routines
optimal crashworthiness performance, they worked with
that come after just as a change in routines is linked to the
Techpro, which specialized in the maintenance and deploy-
technologies that preceded and will follow it. If we drop the
ment of finite element technologies, to develop a new group
language of routines and technologies and speak at the lowest
that could analyze performance data from across Autoworks.
level of abstraction, human and material agencies are con-
This new Robust Synthesis Analysis (RSA) group was formed
from a mix of engineers with backgrounds in optimization and stantly imbricated with one another and this chain of imbrica-
statistical analyses. These engineers created new routines for tions occurs in a path dependent manner.
coding, analyzing, and comparing the data that engineers
generated in their simulations. By the end of my study in late What keeps human and material agencies in a continued
2006, the RSA group was planning to develop a new tool sequence of imbrication is that people draw on the infrastruc-
called CARDS (Computer-Aided Robust Design Solutions) ture created out of past imbrications (routines or technologies)
to automatically analyze the vast amounts of data generated to construct perceptions of affordances and constraint. The
by engineers in their simulation activities. construction of these perceptions creates a space of oppor-
tunity or frustration in which people are motivated to act
(Gaver 1996; Karat et al. 2000). To the extent that people
have the ability to alter their goals (a presumption that has
Discussion and Implications long undergirded discussions of human agency) and the
ability to change a technology’s features (a capability
This article began with the suggestion that employees in increasingly made common by flexible technologies), their
today’s organizations increasingly find themselves working perceptions of affordances and constraints may compel them
with flexible routines and flexible technologies. In recent to make new imbrications of human and material agencies,
years, scholars have argued that organizational routines are which then continues to produce changes in routines and
often designed to be flexible (Essen 2008; Howard-Grenville technologies. This perspective has several implications for a
2005). That is, people can alter the performance of a human agency approach to the organizing process.
routine—their patterns of social interaction—while still main-
taining its ostensive qualities—the broad understanding of
what the routine should do (Feldman and Pentland 2003). Incorporating the Role of Material Agency
Technologies are also increasingly flexible in the sense that into a Human Agency Approach
people have resources to reinvent, redesign, and reconfigure
their material features so that the technology does new things. For a number of years, IS researchers have been skeptical of
Thus, when people work with both flexible routines and talking about technology’s material agency because they have
flexible technologies and wish to change their work practices, not found a successful way of combining the important
it seems that they have a choice. Do they change the routine, insights from structuration and actor–network approaches to
or do they change the technology?

explicitly theorize the role of technologies in the production, Reconciling Ontological and Empirical
maintenance, and change of the organizing process. The Specifications of Routines and
imbrication metaphor recognizes that humans have agency, Technologies
and operationalizes human agency as people’s ability to form
and realize their goals. It also recognizes that technologies At the ontological level, scholars are beginning to suggest that
have agency, operationalizing material agency as tech- routines and technologies are indistinguishable phenomena
nology’s ability to act on its own. For human and material (Baptista 2009; Orlikowski and Scott 2008). As Leonardi
agencies to become imbricated, someone has to arrange them (2009, p. 299) suggests, “technologies are as much social as
in particular sequences. Thus, technology developers and they are material (in the sense that material features were
users actively imbricate their human agency with the material chosen and retained through social interaction) and [routines]
agency of the artifacts with which they work, as they experi- are as much material as they are social (in the sense that social
interactions are enabled and constrained by material pro-
ence that material agency in the real-time of their practice.
perties).” Yet, at the empirical level, technologies and rou-
tines are relatively easy to distinguish (Edmondson et al.
The ability to recognize the role of material agency within the
2001; Pentland and Feldman 2008). As Barley (1988, p. 46)
axiology of the human agency perspective means that the
notes, “if one were to ask individuals in actual organizations
imbrication metaphor may help in specifying why people who
what technologies they use in their work…the technology
can choose to change either their routines or their tech- would have a name and the informant could, at least in
nologies to better execute their work make the choices they principle, point to an instance of its use.” Thus, our current
do. Because its material agency is circumscribed by the set of understanding of the nature of the relationship between
features a technology possesses, the technology can only do routines and technologies evinces dissonance between our
so much. As Pentland and Feldman (2008, p. 243) observe, ontological specifications and our empirical observations.
because its material agency is limited by its feature set, a
toaster simply cannot be used as a cell phone, no matter how The imbrication metaphor provides one way to reconcile this
much someone wishes it could be. Thus technologies can dilemma. If we recognize that routines and technologies are
only do certain things. Of course, how people choose to inter- figurations (to use Latour’s term) or infrastructure (to use Star
pret the things a technology does, or whether they recognize and Ruhleder’s term) created from the imbrication of human
what those things are, is quite variable. For this reason, a and material agencies, it seems more appropriate to make both
technology’s material agency can encourage people to ontological and empirical claims about the relationship
perceive that the technology sometimes constrains their ability between agencies, as opposed to the relationship between
to achieve their goals and other times affords them with routines and technologies. In other words, the imbrication
possibilities to develop new ones. As the case illustrated in metaphor breaks down the walls between studies of routines
this paper demonstrates, a person’s perception of whether a and studies of technology by suggesting that researchers who
technology constrains desired action or affords new action investigate them empirically are actually studying the same
depends on prior imbrications. Without perceptions of con- underlying process: the imbrication of human and material
straint or affordance, people would likely continue to use agencies. When we look at the theoretical string of agencies
existing technologies and routines in inertial ways. As presented in Figure 1, we don’t naturally see any routines or
Orlikowski (2000) shows, to the extent that technologies and technologies; we just see agencies. Of course, our specifi-
routines allow people to accomplish their goals, no change in cation of some imbrication of agencies as either a routine or
either takes place. Thus it is important to recognize the role a technology is a by-product of empirical operationalization.
that changing material agencies (enabled by flexible techn- Thus, we might be more accurate to first conceptualize
organizations as imbricated systems of human and material
ologies) can play in the ongoing process of organizing.
agencies and, if we start from this point, ask how certain
Incorporating the notion of material agencies into a human
imbrications of these agencies produces particular figurations.
agency approach helps to explain how people make choices
Figure 2 performs this exercise by showing how certain
about whether they will change their flexible routines, or their
human and material agencies were imbricated by strategic
flexible technologies. Moreover, in contrast to existing
actors at Autoworks, as they constructed perceptions of
human agency approaches which see material agency as a constraints and affordances, such that routines and
potential threat to human agency, the imbrication lens views technologies were constituted and changed.
material agency more neutrally: its influence as either an
affordance or a constraint depends on the perceptions people The benefit of talking about imbrications of human and
construct about it. material agencies is that imbrication emphasizes interweaving

as opposed to impact. Many IS researchers and even more in which human and material agencies change in response to
organizational researchers have seemed tentative in moving one another. Their concordant changes, influenced by past
technology into a more central role in their theories about patterns of imbrication, constitute and bring reconfigurations
organizational routines because they have been unable to to the routines and technologies through which organizing is
speak of its role without resorting to deterministic language accomplished. As the example at Autoworks has shown,
(Leonardi and Barley 2008). By suggesting that human and when both routines and technologies are flexible, human and
material agencies become interlocked in repeated patterns (as material agencies are in a process of continual imbrication
opposed to saying that routines and technologies impact each such that the organizational structures they constitute are
other at a given point in time), the imbrication lens recognizes always in flux. Thus, as scholars begin to push us to think
that both human and material agencies are necessary for orga- about organizing as a sociomaterial process, the imbrication
nizing to occur; their imbrication is what produces comple- metaphor helps us to explain how the social and the material
mentary changes in technologies and in people’s interactions. become interwoven in the first place and continue inter-
locking in ways that produce the infrastructures that people
From this vantage point, students of routines and technology use to get their work done.
can begin to view structure (metaphorically and literally) as
the product of imbrication. Currents in a river have a struc-
ture (a direction of flow) that is made visible through patterns Acknowledgments
of rock imbrication. By examining imbrication patterns in
fluvial settlements, geologist can “see” how a river is flowing, I thank Steve Barley, Wanda Orlikowski, and Carol Saunders for
how it has flowed in the past, and, in some cases, predict how helpful comments on ideas presented in this paper. An earlier
it will flow in the future. The structuring process is a lot like version of this manuscript was presented at the 2008 meeting of the
the flow of current in a river. While the process occurs, there Academy of Management where it won the Gerardine DeSanctis
are few visible traces by which to observe it. Firms often try Dissertation Award from the Organizational Communication and
to do so by the creation of an organization chart, which maps Information Systems Division and was chosen as a finalist for the
ideal flows of interaction based on hierarchical relations, and William H. Newman Award. Funding for this research was made
organizational researchers often use such charts to capture a possible by a grant from the National Science Foundation (ITR
0427173).
glimpse of structure at a given point in time. Organization
charts, however, only capture the social components of the
organizing process. For example, if Figure 2 were comprised
only of square-shaped boxes (routines), the only other attri- References
bute we would be able to infer about the dynamics of the
Alavi, M., and Leidner, D. E. 2001. “Review: Knowledge Manage-
work system would be the hexagonal-shaped boxes (human
ment and Knowledge Management Systems: Conceptual Found-
agency). We would not know why, over time, the content of ations and Research Issues,” MIS Quarterly (25:1), pp. 107-136.
the square or the diamond-shaped boxes changed. To know Azad, B., and King, N. 2008. “Enacting Computer Workaround
these reasons, we also must treat the square-shaped boxes Practices Within a Medication Dispensing System,” European
(technology’s features) and diamond-shaped boxes (material Journal of Information Systems (17:3), pp. 264-278.
agency) as constitutive features of the structuring process. By Baptista, J. 2009. “Institutionalization as a Process of Interplay
focusing our efforts on how human and material agencies Between Technology and its Organizational Context of Use,”
become imbricated, we can begin to visually conceive what Journal of Information technology (24), pp. 305-319.
the structuring process looks like. As the imbrication frame- Barad, K. 2003. “Posthumanist Performativity: Toward an Under-
work suggests, structuring involves simultaneous and standing of How Matter Comes to Matter,” Signs (28:3), pp.
interactive changes between the features and routines of the 801-831.
technology. By mapping these changes over time, we may be Barley, S. R. 1988. “Technology, Power, and the Social Organiza-
able to gain new insights into the dynamics of socio-technical tion of Work: Towards a Pragmatic Theory of Skilling and
change and its role in the constitution of organizations. Deskilling,” Research in the Sociology of Organizations (6), pp.
33-80.
Barley, S. R. 1990. “The Alignment of Technology and Structure
Through Roles and Networks,” Administrative Science Quarterly
Conclusion (36:1), pp. 61-103.
Boudreau, M.-C., and Robey, D. 2005. “Enacting Integrated Infor-
Routines are intimately tied to the technologies that enable mation Technology: A Human Agency Perspective,” Organiza-
social interaction because of the imbricated nature of human tion Science (16:1), pp. 3-18.
and material agencies. As this paper has shown, the Callon, M. 1991. “Techno-Economic Networks and Irreversibility,”
increasing flexibility of routines and technologies in organi- in A Sociology of Monsters: Essays on Power, Technology and
zations affords an opportunity to look more closely at the way Domination, J. Law (ed.), London: Routledge, pp. 132-161.

Chae, B., and Poole, M. S. 2005. “The Surface of Emergence in Gilbreth, F. B. 1911/1993. Motion Study: A Method for Increasing
Systems Development: Agency, Institutions, and Large-Scale the Efficiency of the Workman, London: Routledge/Thoemmes
Information Systems,” European Journal of Information Systems Press.
(14), pp. 19-36. Heidegger, M. 1959. An Introduction to Metaphysics, New Haven,
Chu, T. H., and Robey, D. 2008. “Explaining Changes in Learning CT: Yale University Press.
and Work Practice Following the Adoption of Online Learning: Howard-Grenville, J. A. 2005. “The Persistence of Flexible Organi-
A Human Agency Perspective,” European Journal of Infor- zational Routines: The Role of Agency and Organizational
mation Systems (17:1), pp. 79-98. Context,” Organization Science (16:6), pp. 618-636.
Ciborra, C. 2006. “Imbrication of Representations: Risk and Hutchby, I. 2001. “Technologies, Texts and Affordances,” Sociology
Digital Technologies,” Journal of Management Studies (43:6), (35:2), pp. 441-456.
pp. 1339-1356. Jones, M. R. 1998. “Information Systems and the Double Mangle:
Collins and Kusch 1998 Steering a Course between the Scyla of Embedded Structure and
Constantinides, P., and Barrett, M. 2005. “Negotiating ICT the Charydbis of Strong Symmetry,” in Information Systems:
Development and Use: The Case of a Telemedicine System in Current Issues and Future Changes, T. J. Larsen, L. Levine, and
the Healthcare Region of Crete,” Information and Organization J. I. DeGross (eds.), Laxenburg, Austria: IFIP Press, pp. 287-302.
Jones, M. R., and Karsten, H. 2008. “Giddens’s Structuration
(16:1), pp. 27-55.
Theory and Information Systems Research,” MIS Quarterly (32:1),
Cooren, F. 2004. “Textual Agency: How Texts Do Things in
pp. 127-157.
Organizational Settings,” Organization (11:3), pp. 373-393.
Karat, J., Karat, C.-M., and Ukelson, J. 2000. “Affordances, Moti-
Cyert, R. M., and March, J. G. 1963. A Behavioral Theory of the
vation and the Design of User Interfaces,” Communications of the
Firm (2nd ed.), Englewood Cliffs, NJ: Prentice Hall.
ACM (43:8), pp. 49-51.
DeSanctis, G., and Poole, M. S. 1994. “Capturing the Complexity
Lapointe, L., and Rivard, S. 2005. “A Multilevel Model of
in Advanced Technology Use: Adaptive Structuration Theory,” Resistance to Information Technology Implementation,” MIS
Organization Science (5:2), pp. 121-147. Quarterly (29:3), pp. 461-491.
Dodgson, M., Gann, D. M., and Salter, A. 2007. “‘In Case of Fire, Latour, B. 1992. “Where Are the Missing Masses? The Sociology
Please Use the Elevator’: Simulation Technology and Organi- of a Few Mundane Artifacts,” in Shaping Technology/Building
zation in Fire Engineering,” Organization Science (18:5), pp. Society: Studies in Sociotechnical Change, W. E. Bijker, and J.
849-864. Law (eds.), Cambridge, MA: MIT Press, pp. 225-258.
Dougherty, D. J. 1992. “Interpretive Barriers to Successful Product Latour, B. 1993. We Have Never Been Modern, Cambridge, MA:
Innovation in Large Firms,” Organization Science (3:2), pp. Harvard University Press.
179-202. Latour, B. 1994. “On Technical Mediation: Philosophy, Sociology,
Edmondson, A. C., Bohmer, R. M. and Pisano, G. P. 2001. Geneaology,” Common Knowledge (3), pp. 29-64.
“Disrupted Routines: Team Learning and New Technology Latour, B. 1999. Pandora’s Hope, Cambridge, MA: Harvard
Implementation in Hospitals,” Administrative Science Quarterly University Press.
(46:4), pp. 685-716. Latour, B. 2005. Reassembling the Social: An Introduction to
Emirbayer, M., and Mische, A. 1998. “What Is Agency?,” Actor–Network Theory, Oxford, UK: Oxford University Press.
American Journal of Surgery (103:4), pp. 962-1023. Leonardi, P. M. 2007. “Activating the Informational Capabilities of
Essen, A. 2008. “Variability as a Source of Stability: Studying Information Technology for Organizational Change,” Organi-
Routines in the Elderly Home Care Setting,” Human Relations zation Science (18:5), pp. 813-831.
(61:11), pp. 1617-1644. Leonardi, P. M. 2009. “Crossing the Implementation Line: The
Fayard, A.-L., and Weeks, J. 2007. “Photocopiers and Water- Mutual Constitution of Technology and Organizing Across
Coolers: The Affordances of Informal Interaction,” Organization Development and Use Activities,” Communication Theory (19),
Studies (18:5), pp. 605-634. pp. 277-309.
Feldman, M. S., and Pentland, B. T. 2003. “Reconceptualizing Leonardi, P. M. 2010. “Innovation Blindness: Culture, Frames, and
Organizational Routines as a Source of Flexibility and Change,” Cross-Boundary Problem Construction in the Development of
Administrative Science Quarterly (48), pp. 94-118. New Technology Concepts,” Organization Science (DOI:
Garud, R., and Rappa, M. A. 1994. “A Socio-Cognitive Model of 10.1287/orsc.1100.0529), forthcoming.
Technology Evolution: The Case of Cochlear Implants,” Organi- Leonardi, P. M., and Barley, S. R. 2008. “Materiality and Change:
zation Science (5:3), pp. 344-362. Challenges to Building Better Theory About Technology and
Gaver, W. 1996. “Affordances for Interaction: The Social Is Organizing,” Information and Organization (18), pp. 159-176.
Material for Design,” Ecological Psychology (8:2), pp. 111-129. Mackay, H., and Gillespie, G. 1992. “Extending the Social Shaping
Gibson, J. J. 1986. The Ecological Approach to Visual Perception, of Technology Approach: Ideology and Appropriation,” Social
Studies of Science (22), pp. 685-716.
Hillsdale, NJ: Lawrence Erlbaum Associates.
Majchrzak, A., Rice, R. E., Malhotra, A., King, N., and Ba, S. L.
Giddens, A. 1984. The Constitution of Society, Berkeley, CA:
2000. “Technology Adaptation: The Case of a Computer-
University of California Press.
Supported Inter-Organizational Virtual Team,” MIS Quarterly
Giddens, A., and Pierson, C. 1998. Conversations with Anthony
(24:4), pp. 569-600.
Giddens, Cambridge, UK: Polity Press.

Markus, M. L. 2004. “Technochange Management: Using IT to Schultze, U., and Boland, R. J. 2000. “Knowledge Management
Drive Organizational Change,” Journal of Information Tech- Technology and the Reproduction of Knowledge Work Practices,”
nology (19), pp. 4-20. Strategic Information Systems (9), pp. 193-212.
Markus, M. L., and Silver, M. S. 2008. “A Foundation for the Schultze, U., and Orlikowski, W. J. 2004. “A Practice Perspective
Study of IT Effects: A New Look at DeSanctis and Poole’s on Technology-Mediated Network Relations: The Use of Internet-
Concepts of Structural Features and Spirit,” Journal of the Based Self-Serve Technologies,” Information Systems Research
Association for Information Systems (9:10/11), pp. 609-632. (15:1), pp. 87-106.
Norman, D. A. 1990. The Design of Everyday Things, New York: Star, S. L., and Ruhleder, K. 1996. “Steps Toward an Ecology of
Doubleday. Infrastructure: Design and Access for Large Information Spaces,”
Norman, D. A. 1999. “Affordance, Conventions, and Design,” Information Systems Research (7:1), pp. 111-134.
Interactions (6:3), pp. 38-43. Tabrizi, B. N. 2005. Accelerating Transformation: Process Innova-
Orlikowski, W. J. 1992. “The Duality of Technology: Rethinking tion in the Global Information Technology Industry, Sydney:
the Concept of Technology in Organizations,” Organization Universal Publishers.
Science (3:3), pp. 398-427. Taylor, F. W. 1911/1998. The Principles of Scientific Management,
Orlikowski, W. J. 2000. “Using Technology and Constituting New York: Dover.
Taylor, J. R. 2001. “Toward a Theory of Imbrication and Organiza-
Structures: A Practice Lens for Studying Technology in Organi-
tional Communication,” The American Journal of Semiotics
zations,” Organization Science (11:4), pp. 404-428.
(17:2), pp. 269-298.
Orlikowski, W. J. 2005. “Material Works: Exploring the Situated
Taylor, J. R., Groleau, C., Heaton, L., and Van Every, E. 2001. The
Entanglement of Technological Performativity and Human
Computerization of Work: A Communication Perspective,
Agency,” Scandinavian Journal of Information Systems (17:1),
Thousand Oaks, CA: Sage Publications.
pp. 183-186.
Taylor, J. R., Groleau, C., Heaton, L., and Van Every, E. 2007.
Orlikowski, W. J., and Scott, S. V. 2008. “Sociomateriality: Chal- “Communication as the Modality of Structuration,” in Theorizing
lenging the Separation of Technology, Work and Organization,” Communication Readings Across Traditions, R. T. Craig and H. L.
The Academy of Management Annals (2:1), pp. 433-474. Mueller (eds.), Thousand Oaks, CA: Sage Publications, pp.
Pentland, B. T., and Feldman, M. S. 2007. “Narrative Networks: 391-404.
Patterns of Technology and Organization,” Organization Science Vaast, E., and Walsham, G. 2005. “Representations and Actions:
(18:5), pp. 781-795. The Transformation of Work Practices with IT Use,” Information
Pentland, B. T., and Feldman, M. S. 2008. “Designing Routines: and Organization (15), pp. 65-89.
On the Folly of Designing Artifacts, While Hoping for Patterns Volkoff, O., Strong, D. M., and Elmes, M. B. 2007. “Technological
of Action,” Information and Organization (18:4), pp. 235-250. Embeddedness and Organizational Change,” Organization Science
Pentland, B. T., and Rueter, H. H. 1994. “Organizational Routines (18:5), pp. 832-848.
as Grammars of Action,” Administrative Science Quarterly (39), Walsham, G. 2002. “Cross-Cultural Software Production and Use:
pp. 484-510. A Structuration Analysis,” MIS Quarterly (26:4), pp. 359-380.
Pickering, A. 1995. The Mangle of Practice: Time, Agency, and Walsham, G. 2005. “Agency Theory: Integration or a Thousand
Science, Chicago: University of Chicago Press. Flowers,” Scandinavian Journal of Information Systems (17:1), pp.
Pickering, A. 2001. “Practice and Posthumanism: Social Theory 153-158.
and a History of Agency,” in The Practice Turn in Contemporary Zack, M. H., and McKenney, J. L. 1995. “Social Context and Inter-
Theory, T. R. Schatzki, K. Knorr-Cetina, and E. von Savigny action in Ongoing Computer-Supported Management Groups,”
(eds.), London: Routledge, pp. 163-174. Organization Science (6:4), pp. 394-422.
Pollock, N., Williams, R., and D’Adderio, L. 2007. “Global Zamutto, R. G., Griffith, T. L., Majchrzak, A., Dougherty, D. J., and
Software and its Provenance: Generification Work in the Faraj, S. 2007. “Information Technology and the Changing Fab-
Production of Organizational Software Packages,” Social Studies ric of Organization,” Organization Science (18:5), pp. 749-762.
of Science (37), pp. 254-280.
Poole, M. S., and DeSanctis, G. 1992. “Microlevel Structuration in
Computer-Supported Group Decision Making,” Human Com- About the Author
munication Research (19:1), pp. 5-49.
Poole, M. S., and DeSanctis, G. 2004. “Structuration Theory in Paul M. Leonardi is an assistant professor of Communication
Information Systems Research: Methods and Controversies,” Studies, Industrial Engineering and Management Sciences, and (by
Handbook of Information Systems Research, M. E. Whitman and courtesy) Management and Organizations at Northwestern Univer-
A. B. Woszczynski (eds.), Hershey, PA: Idea Publishing, pp. sity, where he holds the Breed Junior Chair in Design. He earned
206-249. his Ph.D. in Management Science and Engineering from Stanford
Rose, J., Jones, M. R., and Truex, D. 2005. “Socio-Theoretic University. Paul’s research explores how the process of organizing
Accounts of IS: The Problem of Agency,” Scandinavian Journal shapes new technological artifacts and how those very technologies
of Information Systems (17:1), pp. 133-152. that organizations develop either reinforce or change the process of
Sassen, S. 2006. Territory, Authority, Rights: From Medieval to organizing. He has explored these issues in the contexts of high-
Global Assemblages, Princeton, NJ: Princeton University Press. tech mergers, IT service work, and automotive design.

Copyright of MIS Quarterly is the property of MIS Quarterly & The Society for Information Management and
its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for individual use.

Tech Affordance Mate Agenc

Uploaded by

Copyright:

Available Formats

Tech Affordance Mate Agenc

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Tech Affordance Mate Agenc

Uploaded by

Copyright:

Available Formats

RESEARCH ARTICLE

WHEN FLEXIBLE ROUTINES MEET FLEXIBLE TECHNOLOGIES:

Introduction1 geographical positioning systems to accurately identify the

MIS Quarterly Vol. 35 No. 1 pp. 147-167/March 2011 147

148 MIS Quarterly Vol. 35 No. 1/March 2011

MIS Quarterly Vol. 35 No. 1/March 2011 149

150 MIS Quarterly Vol. 35 No. 1/March 2011

MIS Quarterly Vol. 35 No. 1/March 2011 151

152 MIS Quarterly Vol. 35 No. 1/March 2011

MIS Quarterly Vol. 35 No. 1/March 2011 153

154 MIS Quarterly Vol. 35 No. 1/March 2011

MIS Quarterly Vol. 35 No. 1/March 2011 155

156 MIS Quarterly Vol. 35 No. 1/March 2011

Table 1. Summary of Data Collection

MIS Quarterly Vol. 35 No. 1/March 2011 157

Note: Perforated boxes represent process of Imbrication 5

158 MIS Quarterly Vol. 35 No. 1/March 2011

MIS Quarterly Vol. 35 No. 1/March 2011 159

Everything here is about speed. Simulations are

160 MIS Quarterly Vol. 35 No. 1/March 2011

MIS Quarterly Vol. 35 No. 1/March 2011 161

162 MIS Quarterly Vol. 35 No. 1/March 2011

MIS Quarterly Vol. 35 No. 1/March 2011 163

164 MIS Quarterly Vol. 35 No. 1/March 2011

MIS Quarterly Vol. 35 No. 1/March 2011 165

166 MIS Quarterly Vol. 35 No. 1/March 2011

MIS Quarterly Vol. 35 No. 1/March 2011 167

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.