You are on page 1of 41

AOM Submission # 17455

PATTERNS OF DISCOURSE FOR DIGITAL TECHNOLOGIES IN THE ERA OF

INCREMENTAL CHANGE

Uri Gal

University of Sydney

Nicholas Berente

University of Georgia

Friedrich Chasin

University of Münster

Abstract

In this paper we explore the discourse around six digital technologies after their architectures are

stabilized - the period of time known as the "era of incremental change." Using a novel computational

technique to assess discourse from Wikipedia, we find that the patterns of discourse around digital

technologies vary based on the level of abstraction for the digital technology. The discourse around

digital devices, which represent the lowest level of abstraction, stabilizes relatively quickly. Whereas

digital concepts, which represent technologies at the highest level of abstraction, never fully stabilize and

emergence and contestant continues. The discourse around digital services, which are at a mid-level of

abstraction, display features of both stabilization and continued dynamism. This research extend the work

on technology lifecycles to begin addressing the digital technologies - which are particularly open-ended,

emergent, and discourse-intensive.


INTRODUCTION

It has long been understood that social and political dynamics within industries shape patterns of

technological innovation as much as the technologies themselves do (Tushman & Rosenkopf

1992; Dokko et al 2012). After a period of competitive turmoil which is marked by political

contestation, competition, negotiation, and sensemaking around alternative new technologies,

“dominant design” architecture emerges, which is followed by a period of social stabilization

until the next architectural upheaval (Anderson & Tushman 1990; Henderson & Clark 1990).

This period that follows the architectural stabilization of a dominant design is typically referred

to as the “era of incremental change” in the technology lifecycle (Anderson & Tushman 1990).

Although the era of incremental change is typically thought to be theoretically uninteresting

(Dokko et al 2012), recent work has found that the period that follows the stabilization of a

technology architecture can actually be quite dynamic. There is often still be a significant

amount of innovation with technologies within the bounds of the architecture (Murmann and

Frenken. 2006; Funk. 2009; Lee & Berente 2013). Further, technology stabilization does not

necessarily mean that there is lack of social dynamics – indeed technological stabilization might

be better conceived of as an ongoing accomplishment that represents a continually contested,

negotiated order among stakeholders (Dokko et al 2012).

Recent decades have witnessed the increasing ubiquity of digital technologies whose nature

increases the significance of social dynamics to the emergence and evolution of those

technologies. Digital technologies are products and services that are based on microprocessors

and software, and, as such, are programmable, editable and extensible and therefore continually

evolve as they are appropriated (Yoo et al 2010). For these reasons, digital technologies are

2
distributed and open – they evolve as people interact with them and with each other – and are

never fixed in the way that fully physical technologies are (Kallinikos et al 2013; Germonprez et

al 2011). Thus, social dynamics continue to shape digital technologies and their use long after

architectural stabilization (Yoo et al 2010). Although others have found that the era of

incremental change can be interesting from a perspective of the social dynamics (Dokko et al

2012), these social dynamics become even more relevant in the context of digital technologies.

In this paper we explore the social dimensions involved in the evolution of digital technologies

during the era of incremental change - after the architecture of the digital technologies have

stabilized. We do so by analyzing the discourse that emerges in the wake of a stabilized

technological architecture. We focus on the discourse around the technologies to understand

social dynamics because discourses reflect, carry, and instantiate those dynamics and the relevant

interests, interpretations, and political concerns. It is through discourse that organizational,

social, and material realities take form and gain meaning (Heracleous & Barrett 2001; Phillips et

al 2004; Taylor & Van Every 2000). Because people make sense of digital technologies

discursively (Griffith 1999; Berente et al 2011), these technologies are not merely described by

the discourse that surrounds them, but their very nature and meaning form and morph as this

discourse emerges and evolves over time.

To examine the discourse around the evolution of these digital technologies we draw upon a

novel computational method rooted in Moscovici’s theory of social representations (Moscovici

2000). We do this by systematically tracking changes in the content and collaboration patterns of

contributors to online Wikipedia articles about different digital technologies. We distinguish

among three broad categories of technologies based on their level abstractness: digital devices

(physical artifacts, e.g. iPad and Wii) which have a distinct material manifestation are the least

3
abstract; digital services (Internet-based applications, e.g. Tumblr and Twitter) which do not

have a tactile manifestation are more abstract; and digital concepts (categories of computing

phenomena, e.g. “cloud computing” and “smartphones”) are the most abstract.

We trace the evolution of the discourse during the era of incremental change for different digital

innovations at varying levels of abstraction to identify how the social dynamics around those

technologies change and stabilize over time. We find that the discourse around and meaning of

digital technologies stabilize fairly quickly after an initial sensemaking process. With more

abstract digital services, their discourse and meaning continue to fluctuate long into the era of

incremental change before it stabilizes. With the most abstract digital concepts, their discourse

and meaning never reach stabilization.

The remainder of the paper is organized as follows. First, we introduce the concepts of

technology lifecycle and the era of incremental change, and motivate the need to understand

discourse in the evolution of digital technologies. Second, we describe our approach for

understanding discourse using the theory of social representations, and outline a novel

computational technique. We conclude by presenting the outcomes of our analysis and proposing

patterns of discursive evolution for different categories of digital technologies.

THE ERA OF INCREMENTAL CHANGE, DIGITAL TECHNOLOGIES, AND

DISCOURSE

The technology lifecycle model (Anderson & Tushman 1990) describes the process by which the

structure of technologies changes and stabilizes in an industry. Technology architectures undergo

cycles of punctuated stabilization and change. At first, there are typically a variety of

4
architectural designs competing for dominance in an industry for a given class of technology,

and there is a period of competition among potential architectures known as the “era of turmoil.”

At some point one of those architectural designs “wins” and that design becomes the standard

architecture for that class of technology – what is sometimes referred to as a “dominant design”

(Anderson & Tushman 1990). After a standard architecture is established, there is a period of

incremental innovation around that architecture known as the “era of incremental change” which

continues until at some point alternative architectures drive another period of turmoil. These

three stages: turmoil, stabilization, and incremental innovation, cycle throughout the life of a

technology.

The transition between the stages is driven by social, political, and technological mechanisms

(Tushman & Rosenkopf 1992). Technology does not stabilize as a dominant design in an

industry solely because a new technology is available. People must make sense of this

technology. As they interpret the possibilities for its use, the technology itself changes to

accommodate perceptions of its potential (Bijker 1997). Technological standards have as much

to do with negotiation and social agreements about the technology as with the technology itself

(Williams & Edge 1996). Thus, discursive practices such as negotiation and argumentation are

critical in the era of turmoil leading up to stabilization of the architecture. However, after this

stabilization takes place – in the era of incremental change – social dynamics have traditionally

been thought to settle down. The rich discourse leading up to the emergence of a standard will

level off as people accept the standard and move on to appropriation of the standard and

complementary, incremental innovation. This is what Tushman & Rosenkopf (1992) imply when

they indicate that social dynamics stabilize in the era of incremental change.

5
Despite the centrality of social dynamics to the technology lifecycle model, there has been little

research into the social elements of lifecycle dynamics around any technology, let alone digital

technology. Recently, however, Dokko and colleagues (2012) point out that this social

stabilization more closely resembles Strauss’s negotiated order. Although social dynamics may

reach a state where they appear stable, stability across a diverse set of stakeholders must be

continually negotiated in a dynamic process where the meaning of the technology is shaped.

Thus the social stabilization is enabled through continuous discursive practice.

This observation is consistent with the view of how social phenomena are interpreted and

reproduced through discourse (Phillips et al 2004). Discourses are collections of spoken or

written texts which are produced and interpreted by a community (Fairclough 1992).

Communities make sense of situations through discourse, and it is this discourse that then shapes

subsequent community action and continued discourse. Indeed, key social structures and forms

such as organizations and institutions, as well as changes in these structures and forms, are

constructed and enabled largely through discourse (Heracleous & Barrett 2001; Taylor & Van

Every 2000).

Because discourse is at the core of social action (Habermas 1984; Wittgenstein 1969), and

technologies are in no small part socially constructed (Latour 1990, Bijker 1997), we would

expect discourse to be important in the construction and stabilization of technological

architectures. In the era of incremental change, one would expect the changes in discourse to

diminish (as compared to previous levels, see Tushman & Rosenkopf 1992) and persist at a low

level in order to reflect and maintain that stability (Dokko et al 2012). However, we anticipate

that due to the uniquely open nature of digital technologies, the discourse around them will not

stabilize so readily. Digital technologies are programmable, addressable, sensible,

6
communicable, memorizable, traceable, and associable – qualities which together enable them to

always be evolving in use (Yoo et al 2011). Further, they are socially distributed in the sense that

they span multiple modes of life (Kallinikos et al 2013). Therefore digital technologies enable

fundamentally more dynamic modalities for action than their analog counterpart artifacts

(Kallinikos 2009), which will be reflected in accordingly dynamic discursive patterns. Thus,

even after the architecture for a particular digital technology is stabilized, we expect the

discourse around it to keep evolving as the meaning of the technology continues to shift.

Next we introduce the theory of social representations as a way of understanding discourse, and

then present our study.

SOCIAL REPRESENTATIONS THEORY (SRT)

The central idea of social representations theory (SRT) is that people’s relationship with the

world is invariably mediated by a layer of socially constructed and continuously-evolving

symbols, or representations, which serve to render the world meaningful for social actors

(Moscovici 2000). According to the theory, things in the world – conceptual or material - do not

have an inherent meaning. Rather, they acquire meaning when they are represented by social

groups in an ongoing discursive process. Phenomena, events, or artifacts only become social

reality by virtue of their representations that the community forms. Only by being represented by

a group of people by means of familiar conceptual devices can a phenomenon, event, or artifact

7
become a social object that can be perceived, characterized, compared to other social objects,

and used in language and action (Wagner et al. 1996). Digital technologies evolve in conjunction

with a continuous process of social representation (Gal & Berente 2008).

The Process of Representing

Representational activity is a social process that is often triggered when something disruptive

threatens socially shared perceptions of reality (Moscovici 2000). This can be a new and

unfamiliar phenomenon, or an unexpected characteristic of a familiar phenomenon, which group

members lack the cognitive vocabulary to describe and name. This unfamiliarity creates a sense

of incompleteness and emphasizes the “actuality of something absent” (Moscovici 2000 p.38).

To familiarize the unfamiliar, group members form new social representations in a discursive

process that has two components: anchoring and objectification.

Anchoring. Whenever something unfamiliar is experienced, such as a radical societal change or

technological innovation, a process of familiarizing is initiated by means of anchoring (Wagner

et al. 1996). Anchoring entails the positioning of the unknown in familiar conceptual categories.

The choice of a suitable class of categories is based on a comparison of the unfamiliar to

prototypes considered to represent the corresponding class (Moscovici 1984). For example, in its

early stage, the unfamiliar phenomenon of HIV/AIDS (before acquiring this name) was anchored

in terms of a ‘gay plague’ or ‘gay cancer’ (Farr 1993). Thus, the new phenomenon of HIV/AIDS

was initially understood in terms of, and took on the qualities associated with, the plague or

cancer. Anchoring is a dynamic process that reflects group members’ changing perceptions of

different aspects in their environment. Anchors are thus an integral part of thinking in general;

“there is no thought or perception without anchor” (Moscovici 2000, p.48).

8
Objectification. In addition to anchoring a new phenomenon and placing it in familiar categories,

further discursive activities among community members lead to an objectified representation in

the form of a metaphor, symbol, or image (Wagner et al, 1996). Objectification is the process

whereby socially represented knowledge acquires a concrete and distinct form, or representation.

Objectification involves the development of a signifier that stands for the phenomenon or object

that it represents (Gal and Berente 2008). The representation captures the essence of the

phenomenon and weaves it into the social fabric of the group’s common stock of knowledge. For

example, that HIV/AIDS is readily and widely recognized today as a medical condition in its

own right, is an indication that this representation has been objectified.

Social Representations in Wikipedia

Wikipedia is not just an encyclopedia that holds knowledge about the world. It is also the place

where this knowledge is created: a place of dynamic discourse and social engagement, where

contributors collaborate, but also intensely debate and disagree about how to represent events,

people, and objects in their world. These debates take place as contributors express their opinions

by changing article content and participate in discussions on the corresponding talk pages

(Hansen et al 2009). Contributors can see and engage with others’ views by accessing the

revision history and observing previous discussions regarding the article at hand. In doing so,

contributors build upon the efforts of others in order to reach, if only temporarily, a collective

agreement about the meaning of the topic at hand. This reflects the discursive process of

representing, wherein meaning is collectively and continually shaped through anchoring and

objectification.

9
To examine the evolution of social representations on Wikipedia it is necessary to operationalize

the theory and identify corresponding Wikipedia concepts. Starting with social representations

themselves, we maintain that each one is reflected in a Wikipedia article. Social representations

name various aspects of the world, thereby orientating people and providing a basis for their

communication. Wikipedia articles serve this very function; they give a unique name to different

domains of the world and reflect a socially negotiated meaning of that domain in a particular

point in time (Hepp et al 2007).

Anchoring and objectification occur within and across Wikipedia articles. Anchoring can be

traced by examining internal links in one article that point to other articles. Anchoring involves

explaining an unfamiliar phenomenon in terms of a familiar one, whose nature is deemed to be

relevant to the understanding of the unfamiliar phenomenon (Moscovici 2000). This is what

internal links on Wikipedia aim to achieve: to create an association between a topic that requires

elaboration and explanation, and other known topics that are perceived to be relevant to

understanding the topic at hand. An internal link anchors the current representation in terms of an

existing one.

Objectification is the process whereby socially represented knowledge gains a distinct and

concrete form. Objectification in Wikipedia is best operationalized as the act of linking other

articles to the article at hand. Wikipedia articles that contain a reference to the social

representation under study in the form of a link are therefore relevant for tracing the

objectification process. A growing number of links that point to the social representation under

study would thus reflect an ongoing objectification process, as it would indicate that other

phenomena are explained (i.e. anchored) in terms of the representation under study1.

1
Due to the AoM page limit, in this paper we will focus exclusively on the anchoring process.

10
WIKIGEN: A NOVEL COMPUTATIONAL TOOL FOR STUDYING SOCIAL

REPRESENTATIONS ON WIKIPEDIA

In order to trace the evolution of discourse around digital technologies in Wikipedia, we utilize a

web-based analytical tool, WikiGen (Chasin et al 2013). By connecting to Wikipedia’s live

databases over the platform’s Application Programming Interface (API), WikiGen generates

collaboration and anchoring statistics based on past revisions of any chosen article (in 196

languages). These statistics are instrumental to studying how social representations emerge and

evolve on Wikipedia. In what follows we will outline the tool’s main features that we use in our

analysis.

Collaboration Statistics

Representing unfolds over time through collaboration among contributors and plays out through

the editing of Wikipedia articles. To track this process, WikiGen generates two collaboration

statistics: (1) number of edits; and (2) number of editors. To capture the dynamics of the

collaboration process in greater detail, WikiGen differentiates between minor and major edits

based on data provided by Wikipedia. Minor edits include grammatical and formatting changes.

Major edits involve changes to article content, references, or visible tags. Figures 1a and 1b

illustrate the overall (minor and major) collaboration statistics plotted for the article Adobe since

its inception on Wikipedia until the end of 2014:

11
Figure 1a. Number of overall edits for the article Adobe

Figure 1b. Number of overall editors for the article Adobe

Anchoring Statistics

To measure how anchors emerge and change over time WikiGen generates four anchoring

statistics which reflect shifts in the composition and stability of anchors across selected time

periods: New and removed anchors, anchoring dissimilarity, edit-war level, and anchor strength.

New and Removed Anchors. WikiGen displays in a bar chart the number of introduced and

removed anchors (represented by green and red bars respectively) in a chosen period of time.

The chart helps to identify periods with extensive anchoring activity (i.e., periods in which many

anchors were either removed or introduced, or both). Figure 2 below shows new and removed

anchors for the article Adobe.

12
Figure 2: New and removed anchors for the article Adobe.

Anchoring Dissimilarity. This measure captures the extent to which anchors in time period t are

dissimilar to anchors in the previous period t-1. It is based on the number of days an anchor was

present during the selected period and only includes anchors that were present for at least one

day to eliminate the temporary influence of anchors that were introduced to the article due to

vandalism. Because it is sensitive to the difference in the composition of anchors across time

periods, anchor dissimilarity can be used to gauge the difference in the meaning of an article

over time.

The dissimilarity score ranges from 1 (anchors are completely dissimilar across time periods) to

0 (anchors are completely similar across time periods). The logic behind the calculation is to take

the sum of least common days present for all anchors present in periods t and t-1 and set it in

relation to the total sum of maximum days anchors were present for every anchor in periods t and

t-1. Figure 3a shows the formula for anchor dissimilarity and figure 3b shows the anchor

dissimilarity for the article Adobe:

13
Figure 3a: Anchor dissimilarity formula

Figure 3b: Anchor dissimilarity for the article Adobe

Edit-war level. This statistic measures the intensity of disagreement among users by relating the

number of anchor introductions and disappearances to the total number of unique anchors in a

period of time. Edit-war ranges from 0 (no introductions or disappearances of anchors) indicating

no disagreement among contributors, to infinity (x anchors are introduced and removed y times

where y -> inf) indicating intense disagreement among contributors. Figure 4a shows the formula

for edit-war level and figure 4b shows the edit-war level for the article Adobe:

14
Figure 4a: Edit war level formula

Figure 4b: Edit war level for the article Adobe

Anchor Strength. This measure gauges the resilience of individual anchors within a time period.

It does so by linearly combining the number of days an anchor was present and the number of

revisions it survived during that time period. Anchor strength ranges from ‘0’ to ‘1’. Anchors

that are present for long periods or that are quickly reintroduced after being removed will have a

high score. A score of ‘1’ indicates that an anchor has survived all revisions and was present in

the article for the entire measured time period. Table 1 displays the 10 strongest anchors for the

article Adobe in 2014:

15
Table 1: Ten strongest anchors in 2014 for the article Adobe

FINDINGS: THREE CATEGORIES OF DIGITAL TECHNOLOGIES

Below we present the changes in discourse on Wikipedia for three categories of digital

technologies which reflect increasing levels of abstractness: digital devices, digital services, and

digital concepts. Within each category, we describe the discursive patterns of two sample cases

which were selected based on two main criteria: first, they describe a technological phenomenon

whose design has reached stabilization, and; second, the discourse that surrounds them has

sufficient volume (measured by the number of editors and edits) to render our analyses valid.

Digital Devices

iPad

iPad is a line of tablet computers produced by Apple Inc. The article ‘iPad’ was created on

December 26, 2009, a few months before the release of the first generation of the product in

April 2010. Between the creation of the article and the day we conducted the analysis (December

29, 2014), the article was edited 5183 times by 1711 editors.

16
As can be seen in figures 5 and 6, the discursive activity within the article peaked in January

2010 with 1177 edits done by 421 editors. This activity gradually waned over the following

months.

Figure 5. Number of overall edits for the article iPad

Figure 6. Number of overall editors for the article iPad

During this period, the meaning of ‘iPad’ was being shaped as can be seen by the relatively high

number of new and removed anchors, which gradually diminished by the end of 2010 (figure 7).

17
Figure 7. New and removed anchors for the article for the article iPad

The forming of the meaning of ‘iPad’ is also reflected in the anchor dissimilarity measure which

reached its highest level in February 2010 with a score of 0.91, indicating that 91% of the

anchors in present February were different to the anchors present in the preceding month (figure

8).

Figure 8. Anchor dissimilarity for the article iPad

Furthermore, this period was characterized by a high level of disagreement among editors

regarding the meaning of ‘iPad’. This can be observed in the high edit-war level in January 2010,

which indicates a high number of anchor introductions and disappearances relative to the total

number of unique anchors in this month. Similar to anchor dissimilarity, the edit-war level

gradually decreased in the second half of 2010.

18
Figure 9. Edit-war level for the article iPad

Apart for small shifts in the meaning of ‘iPad’ in October and December of 2012 (reflected in the

heightened anchor dissimilarity levels seen in figure 8), we can see that the meaning of ‘iPad’,

and the level of consensus around this meaning, stabilized within the first year of the creation of

the article and the release of the product to market. This can be further observed in the generally

increasing trend of average anchor strength between 2009 and 2014 (table 2). This trend

indicates that anchors in 2014 were, on average, more durable (survived more revisions and were

present for more days) than anchors in each of the preceding years.

Ave.
anchor
Year strength
2009 0.31
2010 0.18
2011 0.69
2012 0.38
2013 0.73
2014 0.87
Table 2. Average anchor strength for the article iPad

19
Wii

Wii is a home gaming console designed by Nintendo. The article ‘Wii’ was created on January 5,

2004, about three years before the release of the product to market, which occurred over

November-December 2006. Between the creation of the article and the day we conducted the

analysis, the article was edited 21670 times by 6998 editors.

As can be seen in figures 10 and 11, the discursive activity within the article peaked in May 2006

2450 edits done by 898 editors. This activity gradually dissipated over the following 18 months.

Figure 10. Number of overall edits for the article Wii

Figure 11. Number of overall editors for the article Wii

20
The bulk of the meaning-shaping activity of ‘Wii’ took place during the initial period of the

article’s existence, indicated by the large number of introduced and removed anchors, which

persisted until April 2007 and then diminished abruptly (figure 12)

Figure 12. New and removed anchors for the article for the article Wii

The changes in the meaning of ‘Wii” can also be seen in the relatively high anchor dissimilarity

scores which went up during 2004, peaked in June 2005, fluctuated during 2006 and the first half

of 2007, and subsequently decreased significantly (except for two brief surges in August 2010

and November 2013).

Figure 13. Anchor dissimilarity for the article Wii

The forming of the meaning of ‘Wii’ are also reflected in the high edit-war level scores between

may 2006 and April 2007.

21
Figure 14. Edit-war level for the article Wii

Apart for small shifts in the meaning of ‘Wii’ in August 2010 and November 2013 (reflected in

see figures 12 and 13), we can see that the meaning of ‘Wii, and the level of consensus around

this meaning, stabilized within a period of about one and a half years (between the beginning of

2006 and mid 2007), and roughly within six months of the release of Wii to the market. This can

also be observed in the generally increasing trend of average anchor strength between 2004 and

2014 (table 3).

Ave.
anchor
Year strength
2004 0.33
2005 0.23
2006 0.18
2007 0.36
2008 0.95
2009 0.71
2010 0.7
2011 0.95
2012 0.73
2013 0.88
2014 0.94
. Average anchor strength for the article Wii

22
Digital Services

Tumblr

Tumblr is a micro-blogging and social networking service that was launched in February 2007.

The Wikipedia ‘Tumblr’ article was created on August 3, 2008. Between the creation of the

article and the day we conducted the analysis, the article was edited 1593 times by 885 editors.

As can be seen in figures 15 and 16, the discursive activity within the article displays a different

pattern to that of the digital devices examined above. Rather than having a clear peak that

gradually diminishes, the discursive activity unfolds through several distinct peaks, which last

over an extended period of time.

Figure 15. Number of overall edits for the article Tumblr

23
Figure 16. Number of overall editors for the article Tumblr

The formation of the meaning of ‘Tumbler’ occurred in a less orderly fashion than did the

meaning of the digital devices. This happened as anchors were removed from and introduced

into the article in several spurts of activity, mostly during 2009 and later in 2011-2012 (figure

17).

Figure 17. New and removed anchors for the article for the article Tumblr

The ongoing change in the meaning of ‘Tumblr’ is also reflected in the multiple peaks in the

anchor dissimilarity score, which are visible around the same time (figure 18).

24
Figure 18. Anchor dissimilarity for the article Tumblr

The same pattern can also be observed in the ongoing fluctuations in the edit-war level which,

like the discursive activity, display multiple peaks that reflect recurring high levels of

disagreement among editors regarding the meaning of ‘Tumblr’ (figure 19).

Figure 19. Edit-war level for the article Tumblr

The uncertainty around the meaning of ‘Tumblr’ is also visible in the average anchor strength.

Rather than displaying a clear upward trend, average scores oscillate during the lifetime of the

article: they increase from 2008 to 2010, decrease in 2011 and 2012, and increase again in 2013

and 2014 (table 4).

25
Ave.
anchor
Year strength
2008 0.21
2009 0.59
2010 0.72
2011 0.54
2012 0.45
2013 0.66
2014 0.82
. Average anchor strength for the article Tumblr

Twitter

Twitter is a micro-blogging and social networking service that was launched in July 2006. The

Wikipedia ‘Twitter’ article was created on February 12, 2007. Between the creation of the article

and the day we conducted the analysis, the article was edited 5519 times by 2662 editors. As can

be seen in figures 20 and 21, the discursive activity within the article displays a different pattern

to that of the digital devices examined above. Rather than having a clear peak that gradually

diminishes, the discursive activity unfolds through two distinct peaks, which persist for over

more than four years, before it diminishes.

Figure 20. Number of overall edits for the article Twitter

26
Figure 21. Number of overall editors for the article Twitter

The formation of the meaning of ‘Twitter’ occurred in a less orderly fashion than did the

meaning of the digital devices. This happened as anchors were removed from and introduced

into the article in several spurts of activity, mostly during the first half of 2007, first half of 2009,

and in 2011 (figure 22).

Figure 22. New and removed anchors for the article for the article Twitter

The meaning of ‘Twitter’ was in flux for much of the lifespan of the article as is evident from the

multiple peaks in the anchor dissimilarity scores since the creation of the article until they

gradually dissipate in 2012 (figure 23).

27
Figure 23. Anchor dissimilarity for the article Twitter

A similar pattern can also be observed in the recurring fluctuations in the edit-war level which

display multiple peaks in the first half of 2008, the first half of 2009, and in the end of 2010.

These peaks reflect high levels of disagreement among editors regarding the meaning of

‘Twitter’ (figure 24).

Figure 24. Edit-war level for the article Twitter

The instability of the meaning of ‘Twitter’ is also visible in the average anchor strength, which

hover under the 0.5 mark from 2007 to 2011, and only start increasing in 2012 (table 5).

28
Ave.
anchor
Year strength
2007 0.25
2008 0.48
2009 0.34
2010 0.36
2011 0.47
2012 0.68
2013 0.75
2014 0.72
. Average anchor strength for the article Twitter

Digital Concepts

Smartphone

The article ‘Smartphone’ was created on January 9, 2003 to describe a class of digital technology

that combines the features of a mobile phone with advanced computing capabilities used as a

platform to run different types of applications.

Between the creation of the article and the day we conducted the analysis, the article was edited

9034 times by 3944 editors. As can be seen in figures 25 and 26, the discursive activity within

the article remains at a high level throughout the lifespan of the article.

29
Figure 25. Number of overall edits for the article Smartphone

Figure 26. Number of overall editors for the article Smartphone

Similarly, the formation of the meaning of ‘Smartphone’ unfolded as anchors were removed

from and introduced into the article in several surges of activity, mostly around the beginning of

2006, mid-2011, and the beginning of 2014 (figure 27).

Figure 27. New and removed anchors for the article for the article Smartphone

30
An examination of the anchor dissimilarity scores reveals that the meaning of ‘smartphone’

continues to fluctuate and in fact is never solidified. This is apparent from the recurring peaks in

anchor dissimilarity scores throughout the entire lifespan of the article (figure 28).

Figure 28. Anchor dissimilarity for the article Smartphone

Furthermore, the level of disagreement among editors remains considerable over time as is

evident in the recurrent spikes in the edit-war level scores (figure 29).

Figure 29. Edit-war level for the article Smartphone

The instability of the meaning of the article is also reflected in table 6, which shows that the

average anchor strength fluctuates over time and is at or under the 0.5 mark for 7 out of the 12

years of the article’s existence, including 2014. This indicates that in the majority of the time,

anchors in the article have been rather ephemeral.

31
Ave.
anchor
Year strength
2003 0.42
2004 0.73
2005 1
2006 0.33
2007 0.48
2008 0.38
2009 0.5
2010 0.56
2011 0.29
2012 0.78
2013 0.86
2014 0.32

. Average anchor strength for the article Twitter

Cloud Computing

The article ‘Cloud computing’ was created on March 3, 2007. Between the creation of the article

and the day we conducted the analysis, the article was edited 9039 times by 3947 editors. As can

be seen in figures 30 and 31, the discursive activity within the article remains at high to moderate

levels throughout the lifespan of the article.

Figure 30. Number of overall edits for the article Cloud computing
32
Figure 31. Number of overall editors for the article Cloud computing

The formation of the meaning of ‘Cloud computing’ unfolded as anchors were removed from

and introduced into the article in several surges of activity, mostly around the mid-2008, the

beginning 0f 2010, the first half of 2011, the beginning of 2012, and mid-2014 (figure 32).

Figure 32. New and removed anchors for the article for the article Cloud computing

An examination of the anchor dissimilarity scores shows that the meaning of ‘Cloud computing’

continues to fluctuate and in fact is never solidified. This is apparent from the recurring peaks in

anchor dissimilarity scores throughout the lifespan of the article (figure 33).

33
Figure 33. Anchor dissimilarity for the article Cloud computing

Furthermore, the level of disagreement among editors remains considerable over time as is

evident in the recurrent spikes in the edit-war level scores (figure 34).

Figure 34. Edit-war level for the article Cloud computing

The instability of the meaning of the article is also reflected in table 7, which shows that the

average anchor strength remains low throughout the entire lifespan of the article. This indicates

that since 2007, anchors in the ‘Cloud computing’ article have been relatively unstable and

replaced frequently.

Ave.
anchor
Year strength
2007 0.24
2008 0.25
2009 0.37
2010 0.32

34
2011 0.15
2012 0.38
2013 0.28
2014 0.55
. Average anchor strength for the article Cloud computing

DISCUSSION

In this paper we have investigated the discourse within the era of incremental change in the

technology lifecycles of six digital technologies that reflect varying levels of abstraction. The

results for the cases within each level were quite consistent with each other. We found that the

volume of discourse around the least abstract digital devices (iPad and Wii) peaked at first, as

measured by the number of edits and number of editors – presumably as individuals were trying

to make sense of the new technology – but then stabilized as the volume of edits and number of

editors tapered off over time. This is consistent with Tushman & Rosenkopf’s (1992) expectation

for social stability during this period. Their expectation was further validated by the increasing

anchor strength measure over the lifespan of the iPad and Wii articles, which indicates that the

discourse around these devices gradually solidified. Additionally, after an initial spike,

disagreement among editors was all but eliminated, as measured by edit wars. Finally, anchor

dissimilarity – our measure for meaning creation – presented its highest values within the first

year following the release of the iPad and Wii to market and maintained punctuated activity

throughout, albeit at low levels. Since this did not coincide with increased conflict, it suggests

that there was a limited generation of new meanings associated with the technologies over time –

supporting the view of an inherent open-endedness view of digital artifacts (Yoo et al 2011;

Kallinikos et al 2013).

35
Our second class of digital technologies – digital services - followed a different pattern. The

volume of the discourse around the Tumblr and Twitter articles had multiple peaks which

covered a much longer time frame (up to four years for the bulk of edits editors for digital

services compared to approximately one year for digital devices). Multiple peaks indicate a much

more punctuated process of discursive sensemaking as different narratives ebb and flow and gain

salience as the meanings of the services are transformed during their adoption by multiple users.

This four-year process of transformation in meanings was further reflected in the anchor

dissimilarity measure, which too maintained relatively high levels for a period of up to four years

before it settled. Similarly, contestation, as measured by anchor wars, also remained high for a

similar period of time before it tapered off.

Importantly, the architecture of both Tumblr and Twitter had already been established at the time

the two articles started gaining momentum and did not change throughout. However the meaning

of the digital services only solidified several years after the technology stabilized. This continued

contestation and discourse evokes an image similar to Dokko et al’s (2012) view of negotiated

order as continued enactment, but it is different in a fundamental way. Negotiated order implies

that social dynamics take place to keep technology standards stable. Our findings suggest that

social dynamics took place to extend the meanings of the technology and generate new

possibilities for its interpretation and use.

Finally, the most abstract digital technologies we studied – digital concepts (cloud computing

and smartphones) – least resembled the period of social stability noted by Tushman & Rosenkopf

(1992). The level of discourse within the two digital concepts articles remained high and even

appeared to increase throughout their lifespan. The contestation and meaning making also

remained high throughout, as reflected in the edit-war and anchor dissimilarity levels, which

36
never tapered off. The continuing instability in the meaning of cloud computing and smartphones

is also evident from the low and fluctuating levels of their respective average anchor strength

measures. The basic architecture of both cloud computing and smartphones emerged rapidly and

was mostly established by the period in question (although incremental innovations within these

architectures remained in later years). Our findings indicate that the continued contestation and

flux in the meanings of the two services reflect their vagueness long after their architecture had

been stabilized. The meaning, nature, possible uses, and business and social significance of the

two concepts remain a subject of ongoing discursive negotiation and debate.

In general, this study joins with the recent flurry of work on digital innovation which find the

innovative dynamics around digital technologies simply behave differently than their analog

predecessors (e.g., Yoo et al 2011; Yoo et al 2012; Kallinikos 2009; Kallinikos et al 2013;

Leonardi 2010). However, we extend this tradition of research to begin unpacking industrial

dynamics associated with technological evolution. Although research on technology and

innovation management has studied the relationship between patterns of industrial evolution to

product architectures (particularly the work on modularity – see Baldwin & Clark 2000; Garud et

al 2009), very little of this work has taken into consideration the changes associated with digital

forms of technology (with some exceptions, see Hobday 1998 and Lee & Berente 2012). In

attending to digital technologies, we answer the call to draw upon the field of information

systems for informing theories of technology and innovation management (Yoo 2013). Further,

of the work on technology lifecycles in the tradition of Anderson and Tushman (1990), there is

little emphasis on the social dynamics that are so important to technology evolution. By focusing

on discourse, we take a step toward bringing this tradition together with the “social shaping” of

technology tradition (e.g. Latour 2000; Bijker 1997; Williams & Edge 1996).

37
CONCLUSIONS

Our findings indicate that, contrary to the assumption made in the technology lifecycle model,

the era of incremental change is not theoretically uninteresting due to a lack of social activity

(Anderson & Tushman 1990). In fact, this phase can be socially dynamic. This dynamism does

not only imply social negotiations in an effort to maintain technology standards stable (Dokko et

al 2012), but also an ongoing discursive process which reflects repeated efforts by multiple

people to interpret and understand the meaning of new technologies.

Importantly, we saw that this process follows a different trajectory depending on the level of

abstractness of the digital technology that people attempted to make sense of. The most concrete

digital devices followed a process that best resembled the original lifecycle model where the

meaning of the device and the discursive activity around it mostly ceased during the era of

incremental change. However, the discursive activity and meaning of more abstract digital

services took much longer to stabilize and the discursive activity and meaning of the most

abstract digital concepts never stabilized during the timeframe of our analysis.

The differences among the three categories indicate that people have an easier time agreeing on

the meaning of technologies which have a distinct physical manifestation and which they can use

and interact with in an unmediated fashion. The lack of a clear physical manifestation and the

absence of a tactile relationship between users and digital services create a certain distance

between the users and the services. This can be seen in the discursive process around the services

which takes longer to stabilize. Whereas digital services reflect recognizable patterns of

processes and relationships between actions and outcomes, digital concepts refer to entire classes

38
of products or services and are therefore even further removed from users. This was evident in

discursive processes wherein the meaning of the examined concepts was in constant flux.

In this study we examined the discourse for six digital technologies. Despite describing distinct

patterns across the three categories of technology, our findings do not thoroughly explain the

significance of the identified patterns. Doing so would require adding a qualitative dimension to

the current investigation to identify not only the contours of the discursive changes, but also their

content. Future research will involve digging more deeply into the differences among the

patterns that we identified.

REFERENCES

Anderson, P., & Tushman, M. L. (1990). Technological discontinuities and dominant designs: A cyclical
model of technological change. Administrative science quarterly, 604-633.
Berente, N., Hansen, S., Pike, J. C., & Bateman, P. J. (2011). Arguing the value of virtual worlds: patterns
of discursive sensemaking of an innovative technology. MIS Quarterly, 35(3), 685-709.
Bijker, W. E. (1997). Of bicycles, bakelites, and bulbs: Toward a theory of sociotechnical change. MIT
press.
Chasin, F., & Gal, U. & Riemer, K. (2013) 'The Genealogy of Knowledge: Introducing a Tool and
Method for Tracing the Social Construction of Knowledge on Wikipedia', in Proceedings of the 24th
Australasian Conference on Information Systems, Melbourne, Australia, 4 - 6 December.
Dokko, G., Nigam, A., & Rosenkopf, L. (2012). Keeping steady as she goes: A negotiated order
perspective on technological evolution. Organization Studies, 33(5-6), 681-703.
Fairclough, N. (1992). Discourse and text: Linguistic and intertextual analysis within discourse analysis.
Discourse & Society, 3(2), 193-217.
Funk, J. (2009). Components, systems and discontinuities: The case of magnetic recording and playback
equipment. Research Policy, 38(7), 1192-1202.
Gal, U. & Berente, N. (2008) “A Social Representations Perspective on Information Systems
Implementation: Rethinking the Concept of ‘Frames’”. Information Technology and People. 21, 2, p.
133-154.
Habermas, J. (1984). The Theory of Communicative Action, Vol. 1,'Reason and the Rationalization of
Society'.
Hansen, S., Berente, N., & Lyytinen, K. (2009). Wikipedia, Critical Social Theory, and the Possibility of
Rational Discourse 1. The Information Society, 25(1), 38-59.
Henderson, R. M., & Clark, K. B. (1990). Architectural innovation: the reconfiguration of existing
product technologies and the failure of established firms. Administrative science quarterly, 9-30.

39
Garud, R., Kumaraswamy, A., & Langlois, R. (Eds.). (2009). Managing in the modular age: architectures,
networks, and organizations. John Wiley & Sons.
Germonprez, M & Hovorka, D. & Gal, U. (2011) “Secondary design: Mid-range theorizing”. Journal of
the Association of Information Systems, 12, 10, p. 662-683.
Griffith, T. L. (1999). Technology features as triggers for sensemaking. Academy of Management review,
24(3), 472-488.
Hansen, S., Berente, N., & Lyytinen, K. (2009). Wikipedia, Critical Social Theory, and the Possibility of
Rational Discourse 1. The Information Society, 25(1), 38-59.
Hepp, M. & Siorpaes, K., & Bachlechner, D. (2007). Harvesting Wiki Consensus: Using Wikipedia
Entries as Vocabulary for Knowledge Management, IEEE Internet Computing, 11, 5, 54–65.
Heracleous, L., & Barrett, M. (2001). Organizational change as discourse: Communicative actions and
deep structures in the context of information technology implementation. Academy of Management
Journal, 44(4), 755-778.
Hobday, M. (1998). Product complexity, innovation and industrial organisation. Research policy, 26(6),
689-710.
Kallinikos, J. (2009). On the computational rendition of reality: Artefacts and human agency.
Organization, 16(2), 183-202.
Kallinikos, J., Aaltonen, A., & Marton, A. (2013). The ambivalent ontology of digital artifacts. Mis
Quarterly, 37(2), 357-370.
Latour, B. (1990). Technology is society made durable. The Sociological Review, 38(S1), 103-131.
Lee, J., & Berente, N. (2012). Digital innovation and the division of innovative labor: digital controls in
the automotive industry. Organization Science, 23(5), 1428-1447.
Lee, J., & Berente, N. (2013). The era of incremental change in the technology innovation life cycle: An
analysis of the automotive emission control industry. Research Policy, 42(8), 1469-1481.
Leonardi, P. M. (2010). Digital materiality? How artifacts without matter, matter. First Monday, 15(6).
Moscovici, S. (2000). The Phenomenon of Social Representations. In S. Moscovici & G. Duveen (Eds.),
Social Representations: Explorations in Social Psychology. Polity Press Cambridge.
Murmann, J. P., & Frenken, K. (2006). Toward a systematic framework for research on dominant designs,
technological innovations, and industrial change. Research Policy, 35(7), 925-952.
Phillips, N., Lawrence, T. B., & Hardy, C. (2004). Discourse and institutions. Academy of management
review, 29(4), 635-652.
Taylor, J. R., & Van Every, E. J. (1999). The emergent organization: Communication as its site and
surface. Routledge.
Tushman, M. L., & Rosenkopf, L. (1992). Organizational determinants of technological-change-toward a
sociology of technological evolution. Research in organizational behavior, 14, 311-347.
Wagner, W., & Valencia, J., & Elejabarrieta, F. (1996). Relevance, discourse and the ‘hot’ stable core
social representations - A structural analysis of word associations, British Journal of Social
Psychology, 35, 3, 331–351.
Williams, R., & Edge, D. (1996). The social shaping of technology. Research policy, 25(6), 865-899.
Wittgenstein, L. (1969). Preliminary Studies for the'Philosophical Investigations'.

40
Yoo, Y., Henfridsson, O., & Lyytinen, K. (2010). Research commentary-The new organizing logic of
digital innovation: An agenda for information systems research. Information Systems Research, 21(4),
724-735.
Yoo, Y., Boland Jr, R. J., Lyytinen, K., & Majchrzak, A. (2012). Organizing for innovation in the
digitized world. Organization Science, 23(5), 1398-1408.
Yoo, Y. (2013). The Tables Have Turned: How Can the Information Systems Field Contribute to
Technology and Innovation Management Research?. Journal of the Association for Information
Systems, 14(5), 227-236.

41

You might also like