generativity in digital ecosystems: how distributed ... · the contemporary digital age sees a...

Generativity in Digital Ecosystems:

How Distributed Networks Organize for Continuous Change

D I S S E R T A T I O N

of the University of St. Gallen,

School of Management,

Economics, Law, Social Sciences

and International Affairs

to obtain the title of

Doctor of Philosophy in Management

submitted by

Alexander Eck

from

Germany

Approved on the application of

Prof. Dr. Walter Brenner

and

Prof. Dr. Jan vom Brocke

Dissertation no. 4763

Digitaldruckhaus GmbH, Konstanz 2018

The University of St. Gallen, School of Management, Economics, Law, Social Sciences

and International Affairs hereby consents to the printing of the present dissertation,

without hereby expressing any opinion on the views herein expressed.

St. Gallen, May 22, 2018

The President:

Prof. Dr. Thomas Bieger

Acknowledgements I

Acknowledgements

This thesis is the result of a journey that started in early 2013. Between then and now,

many people contributed to making this thesis a reality. First and foremost, I thank Prof.

Dr. Walter Brenner for creating a unique work environment that is in touch with

practitioners, yet fosters the creation of independent and creative scholarly work. I am

truly grateful to Prof. Brenner for letting me become part of his team at University of

St. Gallen and for the numerous discussions that were crucial to form this thesis.

Furthermore, I thank Prof. Dr. Jan vom Brocke from University of Liechtenstein for his

willingness to co-supervise my thesis and for giving vital advice in critical times. My

thanks also go to Prof. Dr. Falk Uebernickel and Dr. Jochen Wulf, who were always

willing to review my research projects and reliably identified areas to improve them.

Organizational advice and support was impeccable, for which I would like to thank

Barbara Rohner, Dr. Jochen Müller, Dr. Peter Gut and Susanne Gmünder. Special

appreciation goes to Markus Handke and Sandro Storchenegger for providing superb IT

infrastructure. My first two years in St. Gallen were marked by close collaboration with

practitioners at UBS AG, and I thank Stefan Keidel, Dr. Thomas Schneider and all others

for their support and leadership. In this context I would also like to thank Prof.

Uebernickel for masterfully managing the project setting.

A large part of the overall experience was shaped by my fellow colleagues. Without

neglecting the others, I am deeply grateful to Benjamin Spottke and Matthias Herterich

for countless discussions, reviews, brilliant ideas and suggestions, most of which found

their way into this thesis and even proved useful beyond that. A highlight of my PhD

experience was the 10-weeks research stay at Warwick Business School in early 2016.

I thank Prof. Dr. Ola Henfridsson for hosting this visit and giving invaluable advice,

Philipp Hukal for the many debates on generativity and taking initiative in many more

scholarly and organizational matters than I could have hoped for, Dr. Aleksi Aaltonen

for discussing graph theory and Python with me, and all staff and fellow PhD students

for welcoming me into the WBS family.

A special thank you goes to my family for their encouragement throughout the years

and of course to Dr. Olga Willner, who helped and supported me in more ways than I

could ever express with words.

Berlin, January 2018

Alexander Eck

Table of contents III

Table of contents

Acknowledgements ........................................................................................................ i

Table of contents .......................................................................................................... iii

List of abbreviations ................................................................................................... vii

List of figures ................................................................................................................ ix

List of tables .................................................................................................................. xi

Abstract ....................................................................................................................... xiii

Kurzfassung ................................................................................................................. xv

Part A ............................................................................................................................. 1

1 Introduction ............................................................................................................. 1

1.1 Research motivation ......................................................................................... 1

1.2 Research objective and research questions ...................................................... 2

1.3 Research design ................................................................................................ 4

1.4 Thesis structure ................................................................................................ 5

2 Research results ...................................................................................................... 7

2.1 On the generativity term – Article I ................................................................. 7

2.2 Generative properties and generative patterns – Article II .............................. 9

2.3 Detection of open source software ecosystems – Article III ......................... 11

2.4 Cross-community coordination as means of generative system change –

Article IV ....................................................................................................... 13

3 Discussion and future research ............................................................................ 17

3.1 Theoretical contributions ............................................................................... 17

3.2 Practical implications ..................................................................................... 18

3.3 Limitations and future research ..................................................................... 19

4 Reference overview of articles in this thesis ....................................................... 21

4.1 The generative capacity of digital artifacts: a mapping of the field – Article I

........................................................................................................................ 21

4.2 Untangling generativity: two perspectives on unanticipated change produced

by diverse actors – Article II .......................................................................... 22

IV Table of contents

4.3 Reconstructing open source software ecosystems: finding structure in digital

traces – Article III .......................................................................................... 23

4.4 Coordination across open source software communities: findings from the

Rails ecosystem – Article IV ......................................................................... 24

Part B ........................................................................................................................... 25

I The generative capacity of digital artifacts: a mapping of the field ................ 25

I.1 Introduction .................................................................................................... 26

I.2 Generativity in the context of digital artifacts ............................................... 27

I.3 Research method ............................................................................................ 31

I.4 The many meanings of generativity in IS research ........................................ 33

I.5 The generative capacity of digital artifacts .................................................... 36

I.5.1 Approaching the scholarly discourses related to generativity from two

different angles ................................................................................... 36

I.5.2 Digital infrastructures ......................................................................... 38

I.5.3 Digital platforms ................................................................................. 40

I.5.4 Digital artifacts, innovation, and modularity ...................................... 42

I.6 Conclusion and further research .................................................................... 44

II Untangling generativity: two perspectives on unanticipated change produced

by diverse actors .................................................................................................... 47

II.1 Introduction .................................................................................................... 48

II.2 Two perspectives on generativity .................................................................. 49

II.2.1 Generative properties: generativity as consequence of system design

............................................................................................................ 50

II.2.2 Generative patterns: generativity as consequence of system evolution

............................................................................................................ 52

II.3 Generative properties and generative patterns in literature ........................... 53

II.4 Generativity in digital platform ecosystems .................................................. 56

II.4.1 Modeling a digital platform ecosystem .............................................. 56

II.4.2 Generative properties in digital platform ecosystems ........................ 58

II.4.3 Generative patterns in digital platform ecosystems ............................ 60

Table of contents V

II.5 The promise of generativity for explaining digital innovation ...................... 63

II.6 Conclusion ..................................................................................................... 66

II.7 Appendix: descriptions of generativity in extant literature ........................... 68

III Reconstructing open source software ecosystems: finding structure in digital

traces ...................................................................................................................... 71

III.1 Introduction .................................................................................................... 72

III.2 Distributed innovation in open source software ecosystems ......................... 73

III.2.1 Open source software ecosystems ...................................................... 73

III.2.2 Distributed innovation ........................................................................ 75

III.3 Digital trace data as new kind of data source ................................................ 76

III.3.1 Digital trace data in information systems research ............................. 76

III.3.2 Implication for IS: in need of ‘exploratory data loops’ ...................... 78

III.4 Reconstructing open source software ecosystems from digital traces ........... 79

III.5 From data hairball to surprising pattern ......................................................... 81

III.5.1 Exploratory data loop I: assessing data quality .................................. 82

III.5.2 Exploratory data loop II: informing sample selection ........................ 83

III.5.3 Exploratory data loop III: detecting intriguing regularities ................ 84

III.6 Instead of a conclusion: plans for further research ........................................ 86

IV Coordination across open source software communities: findings from the

Rails ecosystem ...................................................................................................... 89

IV.1 Introduction .................................................................................................... 90

IV.2 Background .................................................................................................... 91

IV.2.1 Coordination, coordination episodes, and coordination mechanisms 91

IV.2.2 Coordination mechanisms across communities of open source

software ............................................................................................... 91

IV.3 Research design and methods ........................................................................ 92

IV.3.1 Data collection .................................................................................... 93

IV.3.2 Construct operationalization ............................................................... 95

IV.3.3 Data analysis ....................................................................................... 98

IV.4 Findings: mechanisms of cross-community coordination ............................. 98

VI Table of contents

IV.4.1 The adaptation mechanism ................................................................. 99

IV.4.2 The upgrading mechanism ................................................................ 100

IV.4.3 The positioning mechanism .............................................................. 100

IV.4.4 The departure mechanism ................................................................. 101

IV.5 Discussion: coordination as means of generative change ............................ 101

IV.6 Limitations and conclusion .......................................................................... 102

References .................................................................................................................. 103

Publication list of the author .................................................................................... xvii

Curriculum vitae ........................................................................................................ xix

List of abbreviations VII

List of abbreviations

AMCIS Americas Conference on Information Systems

API Application programming interface

ARF Attractive and repulsive forces

DESRIST Design Science Research in Information Systems and Technology

Dipl.-Ing. Diplom-Ingenieur

ECIS European Conference on Information Systems

ERP Enterprise Resource Planning

FTP File transfer protocol

IBM International Business Machines Corporation

ICIS International Conference on Information Systems

ICT Information and communication technologies

IP Internet protocol

IS Information systems

IT Information technology

IWI-HSG Institute of Information Management, University of St. Gallen

MIS Management Information Systems

MKWI Multikonferenz Wirtschaftsinformatik

OSS Open source software

PACIS Pacific Asia Conference on Information Systems

PhD Doctor of Philosophy

RQ Research question

WBS Warwick Business School

XML Extensible Markup Language

List of figures IX

List of figures

Figure 1. Delineating the intended target discourse by tracing definitions and context

....................................................................................................................... 9

Figure 2. Visualization of Github projects (nodes) with their identified interrelations

(edges) ......................................................................................................... 12

Figure 3. Exemplary screenshot of a discussion board on Github ............................. 14

Figure 4. Number of sources (y-axis) referenced in at least n papers (x-axis) ........... 32

Figure 5. The 14 sample papers and the 22 sources which were referenced at least 4

times ............................................................................................................ 33

Figure 6. Generativity as consequence of system design, and as consequence of

system evolution .......................................................................................... 50

Figure 7. Socio-technical model of a digital platform ecosystem .............................. 57

Figure 8. A macro-micro-macro model of digital platform ecosystem change .......... 62

Figure 9. Generic process to harvest and analyze digital trace data ........................... 77

Figure 10. Artifact dependencies and number of episodes per artifact pair ................. 97

List of tables XI

List of tables

Table 1. Summary of research methods applied in Articles I-IV ................................ 4

Table 2. Overview of articles in this thesis and their contributions ............................ 7

Table 3. Differences of properties and patterns perspective along research

dimensions (indicative) ............................................................................... 10

Table 4. Identified mechanisms of cross-community coordination (Eck 2018) ....... 15

Table 5. Bibliographic information for Article I ....................................................... 21

Table 6. Bibliographic information for Article II ...................................................... 22

Table 7. Bibliographic information for Article III .................................................... 23

Table 8. Bibliographic information for Article IV .................................................... 24

Table 9. Bibliographic information for Article I ....................................................... 25

Table 10. Illustration of links between digital artifacts attributes and generativity

attributes ...................................................................................................... 30

Table 11. Literature search results ............................................................................... 32

Table 12. Same term, different things: meanings of generativity in identified articles

..................................................................................................................... 34

Table 13. Types of digital artifacts and themes discussed in the literature review

sample .......................................................................................................... 36

Table 14. Scholarly discourses commonly discussed in conjunction with generativity

..................................................................................................................... 38

Table 15. Bibliographic information for Article II ...................................................... 47

Table 16. Descriptions of generativity in IS literature ................................................ 54

Table 17. Descriptions of generativity as consequence of system design................... 68

Table 18. Descriptions of generativity as consequence of system evolution .............. 69

Table 19. Bibliographic information for Article III .................................................... 71

Table 20. Ecosystems ‘rails/rails’ and ‘minecraftforge/minecraftforge’ .................... 82

Table 21. Test results for small-world and scale-free structure .................................. 84

Table 22. Bibliographic information for Article IV .................................................... 89

Table 23. Artifacts included in the case study ............................................................. 94

Table 24. Exclusion criteria and number of excluded episodes .................................. 95

Table 25. Operationalization of constructs and number of data points per construct . 96

Table 26. Cross-community coordination mechanisms .............................................. 99

Table 27. Comprehensive publication list with participation of the author .............. xvii

Abstract XIII

Abstract

The contemporary digital age sees a surge in digital ecosystems, which are productive

networks of interdependent digital artifacts and social actors. Long-lived and innovative

digital ecosystems such as the internet, Linux, and other ecosystems based on open

source software showcase an organizing logic that exploits the unbounded and malleable

nature of digital technology. They are highly generative, meaning they create

unanticipated and continuous changes driven by diverse groups distributed throughout

the ecosystem. While the existence of generative ecosystems is described in prior

literature, it is not well-understood how to detect them, nor how they organize to remain

open to changes yet sufficiently consistent necessary to retain their participating actors.

This thesis, consisting of four individual articles, investigates generativity in digital

ecosystems, both conceptually and empirically. A systematic review of the knowledge

base summarizes the state-of-the-art in generativity research, focusing on platform-

based and infrastructure-based digital ecosystems. It further contributes to research by

identifying the many scholarly discourses to which the generativity term has been

attached. Conceptually, this thesis develops two complementary perspectives on

generativity and illustrates them on the example of platform-based digital ecosystems.

While the ‘generative properties’ lens asks for artifact and governance designs that

induce generativity, the ‘generative patterns’ lens is concerned with explaining how

digital ecosystems evolve in unanticipated ways.

As generative digital ecosystems are not created by a grand designer, they are difficult

to identify. This thesis presents a computational approach to identify digital ecosystems,

drawing from digital trace data collected on Github, a popular service for hosting open

source software projects. This approach is applied to a dataset of 34.4m open source

software projects, which yields over 20,000 different digital ecosystems, large and

small. While different topologies exist, most large digital ecosystems are platform-

based. Coordination arrangements in one such ecosystem are analyzed in more detail

via an explanatory multiple case study. Four mechanisms are identified that allow

continuous changes in the various parts of the ecosystem yet ensure consistency.

This thesis contributes to theory by (1) providing two perspectives on generativity in

digital ecosystems, (2) showing how to identify digital ecosystems composed of open

source software, and (3) explaining how open source software communities coordinate

across project boundaries. For practitioners, this thesis holds managerial lessons for how

to organize in the contemporary digital age.

Kurzfassung XV

Kurzfassung

Die Bedeutung digitaler Ökosysteme nimmt zu. Diese sind produktive Netzwerke von

voneinander abhängigen digitalen Artefakten und sozialen Akteuren. Langlebige

Ökosysteme wie das Internet, Linux und andere folgen einer Organisationslogik, welche

sich den grenzenlos formbaren Charakter digitaler Technologie zunutze macht. Sie sind

generativ, das heißt sie erzeugen unerwartete und kontinuierliche Veränderungen, die

von unterschiedlichsten Akteuren im Ökosystem beigesteuert werden. Während die

Existenz generativer Ökosysteme in der Literatur beschrieben wird, ist unklar, wie diese

zu erkennen sind und wie sie sich organisieren, um einerseits offen für Veränderung zu

bleiben, andererseits aber beständig genug sind, um ihre beteiligten Akteure zu erhalten.

Diese aus vier Einzelartikeln bestehende Dissertation untersucht Generativität in

digitalen Ökosystemen. Ein Überblick fasst den Stand der Forschung zusammen,

insbesondere mit Blick auf Generativität in Plattform-basierten und Infrastruktur-

basierten digitalen Ökosystemen. Konzeptionell entwickelt diese Arbeit zwei sich

ergänzende Perspektiven auf Generativität und veranschaulicht diese am Beispiel

Plattform-basierter digitaler Ökosysteme. Während die „generative Eigenschaften“-

Linse nach Artefakt- und Steuerungs-Designs fragt, die Generativität induzieren, befasst

sich die „generative Muster“-Linse mit der Erklärung, wie sich digitale Ökosysteme auf

unerwartete Weise entwickeln.

Da generative digitale Ökosysteme nicht von einem allwissenden Designer erstellt

werden, sind sie schwer zu erkennen. Es wird ein computergestützter Ansatz zur

Identifizierung digitaler Ökosysteme vorgestellt, der Daten von Github nutzt, einem

beliebten Service für das Bereitstellen von Open-Source-Software-Projekten.

Angewendet auf einen Datensatz von 34,4 Millionen Projekten werden über 20.000

verschiedene digitale Ökosysteme identifiziert. Während verschiedene Topologien

existieren, sind die meisten großen Ökosysteme Plattform-basiert. Mechanismen zur

Koordination in einem solchen Ökosystem werden anhand einer erklärenden Fallstudie

untersucht. Vier Mechanismen werden identifiziert, die Veränderungen in den

verschiedenen Teilen des Ökosystems ermöglichen und Kontinuität gewährleisten.

Diese Dissertation trägt zur Theorie bei, indem (1) zwei Perspektiven auf Generativität

in digitalen Ökosystemen entwickelt werden, (2) ein Ansatz zur Identifikation von

Open-Source-Software-Ökosystemen vorgestellt und (3) erklärt wird, wie sich Open-

Source-Software-Communities über Projektgrenzen hinweg koordinieren. Für Praktiker

enthält diese Arbeit Hinweise zur Steuerung im digitalen Zeitalter.

Part A: Introduction 1

Part A

1 Introduction

1.1 Research motivation

November 12, 2014 marked a seemingly small, yet remarkable turning point for

Microsoft, one of the world’s largest software providers. On that day, the company made

.Net Core available as open source software (Landwerth 2014). Comparable to the Java

technology stack, .Net Core is a key technology for writing and running software, which

Microsoft held under close protection until then. Moving to open source software was

motivated by the goal to “build and leverage a stronger ecosystem” (ibid.) and the

prospect that external developers would contribute to continuous evolution and

extension of the technology base. This was a stark departure from the Microsoft of old,

which rallied against “communist” open source software and warned that “developing

software requires leadership” (The Economist 2003). Today, open source software

projects that Microsoft curates are among the most active ones, attracting thousands of

contributors (GitHub 2017d), and, despite earlier warnings of the perils of open source,

the company’s stock market performance is convincing (The Economist 2017).1

As this example illustrates, there seems to be tremendous value in opening up digital

artifacts2, letting others extend them with complementary artifacts and find novel use

contexts (Yoo 2013). Over time, these activities create a productive network of

interdependent digital artifacts and social actors, or digital ecosystems (cf. Selander et

al. 2013). Many examples of successful digital ecosystems come to mind, with the app

economy, the internet and Linux being some of the best-known ones. In short, it seems

that digital ecosystems provide a viable organizing logic, or a rational frame to organize

(Sambamurthy and Zmud 2000), in the contemporary digital age.

A way to make sense of digital ecosystems is in terms of their generativity, or their

capability to create unanticipated and continuous changes driven by diverse groups

without central oversight (Zittrain 2008). In general, the argument goes, a more

1 Stock market performance is not necessarily a result of Microsoft’s move to open source software. It is mentioned here to say that there is nothing particularly ‘communist’ about the open source software model. 2 Throughout this thesis, digital artifact refers exclusively to software programs and not to other types of digital artifacts such as digital music.

2 Part A: Introduction

generative digital ecosystem will have a tendency to create more digital innovations, or

more novel combinations of digital artifacts (Yoo et al. 2010), because it creates more

opportunities for diverse social actors to interact freely and tinker with digital artifacts

created in other parts of the ecosystem, which in turn fuels further opportunities for

change and recombination (Arthur 2009; Boland et al. 2007; Yoo 2013).

The argument is far from complete, however (cf. Reuver et al. 2017). For example, it is

unclear whether a digital ecosystem can be generative (an inherent property) or become

generative (a path-dependent trajectory). Moreover, digital ecosystems are notoriously

difficult to delineate, because they are unbounded and their digital fabric is continuously

in the making (cf. Garud et al. 2008). Finally, generative processes in digital ecosystems

regularly need to resolve the tension of allowing independence on actor level despite

being entangled in a web of socio-technical dependencies (Tilson et al. 2010). This

thesis is motivated by the ambition to explore these pertinent topics. Thus, it contributes

to the knowledge on generativity in digital ecosystems as a way to organize for constant

change (and in extension innovation) in the digital age.

1.2 Research objective and research questions

The following overarching research objective frames this thesis:

This research aims to describe and explain generativity in digital ecosystems, or

the capability of distributed networks to organize for continuous change.

In elaborating the thesis, I tackled the objective both conceptually and empirically.

Conceptually, generativity in digital ecosystems needs to be expounded in considerably

more depth than the information systems (IS) literature has been able to provide.

Although generativity is regarded to be potentially as influential to the digital age as

modularity has been to the industrial age (Yoo 2013), extant literature treats it as “a nice

shorthand label for a crucial and complicated thought” (Post 2010: 2758), albeit

without fully explicating the thought, at least not in IS research. Moreover, the

dictionary definition of generativity as the ability to give rise to something (Oxford

English Dictionary 2009) is sufficiently broad to be applied to many seemingly related,

yet widely distinct theoretical concepts. A consistent vocabulary is missing that is

capable to describe in sufficient detail and from a socio-technical account how diverse

groups create unanticipated changes in digital ecosystems without central oversight.

Therefore, research question 1 (RQ.1) is formulated as:

RQ.1: How can generativity in socio-technical digital ecosystems be described?


Generativity in digital ecosystems also poses interesting empirical challenges. For one,

there are methodological issues on data collection and analysis. It is difficult to

distinguish a digital ecosystem of nontrivial complexity from another one, because their

boundaries are not explicitly defined by a grand designer that supervises ecosystem

creation and evolution (Hanseth and Lyytinen 2010). Moreover, depending on framing,

any one element may be part of multiple digital ecosystems, as it maintains multiple

interdependencies (Reuver et al. 2017). Existing accounts of generativity in digital

ecosystems sidestep such issues by defining ecosystem boundaries a priori with

idiosyncratic selection criteria (e.g., Selander et al. 2013; Um et al. 2015). However, to

enable comparison of structure and elements of different digital ecosystems, and thus

set the foundation for possibly generalizable empirical findings (cf. Reuver et al. 2017),

an automated approach of digital ecosystem detection is required that can discern

individual digital ecosystems out of a large dataset based on a set of common rules and

assumptions (cf. Blincoe et al. 2015). For this thesis, I chose the empirical context of

open source software to study generativity in digital ecosystems. Accordingly, RQ.2 is

formulated as:

RQ.2: How can individual digital ecosystems be detected out of a large sample

of open source software projects?

A profound conceptualization of generativity (RQ.1) and a robust approach to detect

digital ecosystems in the context of open source software (RQ.2) enable an empirical

investigation of how diverse actors create continuous changes in a distributed socio-

technical network composed of interdependent elements. As scholars have noted,

generativity in a digital ecosystem paradoxically depends on stable social arrangements

that succeed in coherently integrating changes in its various parts; otherwise, continued

changes in a distributed network would gradually lead to its demise (Tilson et al. 2010).

In other words, coordination mechanisms (Okhuysen and Bechky 2009) help organize

generativity in a digital ecosystem. It is reasonable to assume that mature, long-lived

digital ecosystems which are sufficiently distributed in knowledge and control have

developed sustainable arrangements to coordinate that are supportive of generative

change. Identifying such arrangements thus leads to a better explanation of generativity

in digital ecosystems. As I conducted the empirical part of this thesis in the context of

open source software, RQ.3 is formulated as:

RQ.3: How do actors coordinate changes to digital artifacts in a mature,

distributed digital ecosystem of interdependent open source software projects?


1.3 Research design

This research is guided by a critical realist ontology (Archer 1998), assuming that

generativity in digital ecosystems has an objectively existing deep structure. At the same

time, it acknowledges that our knowledge of the world is socially constructed and thus

limited to what can be perceived and measured (Mingers et al. 2013). This translates to

a research design (cf. Zachariadis et al. 2013) that starts with a delineation of the

research interest, namely generativity in the context of digital ecosystems (RQ.1);

continues by proposing a way to identify a subset of the empirical domain which is

relevant for the research objective, namely digital ecosystems of open source software

projects (RQ.2); and drills down to a particular socio-technical practice that advances

our understanding of the phenomenon, by identifying how changes to digital artifacts

are coordinated in a specific digital ecosystem (RQ.3). Consequently, I employed a

varied methodological apparatus, designed to meet the specifics of the individual

research questions. Table 1 gives a summary of the research methods and data sources.

In the following, I explain the rationale behind the choices.

Table 1. Summary of research methods applied in Articles I-IV

RQ Article Research method(s) Data source(s)

RQ.1

I Systematic literature review 52 papers with references to generativity term

II Development of descriptive theory Zittrain’s (2006, 2008) definition of generativity

RQ.2 III Network analysis Digital trace data extracted from 34.4m open source software projects hosted on Github

RQ.3 IV Explanatory multiple case study 96 coordination episodes among 5 open source software projects that are part of the same digital ecosystem

RQ.1 asks about a socio-technical description of generativity in digital ecosystems. Due

to its versatility, the label generativity may be attached to many distinct theoretical

concepts. Hence, I designed Article I to elucidate those different concepts by means of

systematic, keyword-driven literature review of representative papers. The article then

continues with analyzing the shared references of those 14 papers that use the term in

the sense of Zittrain (2006, 2008). With Article II, I build on these results to develop a

descriptive theory of generativity in digital ecosystems. By dissecting the influential

definition of Zittrain of generativity as “a system’s capacity to produce unanticipated

change through unfiltered contributions from broad and varied audiences” (Zittrain


2008: 70), my aim was to describe generativity in digital ecosystems from two

alternative, yet complementary perspectives.

RQ.2 probes into ways to identify digital ecosystems that consist of open source

software projects. Because open source software projects come into existence and

change in the public domain, they produce large quantities of digital trace data (cf.

Howison et al. 2011) that are potentially useful for this research project. In Article III, I

harness digital trace data by applying a detection technique originally proposed by

Blincoe et al. (2015).3 This approach is capable to computationally discern digital

ecosystems out of a large dataset of open source software projects, with data traces

captured from Github, a popular service for hosting open source software projects. I then

subject a subset of the identified digital ecosystems to complementary means of network

analysis (cf. Watts 2007) informed by the research interest of this thesis.

Finally, RQ.3 aims at the practice of coordination in digital ecosystems composed of

open source software projects. In Article IV, I apply case study methodology to elucidate

stable arrangements to coordinate across five interdependent open source software

projects. For selecting the cases I relied on the results presented in Article III, in which

I identified the digital ecosystem around Ruby on Rails, an artifact that facilitates

development of web-based applications, as an empirical setting suitable for the research

project. I carried out data analysis in a sense-making process that Langley (1999)

describes as quantification strategy: taking rich process data of 96 episodes of

coordination (cf. Annabi et al. 2008), systematically reducing the data corpus to theory-

guided characteristics, and identifying stable coordination arrangements.

Additional explanations of how research was conducted are provided in the respective

methods sections of the individual articles, which are assembled in Part B of this thesis.

1.4 Thesis structure

I elaborated this thesis in agreement with the requirements and guidelines for cumulative

dissertations at the University of St. Gallen. It is composed of two parts: Part A provides

an overview of the research topic, motivates the research questions, and summarizes key

3 The approach employed in Article III makes adjustments to Blincoe et al. (2015) in that a considerably more extensive dataset is sourced and a different algorithm is used to discern digital ecosystems.


research results. Part B is the core of this thesis, which contains four individual articles

that jointly address the research objective.

Part A is organized in four sections. Section 1 describes the motivation for this research

project, formulates the research objective, derives the research questions, and explains

the overall research design. Section 2 presents major research results and clarifies the

contributions of the four individual articles regarding the overall research project.

Section 3 discusses theoretical and practical implications, followed by suggestions for

future research derived from the inherent limitations of this research endeavor. Section 4

gives a reference overview of the four articles included in this thesis.

Part B consists of four self-contained articles that jointly address the research questions

in detail. At time of writing this thesis, Articles I-III had been published, and Article IV

had been accepted for future publication. All four articles went through double-blind

review processes. As the articles were submitted to different scholarly outlets, they vary

in their individual structure and level of detail. To keep presentation consistent, I

reformatted the articles for inclusion in this thesis, with uniform font sizes and styling

for tables, as well as continuous numbering of sections, figures and tables. I also

improved readability of figures if possible. Apart from correcting a few orthographic

and grammatical errors, content of the articles remained unchanged. All references (for

Part A and Part B) were consolidated at the end of this thesis.

Part A: Research results 7

2 Research results

This thesis consists of four individual articles that address the research questions

introduced in Section 1.2, as shown in Table 2. Two articles are dedicated to describing

the generativity concept (RQ.1). Article I, by means of two linked literature reviews,

systematically summarizes the knowledge base on generativity in information systems

research. Based on this foundation, Article II introduces two views on how large and

varied audiences produce change in digital ecosystems – the generative properties and

generative patterns perspective, respectively. Article III presents a method to discern

individual digital ecosystems within a larger population of open source software

projects, thus addressing RQ.2. Finally, Article IV probes into one of the detected digital

ecosystems to explain how large and varied audiences coordinate to introduce change

(RQ.3). As the individual articles are self-contained, it is worth contextualizing them

with regard to the overall research objective of this thesis, which follows next.

Table 2. Overview of articles in this thesis and their contributions

RQ Article Title (outlet) Contribution to thesis

RQ.1

I The Generative Capacity of Digital Artifacts: A Mapping of the Field (PACIS 2015 Proceedings)

On the generativity term

II Untangling Generativity: Two Perspectives on Unanticipated Change Produced by Diverse Actors (ECIS 2016 Proceedings)

Generative properties and generative patterns

RQ.2 III Reconstructing Open Source Software Ecosystems: Finding Structure in Digital Traces (ICIS 2016 Proceedings)

Detection of open source software ecosystems

RQ.3 IV Coordination Across Open-Source Software Communities: Findings from the Rails Ecosystem (MKWI 2018 Proceedings)

Cross-community coordination as means of generative system change

2.1 On the generativity term – Article I

Digital artifacts are infinitely malleable and thus never complete (Garud et al. 2008).

Unlike manufactured products, which advance from design to production to use fairly

linear, digital artifacts carry the option to be changed and recombined at any time, given

one gets access to their programmatic foundation (Yoo et al. 2010). Organizations that

know how to exploit such options gain a competitive advantage in terms of their

innovativeness (Woodard et al. 2013). Often, outsiders identify design change and

8 Part A: Research results

recombination potentials – the app ecosystem comes to mind – which favors a more

open approach to sharing artifact designs and a more networked organizing logic than

what was considered efficient in the past (Svahn and Henfridsson 2012).

Realizing this new organizing logic, in a paper from 2006 and a book published in 2008,

Johnathan Zittrain proposed to judge complex socio-technical systems by their

generativity, or their ability to

“produce unanticipated change through unfiltered contributions from broad

and varied audiences” (Zittrain 2008: 70).

Zittrain offered this definition from the standpoint of a law scholar who was concerned

of losing the distributed and open fabric of Internet-based innovations to short-term

interests of companies and governments (Post 2010; Zittrain 2009). Information systems

scholars soon adopted the concept of generativity to offer a nascent explanation of the

organizing logic in a fully digitized society (Tilson et al. 2010; Yoo et al. 2010). Yoo

(2013) even suggested the idea of generativity might become as central to our

understanding of the digital age as the idea of modularity had become to making sense

the industrial age (cf. Simon 1996).

Despite its expected potential for advancing IS research, generativity theory still

originates from an outside scholarly community. Applying a ‘foreign’ theory to an IS

problem comes with all sorts of pitfalls (Truex et al. 2006), one of which is ambiguity

of terms: the same term might mean different things to different people and different

terms might be attached to the same phenomenon. With Article I, I clarify the

terminology of generativity systematically via two linked literature reviews, as

illustrated in Figure 1.

The first review catalogues the various definitions of the generativity term that IS

researchers had applied over time, extracted from analyzing the basket of eight journals

(AIS 2011). Overall, I identified 18 different meanings, including Zittrain’s definition,

but also covering such diverse scholarly discourses as generative mechanisms in critical

realism (e.g., Bhaskar 1998 applied by Henfridsson and Bygstad 2013); generative

processes in organizational learning (e.g., Senge 1992 applied by Lambert and Peppard

1993); generative grammar in linguistics (e.g., Chomsky 1957 applied by Lyytinen

1985); and generative relationships in complexity theory (e.g., Lane and Maxfield 1996

applied by McBride 2005). Such heterogeneity is not surprising, given the broad

dictionary definition of the term: generativity refers to the ability or function of


producing, either in the productive sense – as in generating – or the reproductive sense

– as in generations (Oxford English Dictionary 2009).

Figure 1. Delineating the intended target discourse by tracing definitions and context

The second review traces the context to which IS researchers had introduced Zittrain’s

generativity concept. To this end, I conducted a citation network analysis on the subset

of 14 identified papers which referenced Zittrain (2006, 2008). This analysis shows that

generativity is commonly discussed in the context of complex socio-technical systems

in the form of platform-based (e.g., Tiwana et al. 2010) and infrastructure-based digital

ecosystems (e.g., Hanseth and Lyytinen 2010). Two main lines of argument are

commonly advanced. First, a system which is perceptive for contributions of broad

audiences – that is a generative system – is thought to be conducive to innovative

outcomes (e.g., Boland et al. 2007). Second, generative systems are thought to strike a

sustainable balance between being open to distributed changes, while remaining

sufficiently consistent and reliable to attract and retain a large number of participating

actors (e.g., Tilson et al. 2010).

With clarifying the generativity term and tracing its context in IS research, Article I lays

the foundation for the rest of the thesis. Crucially, it shows the importance of

investigating not simply the malleability of single digital artifacts in isolation, but rather

how a network of interdependent actors organizes change processes that strike a balance

between openness and stability.

2.2 Generative properties and generative patterns – Article II

In the course of summarizing the knowledge base on generativity in IS research, a major

conceptual inconsistency became apparent. Extant literature does not distinguish

between generativity as consequence of system design and generativity as consequence

of system evolution. In other words: can a system be generative, or does it become


generative? The first perspective would call for the identification of artifact and

governance design elements that instill generative capacity. The second perspective

would ask to trace change processes that lead to generative evolution over time. Both

viewpoints have their merit in studying generativity in complex socio-technical systems,

they are even complementary. Yet, when it comes to research goals, objects, approaches

and expected results, they are starkly different, as Table 3 indicatively illustrates.

Table 3. Differences of properties and patterns perspective along research dimensions (indicative)

Generative properties perspective Generative patterns perspective

Goal Artifact designs and governance arrangements with generative capacity

Contingent explanations for generative ecosystem evolution

Object Digital artifacts and ecosystem governance Emergent change in socio-technical systems

Approach Analyze artifact designs and governance in mature and distributed ecosystems

Observe micro-level behavior of ecosystem actors and resolution of tensions over time

Results Static models; design knowledge Dynamic models; ethnographic studies

Examples Wareham et al. 2014 Eaton et al. 2015; Um and Yoo 2016

With Article II, I describe these two perspectives on three levels. First, by dissecting

Zittrain’s generativity definition I make the point that it carries two perspectives on

generativity. The first one, labeled generative properties, regards generativity as

“a system’s capacity to churn out a variety of novel products or to be put to a

multitude of surprising uses” (Eck and Uebernickel 2016b: 3).

The second perspective, labeled generative patterns, understands generativity as

“how and to which extent a system develops elements, structures and behaviors

that surpass the imagination or ambition of its original designers” (Eck and

Uebernickel 2016b: 4).

Next, I analyzed a collection of 20 papers – selected from the basket of eight journals

(AIS 2011) plus ICIS and ECIS conference proceedings – for the generativity

perspective they assumed. Out of these, 10 papers regard generativity as consequence

of system design, 5 papers see generativity as consequence of system evolution, and 5

papers carry traces of both perspectives in them, usually leaning more towards one of

the two perspectives. For example, while Yoo et al. (2012) mainly discuss how some

companies achieve building “generative platforms” (ibid.: 1401) that harness the

creativity and innovativeness of third parties, they also state that “digital technology

become inherently malleable and dynamic; […] pervasive digital technology creates


generative innovation” (ibid.: 1399). In the first case, the authors refer to the properties

perspective, in the second case to the patterns perspective. This ambiguity is unfortunate,

as it leaves the reader wondering whether generativity refers to an inherent quality or to

a path-dependent process, a distinction which I make with Article II.

Third, I apply the two perspectives to discuss generativity of platform-based digital

ecosystems, or digital ecosystems which are organized around a central digital platform

(Tiwana et al. 2010). The discussion shows how the generative properties perspective

highlights the role of digital platform design and ecosystem governance in creating and

harnessing generative potential. Meanwhile, the generative patterns perspective lends

itself to an analysis of how successive resolutions of tensions between individual

ecosystem actors may change the overall system in unanticipated directions.

2.3 Detection of open source software ecosystems – Article III

With a nuanced description of generativity in digital ecosystems as conceptual

foundation, I continued with designing and carrying out the empirical part of this thesis.

As empirical setting I deliberately selected open source software, because six critical

characteristics are aligned with the objectives of this thesis. First, the open source

software model promotes a distributed way to organize (Raymond 1999). Second, it

developed governance mechanisms that are open to distributed change yet ensure

sufficient consistency and reliability (Kogut and Metiu 2001). Third, the open source

way has proved to be viable and successful (von Hippel and von Krogh 2003). Forth,

individual open source software is often combined with other open source software to

form a larger digital ecosystem (Gonzalez-Barahona et al. 2009). Fifth, open source

software has a tendency to be changed continuously (Howison and Crowston 2014).

Sixth, these change processes leave ample traces of activity in the public domain, which

can be analyzed by the avid researcher (Howison et al. 2011).

The population of open source software goes into the millions (cf. GitHub 2017d).

Because many open source software ecosystems – that is, digital ecosystems composed

of open source software – are distributed and emergent, interdependencies between

individual elements and thus ecosystem boundaries are not easily detectable (cf. Faraj

et al. 2016). Existing research of open source software ecosystems in information

systems (e.g., Singh et al. 2011; Um et al. 2015) followed a top-down approach to

delineate ecosystem boundaries.


Given that generativity emphasizes the distributed and emergent nature of complex

ecosystems, I decided to employ a bottom-up, data-driven approach to detect ecosystem

boundaries with Article III. The computational detection technique, adapted from

Blincoe et al. (2015) and implemented via self-written software, yielded a network graph

that consisted of 144,311 individual open source software projects and how they related

to each other.4 With the help of an algorithm (Rosvall and Bergstrom 2008) I was then

able to detect 22,609 different ecosystems, the largest of which consisted of 941

individual open source software projects. Figure 2 shows a visual representation of the

identified projects (nodes), their interdependencies (edges) and ecosystem boundaries

(colors).5

Figure 2. Visualization of Github projects (nodes) with their identified interrelations (edges)

Data-driven empirical research is prone to some methodical pitfalls (Howison et al.

2011) but also enables worthwhile exploratory sense-making (Grover and Lyytinen

2015). Mindful of this context, Article III proposes and applies three exploratory data

4 This approach is scalable. Between elaborating Article III in April 2016 and starting the design of Article IV in July 2016, I added more data, which let the network grow to 201,606 projects. 5 The visualization is depicted here to illustrate the scale of collected data. It shows the status as of 14 February 2016 and consists of 135,626 nodes and 231,760 edges. I continued to collect data from Github until 01 July 2016. The created database ultimately consisted of 201,606 nodes and 343,344 edges between these nodes.


loops to (1) assess data quality, (2) inform decision-making, and (3) detect regularities.

This methodical guidance is aimed to systematically help a researcher become familiar

with an initially unknown dataset, very much in line with the tradition of exploratory

data analysis pioneered by Tukey (1977).

A major result of exploring the collected dataset was the identification of two very

different topologies of large open source software ecosystems, exemplified by the Ruby

on Rails ecosystem and the Minecraft Forge ecosystem, respectively. While the first

ecosystem is very much centered around one open source software project, the latter is

considerably more distributed. The topologies resemble the digital platform and digital

infrastructure structures distinguished in Article I, which underlines the usefulness of

making this distinction. Moreover, Article III indicates that platform-based digital

ecosystems might be more common than infrastructure-based digital ecosystems, and

suggests that the largest and most popular open source software ecosystems tend to rely

on a central digital platform.

2.4 Cross-community coordination as means of generative system

change – Article IV

The last part of this thesis investigates the practice of cross-community coordination,

that is coordination between open source software projects. Cross-community

coordination is relevant to explain generative change in digital ecosystems, because it is

driven by “unfiltered contributions from broad and varied audiences” (Zittrain 2008:

70). In this case, unfiltered contributions are changes to outside open source software

artifacts on which the focal artifact relies, and broad and varied audiences are the many

participants in distributed open source software communities. The goal of Article IV

was to identify mechanisms with which actors from multiple open source software

projects reliably coordinate changes made in any part of the digital ecosystem.

The ability to detect individual ecosystems in a large dataset of open source software

projects, the major result of Article III, was fundamental to make case selection for this

last article. I chose to probe into the ecosystem around Ruby on Rails, because

exploratory data analysis identified it as a prime exemplar of a platform-based digital

ecosystem that exhibited considerable, constant activity driven by large communities

over an extended period of time. By applying selection criteria that are in line with the

research objective – namely change activity, maturity, distributedness and popularity –

I was able to select five suitable open source software projects within the Ruby on Rails

ecosystem as empirical setting for an explanatory case study.


The case study traces 96 episodes of cross-community coordination and distills their

particular characteristics, such as what triggered the episode, how many actors were

involved, how much time the episode took to resolve, or whether and how much source

code was changed in the process. Data was gathered and coded from Github discussion

boards, illustrated with a screenshot in Figure 3. The process was inspired by research

from Lindberg et al. (2016) who also analyzed Github discussion board entries to study

coordination (albeit within and not across communities).

Figure 3. Exemplary screenshot of a discussion board on Github

In conducting the case study, I identified four distinct arrangements to coordinate

changes across interdependent open source software projects, namely adaptation,

upgrading, positioning and departure, collectively defined in Table 4.


Table 4. Identified mechanisms of cross-community coordination (Eck 2018)

Mechanism Definition

Adaptation A coordination mechanism wherein “one community seeks to adapt its artifact to changes in an upstream artifact as to restore functionality or utilize new functionality” (Eck 2018: 8).

Upgrading A coordination mechanism wherein “one community seeks to understand major changes in an upstream artifact and […] upgrades the downstream artifact to appropriate these changes” (ibid.: 8).

Positioning A coordination mechanism wherein “two communities seek to track down an issue observed from coupling their respective artifacts and introduce change at the position they jointly agree on” (ibid.: 9).

Departure A coordination mechanism wherein “an actor, based on experiences made with a coupled artifact, proposes a change and the focal community decides whether and how to implement the change despite it being a departure from existing design” (ibid.: 9).

In the vocabulary of Article II, these coordination mechanisms represent generative

properties. They are a set of governance arrangements that instill generative capacity in

the digital ecosystem. By actualizing these different arrangements, distributed and

heterogeneous actors can produce unanticipated (and even innovative) change, but in a

reliable manner that neither breaks functioning of an individual open source software

artifact nor leads to a demise of the overall ecosystem. To the contrary, Article IV

indicates that cross-community coordination is an efficient way to organize continuous

change in digital ecosystems, at least in the empirical setting of the case study.

Part A: Discussion and future research 17

3 Discussion and future research

3.1 Theoretical contributions

This thesis contributes to our knowledge of generativity in digital ecosystems, more

specifically to how distributed and heterogeneous actors organize for continuous change

in digital ecosystems. In the following, I briefly summarize major theoretical

contributions along the three research questions.

RQ.1 asked about a description of generativity in digital ecosystems. To this end,

Article I clarifies the pertinent ambiguity around the generativity term (e.g., Bygstad

2015; Förderer et al. 2014). It also summarizes the reasoning of existing research that

generative change in complex digital ecosystems – both platform-based and

infrastructure-based – may lead to innovative outcomes. Building on this foundation,

Article II develops a socio-technical description of generativity in digital ecosystems

from two complementary perspectives. While the generative properties perspective

highlights artifact design and governance arrangements that instill generative capacity

into a digital ecosystem, the generative patterns perspective puts forth the thought that

digital ecosystems may evolve over time such that it exceeds the imagination of its

original designers. This distinction is relevant for theory-building, as each perspective

suggests different, yet complementary research designs.

RQ.2 was concerned with empirically detecting digital ecosystems in a large sample of

open source software projects. Article III adapts the approach of Blincoe et al. (2015) to

computationally detect open source software ecosystems, based on millions of data

points collected on Github, a popular source code hosting and online collaboration

service for open source software projects (cf. Gousios 2013). The article complements

this approach by proposing and applying three exploratory data loops to (1) assess data

quality, (2) inform decision-making, and (3) detect interesting regularities. This

methodical guidance is inspired by the tradition of exploratory data analysis (Tukey

1977) and contributes to the call for more data-driven research in information systems

research in general (Grover and Lyytinen 2015) and digital ecosystem research in

particular (Reuver et al. 2017).

Finally, RQ.3 asked about coordination mechanisms in mature and distributed digital

ecosystems, as an exemplary set of generative governance arrangements. Article IV

identifies four cross-community coordination mechanisms from an explanatory multiple

case study of five interrelated open source software projects. These mechanisms –

18 Part A: Discussion and future research

namely adaptation, upgrading, positioning and departure – demonstrate the broad

arsenal of means that actors employ to accommodate continuous change happening in

the digital ecosystem they are part of. To this end, Article IV contributes to the call for

identifying socio-technical explanations for generative digital ecosystems (Yoo 2013).

3.2 Practical implications

This thesis holds managerial lessons to organizations that are part of a dynamic digital

ecosystem or want to play on this field. First, the infinite malleability and perpetual

incompleteness of digital artifacts favors an organizational logic that is supportive of

and exploits these particular qualities. Such an organizational logic is centered around

digital ecosystems, that is collections of co-evolving, interdependent digital artifacts and

the social actors related to these artifacts (Article I). Thus, managers should seek outside

alliances and actively engage in digital ecosystems to profit from the promise of digital

technologies (cf. Basole 2016). Furthermore, they can expect that continuous change in

all parts of the organization will become the norm (cf. Garud et al. 2008).

Second, digital ecosystems can come in different topologies, ranging from a structure in

which a central digital platform dominates, all the way to a topology that is supported

by an entangled web of diverse infrastructure (Articles I & III). Managers should be

mindful of the various possibilities and weigh the respective advantages and

disadvantages carefully (cf. Albert and Barabási 2002). Without providing (nor seeking)

a theoretical explanation, this thesis indicates that digital ecosystems with many

distributed participants tend to favor a platform-based topology (Article III).

Third, from analyzing the ecosystem around Ruby on Rails, four stable arrangements to

organize continuous change in a complex platform-based digital ecosystem were

identified (Article IV). These coordination mechanisms offer several cues for

managerial advice. The adaptation and upgrading mechanisms show how a digital

platform can be changed continuously (in small and large steps, respectively), without

knowing the dependent elements in the ecosystem (cf. Tilson et al. 2010). The

positioning mechanism illustrates how design of a digital platform is strengthened as a

consequence of maintaining many ties with other elements in the ecosystem. The

departure mechanism suggests that mature open source software projects are guided by

a strong, and commonly accepted, sense of direction.

Fourth, the distinction of generative properties and generative patterns (Article II)

suggests that a generative digital ecosystem not only attracts contributions from broad

Part A: Discussion and future research 19

and varied audiences, but that it may change beyond the imagination and ambition of its

original designers in the process. Organizations that orchestrate a digital ecosystem

should regard such emergent dynamics as entrepreneurial opportunities and actively

seek to support those actors in the ecosystem which produce unanticipated changes,

while being mindful of actors that use these same dynamics to contest ecosystem control

(cf. Eaton et al. 2015).

3.3 Limitations and future research

Organizing for constant change in distributed digital ecosystems is a rich field for

research. While this thesis advances our knowledge in selected areas – conceptually with

a clearer description of the generativity construct and empirically with an identification

of coordination mechanisms across open source software communities – many questions

remain unanswered. This thesis therefore, as any piece of research, has limitations which

invites future research. Each individual article in Part B of this thesis contains a

discussion of possible avenues for further research within the scope of the respective

article. In the following, I outline three major limitations and resulting research

opportunities from the perspective of the overall thesis.

First, there are likely many more arrangements to organize constant change in a digital

ecosystem than the four cross-community mechanisms identified in Article IV. Indeed,

given the employed research methodology, it is unclear how generalizable these results

are beyond the studied empirical context, which invites similar research in other digital

ecosystems. Furthermore, follow-up research could investigate cross-community

coordination in ecosystems that do not rely on a central digital platform such as Ruby

on Rails, and it could study to which extent the identified mechanisms are valid beyond

ecosystems composed of open source software. In addition, the research presented in

this thesis focused only on coordination mechanisms, but additional governance

instruments on ecosystem level exist (cf. Tiwana et al. 2013). For example, future

research could investigate whether and how the overall ecosystem architecture is

negotiated between the members of a distributed digital ecosystem (cf. Weyns and

Ahmad 2013).

Second, this thesis remained silent on dynamic aspects of digital ecosystems, that is how

and why their elements, structure and behavior change over time. The proposed

approach in Article III to identify open source software ecosystems draws from time-

stamped empirical data. Therefore, follow-up research could extract activity measures

from the source data and trace the growth (and possibly decay) of various ecosystems.

20 Part A: Discussion and future research

Based on such an investigation, generative patterns (Article II) could be distilled. In

addition, the research of Article IV could be extended to capture how less mature

ecosystems develop stable arrangements to coordinate over time. To decrease the

influence of such dynamic effects, I purposefully selected a mature digital ecosystem to

study cross-community coordination. Yet, even in mature organizational arrangements

learnt behavior is malleable (cf. Doz 1996). Therefore, future research could analyze

under which circumstances stable mechanisms to coordinate are adjusted, for example

following critical events (cf. Lyytinen and Newman 2008).

Third, complementary to social arrangements, artifact designs contribute to the

organization of change in a digital ecosystem (Hanseth and Lyytinen 2010). While

Article I discusses why digital artifacts are generally well-suited to be changed, this

thought is not further developed in this thesis. Differences in artifact designs make it

easier or harder to actualize change potential (cf. Yoo et al. 2010) and even determine

such potential (Woodard et al. 2013). Therefore, further research could study the

common artifact design features of highly generative ecosystems and how they differ

from less generative ecosystems (Woodard and Clemons 2014).

Part A: Reference overview of articles in this thesis 21

4 Reference overview of articles in this thesis

The core part of this dissertation (Part B) are four articles that jointly address the

research objective of the dissertation endeavor. They are contextualized regarding their

contributions to the thesis in Section 3. In the following, an overview of the included

articles with bibliographic information and abstract is given for reference.

4.1 The generative capacity of digital artifacts: a mapping of the field

– Article I

Table 5. Bibliographic information for Article I

Title The Generative Capacity of Digital Artifacts: A Mapping of the Field

Authors Alexander Eck, Falk Uebernickel, Walter Brenner

Outlet PACIS 2015 Proceedings

Year 2015

Status Published

Abstract. The concept of generativity as the capacity of a technology or a system to be

malleable by diverse groups of actors in unanticipated ways has recently gained

considerable traction in information systems research. We review a sample of the body

of knowledge and identify that scholars commonly investigated generativity in

conjunction with digital infrastructures and digital platforms, both of which are

complex, networked, and evolving socio-technical systems. Interestingly, other types of

digital artifacts have been neglected, despite our initial assumption that the distinct

attributes (e.g., reprogrammability, distributedness) of any digital artifact match well

with generativity. The literature review also reveals that innovation brought about

heterogeneous groups of actors is universally regarded as the goal of generativity,

discounting the possibility of exploiting generative systems towards other valuable ends

such as organizational agility. Furthermore, scholars commonly discuss generativity in

conjunction with the logic of modularity, leading to unresolved questions on how these

two concepts might complement each other. Another important contribution of this

paper is the systematization of various meanings of generativity, spanning from the

philosophical–e.g., generative mechanisms in critical realist research–to a more literal

understanding, for instance generativity as synonym to ‘creation of a particular

solution’.

22 Part A: Reference overview of articles in this thesis

4.2 Untangling generativity: two perspectives on unanticipated

change produced by diverse actors – Article II

Table 6. Bibliographic information for Article II

Title Untangling Generativity: Two Perspectives on Unanticipated Change Produced by Diverse Actors

Authors Alexander Eck, Falk Uebernickel

Outlet ECIS 2016 Proceedings

Year 2016

Status Published

Abstract. Digital platform ecosystems capitalize on the engagement of large groups of

actors with diverse skills that create unexpected services, find novel uses, and move the

ecosystem forward in unanticipated ways. The generativity concept captures this

phenomenon, which some scholars regard as fundamental to our understanding of how

digital innovation plays out. Despite such claims, it is unclear what generativity means

and how it manifests itself. After scrutinizing the foundational definition, we delineate

two perspectives on generativity. The ‘generative properties’ perspective asks which

properties of digital artifacts embedded in social structure invite actors to set free their

creativity and produce unanticipated outcomes. In contrast, the ‘generative patterns’

perspective asks which patterns of events lead to an evolutionary dynamic that produces

unanticipated change. We find cues for both perspectives in literature, which tentatively

validates them. We then formulate a socio-technical model of digital platform

ecosystems and describe ecosystem generativity by applying both perspectives. In this

context, the first perspective highlights the digital platform and its controlling

orchestrator, while the second perspective directs our attention to interactions among

actors and artifacts. We conclude this paper by discussing how generativity can help

explain digital innovation.

Part A: Reference overview of articles in this thesis 23

4.3 Reconstructing open source software ecosystems: finding

structure in digital traces – Article III

Table 7. Bibliographic information for Article III

Title Reconstructing Open Source Software Ecosystems: Finding Structure in Digital Traces


Outlet ICIS 2016 Proceedings

Year 2016

Status Published

Abstract. We report on the computational reconstruction of 273 open source software

ecosystems, consisting of 41,388 artifacts and couplings between them, extracted from

digital traces of 34.4 million software artifacts. We argue that digital traces are a new

kind of data source, and propose ‘exploratory data loops’ to exploit the benefits of digital

trace data in early stages of a research program. We apply this schema to systematically

assess data quality, inform sample selection, and detect patterns. Empirically, we show

that highly distributed networks are unlikely to follow a hierarchically modular

structure, despite popular belief. As is shown visually with two examples, very distinct

structures can emerge from autonomous behavior. The results indicate that different, yet

similarly effective, strategies may exist to organize for distributed innovation in digital

ecosystems. The paper is concluded by outlining how follow-up work will harness the

reconstructed ecosystems for detecting behavioral patterns in distributed networks.

24 Part A: Reference overview of articles in this thesis

4.4 Coordination across open source software communities: findings

from the Rails ecosystem – Article IV

Table 8. Bibliographic information for Article IV

Title Coordination Across Open Source Software Communities: Findings from the Rails Ecosystem

Authors Alexander Eck

Outlet MKWI 2018 Proceedings

Year 2018

Status Published

Abstract. While coordination of work within open source software (OSS) communities

is well-researched, it is virtually unknown how they coordinate across community

boundaries. However, as OSS projects are often part of a larger digital ecosystem of

interdependent artifacts and communities, cross-community coordination is a pertinent

topic. We turn to the ecosystem around Ruby on Rails to empirically explore this

research gap. To this end, we scrutinize 96 coordination episodes among five

interrelated OSS projects and identify four cross-community coordination mechanisms:

adaptation, upgrading, positioning, and departure. Each mechanism describes a distinct

and stable arrangement to integrate contributions across community borders. After

presenting our findings, we reason about the significance of the results on explaining

generative change in digital ecosystems.

Part B: The generative capacity of digital artifacts: a mapping of the field 25

Part B

I The generative capacity of digital artifacts: a mapping of

the field

Table 9. Bibliographic information for Article I

Title The Generative Capacity of Digital Artifacts: A Mapping of the Field

Authors Alexander Eck, Falk Uebernickel, Walter Brenner

Outlet PACIS 2015 Proceedings

Year 2015

Status Published

Abstract. The concept of generativity as the capacity of a technology or a system to be

malleable by diverse groups of actors in unanticipated ways has recently gained

considerable traction in information systems research. We review a sample of the body

of knowledge and identify that scholars commonly investigated generativity in

conjunction with digital infrastructures and digital platforms, both of which are

complex, networked, and evolving socio-technical systems. Interestingly, other types of

digital artifacts have been neglected, despite our initial assumption that the distinct

attributes (e.g., reprogrammability, distributedness) of any digital artifact match well

with generativity. The literature review also reveals that innovation brought about

heterogeneous groups of actors is universally regarded as the goal of generativity,

discounting the possibility of exploiting generative systems towards other valuable ends

such as organizational agility. Furthermore, scholars commonly discuss generativity in

conjunction with the logic of modularity, leading to unresolved questions on how these

two concepts might complement each other. Another important contribution of this

paper is the systematization of various meanings of generativity, spanning from the

philosophical–e.g., generative mechanisms in critical realist research–to a more literal

understanding, for instance generativity as synonym to ‘creation of a particular

solution’.

Keywords: generativity, innovation, digital artifact, infrastructure, platform, modularity

26 Part B: The generative capacity of digital artifacts: a mapping of the field

I.1 Introduction

Fueled by recent calls for research (Tilson et al. 2010; Yoo et al. 2010) the generativity

concept, defined by Zittrain as the capacity of a technology or a system to be malleable

by diverse groups of actors in unanticipated ways (2006; 2008), has gained some

prominence in the information systems (IS) research community. Consistent with the

mentioned calls for research, recent literature explored this concept in the context of

digital infrastructures (cf. Tilson et al. 2010) and digital platforms (cf. Yoo et al. 2010)

and how it drives innovation. For instance, Ghazawneh and Henfridsson (2013)

discussed how platform owners design and tweak boundary resources such as APIs

(application programming interfaces) to balance generative capacity with control

concerns; Boudreau (2012) investigated how the number of third-party application

developers increase the attractiveness of a platform through collective effects of high

variety and innovation; and Selander et al. (2013) ask how third-party actors choose the

platforms in which they intend to participate and how they extract value from their share

of innovation.

Zittrain’s understanding of generativity6 was mostly borne out of law discourses

(Zittrain 2008) before it entered the IS domain. When a new concept is adopted, a clear

characterization in relation to existing scholarly discourses is called for, in order to grasp

the phenomenon behind the term. In light of Yoo’s (2013) call for research to examine

generativity as novel contribution to innovation and technology management research,

we regard such a characterization to constitute a crucial first step towards this goal. With

this paper we therefore seek to trace how IS scholars grafted generativity into existing

IS-related roots (cf. Truex et al. 2006), that is in which established discourses within IS

research this newly adopted generativity concept has been interwoven. We formulate

our goal accordingly:

In relation to which IS discourses has Zittrain’s generativity concept been commonly

discussed?

Before we can tackle this substantial question, we must eliminate any terminological

ambiguity first. As Avital and Te'eni (2009) pointed out in their own alternative

conceptualization of generativity, in its broadest sense the term “refers to a capacity of

producing or creating something” (ibid.: 2), leading them to conclude that the term has

6 When mentioning generativity hereafter, we refer to Zittrain’s conceptualization of the term if not otherwise stated.


been utilized to mean different things in different research contexts within social

sciences. The obvious danger is that the uninitiated scholar is puzzled when the same

term is employed to mean fundamentally different things (cf. Gerring 2012). As a

discipline that habitually borrows theories and concepts from other fields (Baskerville

and Myers 2002), IS research seems particularly prone to evoking such potential

misunderstandings. As preparatory step we hence aim to distinguish Zittrain’s concept

of generativity from other meanings carrying the same term in IS research.

We start by introducing the generativity concept and placing it in the context of digital

artifacts. Then we describe how we tackled the research question and elucidate our

rationale for proceeding as we did. We continue by briefly presenting the different

meanings of generativity, as to unravel any terminology issues. Equipped with a better

grasp of the relevant knowledge base, we focus the following discussion on the scholarly

discourses around generativity, approaching the topic from two slightly different angles:

first, we trace from which roots IS scholars commonly drew when examining

generativity. Second, we analyze which digital artifacts have been proposed in the

salient literature to carry generative capacity. Finally, we identify the main shortcoming

of this paper and suggest selected avenues for future research.

I.2 Generativity in the context of digital artifacts

With an ever increasing share of physical artifacts complemented by digital components

or fully replaced by digital artifacts, digital technologies become the dominant source

for innovation (Lyytinen and Yoo 2002). According to widely accepted estimates about

80% of all innovations in the automotive industry are directly related to novel uses of

digital technologies, just to name one example (Leen and Heffernan 2002; Mossinger

2010). This push into the digital realm increases competitive pressure and

simultaneously affords unprecedented innovation across industries (Bharadwaj et al.

2013). Several scholars argued that digital artifacts differ distinctively from their

physical counterparts and concluded that organizations cannot deal with digitization

without considerably altering their approach to doing business. For instance, Svahn and

Henfridsson (2012) distinguish two innovation regimes, with the product innovation

regime rooted in the physical realm and the IT innovation regime dealing with the

peculiarities of the digital realm. The authors contend that the imbrication of physical

and digital artifacts requires organizations to become proficient in both innovation

regimes simultaneously.


To provide a satisfactory account of the numerous narratives around the reciprocal

relationships of technological advancement and organizational change, it would be

necessary to trace back to the pioneering work on socio-technical systems some six

decades ago (Emery and Trist 1965; Trist and Bamforth 1951). For the purpose of this

paper and acknowledging the resulting limitations, we allow ourselves to regard

Zittrain’s work on generativity as departure point. In his foundational article published

in 2006, the author suggests that the generative capacity of digital technologies in

general and of the internet in particular lies at the core of wide-spread and distributed

innovation, creativity, and entrepreneurial activity associated with these technologies

(cf. Chesbrough 2003; von Hippel 2005). He defines generativity as “a technology’s

overall capacity to produce unprompted change driven by large, varied, and

uncoordinated audiences” (Zittrain 2006: 1980). In his successive book, the author

refined this definition, suggesting that generativity “is a system’s capacity to produce

unanticipated change through unfiltered contributions from broad and varied

audiences” (Zittrain 2008: 70). With this expansion to a systems perspective, the author

arguably intends to capture three aspects of generativity: first, that technologies can

drive individual and collective creativity, for instance through providing tools for artistic

expression, or by facilitating collaboration on long distances (Zittrain 2008). Second,

that only through the participation of humans the generative capacity of a technology

can be realized (Zittrain 2008). And third, that innovation happens on different layers–

e.g., technology, content, and society–each of which may possess generative capacity

on their own (Zittrain 2008).

Further dissecting what makes a system generative, Zittrain (2008) suggests five

characteristics of generativity, namely: leverage, adaptability, ease of mastery,

accessibility, and transferability. Leverage refers to the extent by which a system actor’s

productivity is increased compared to an actor performing outside the system–similar to

the metaphor of a computer being like a bicycle for the mind (Krainin and Lawrence

1990). Adaptability indicates how malleable a system is for application in many and

varied contexts. Ease of mastery denotes how understandable a system is and also how

much effort an actor must put into becoming proficient to adapt it. Accessibility reflects

how low the barriers of entry are. Finally, transferability signifies how readily changes

in one part of the system can be conveyed to other parts of the system or distributed to

anther system instantiation.

This concept of emergent change and innovation resonates within IS research, because

it matches well with characteristics of digital artifacts, which we draw from Kallinikos


et al. (2013). They distill four immediate characteristics (interactive; editable;

reprogrammable; distributed) and three corollary attributes (modular; granular;

reflexive) of digital artifacts. Interactivity denotes the possibility to explore a digital

artifact, its individual components, and dependencies. Editability relates to the

possibility of modifying the artifact while leaving its logical structure unchanged.

Reprogrammability reflects the possibility of releasing a digital artifact from its

immediate use context, modify its structure, and repurpose it. Distributedness signifies

that digital artifacts are not confined to any physical or institutional borders. Modularity

refers to the distinct quality of modularized digital artifacts not to be bound to a fixed

product architecture, meaning that individual modules of a complex digital artifact can

be transferred to completely unrelated use contexts. Granularity stands for the inherent

decomposability of digital artifacts, down to their basic binary representation, and for

the associated possibility to modify both an insignificant and a substantial part of the

artifact on different levels of abstraction. Lastly, the reflexive dynamics of digital

artifacts carries the notion that any access, assembly, or otherwise manipulation can only

be performed through making use of other digital artifacts. Consequently, any domain

in which digital artifacts enter will invariantly see an increase of digital artifacts over

time.

In Table 10, we give some illustrative examples from IS literature in which the use of

digital artifacts are accounted to have led to an innovative outcome. For each presented

example we singled out one digital artifact attribute that we would argue played a central

role in creating the innovation. Furthermore, we highlighted one of the five generativity

attributes to which the example may be associated with. For instance, Boland et al.

(2007) (example 14) demonstrate how the reflexive dynamics of digital artifacts–in this

case novel 3D visualization technology which was adopted by one architect company–

led to a ripple effect, sparking digital innovation throughout the heterogeneous partner

network of the focal firm. These dynamics also constantly reshaped how the focal firm

was able to carry out work, hence suggesting that changes brought about 3D

visualization in one part of the partner network transferred to another part, all without

a central actor planning or overseeing this development.


Table 10. Illustration of links between digital artifacts attributes and generativity attributes

Illustrative example from IS literature (digital artifact highlighted in italics)

Digital artifact attribute

Generativity attribute

1 New mass-personalized content platforms such as Last.fm (Oestreicher-Singer and Zalmanson 2013)

Interactivity Leverage

2 Design features of self-service technologies for encouraging initial use (Meuter et al. 2005)

Interactivity Accessibility

3 Commenting and sharing functionality as means for content to “go viral” (Yoo 2010)

Editability Leverage

4 Hyperlinking as essential mechanism to provide cross-references and to let users browse the web (Shapiro and Varian 1999)

Editability Ease of mastery

5 Achieving high level of product quality in open source software development through social controls (von Krogh et al. 2012)

Reprogrammability Adaptability

6 Improving the control software of embedded systems after the product has been shipped (Lee and Berente 2012)

Reprogrammability Transferability

7 Convergence of devices, distribution channels, and markets (Tilson et al. 2010)

Distributedness Leverage

8 Distributed information gathering and sharing via informally organized social software (von Krogh 2012)

Distributedness Ease of mastery

9 Exploiting existing signal processing module to rapidly develop a new product (Woodard et al. 2013)

Modularity Adaptability

10 Design and tuning of APIs to achieve desired levels of changeability and control (Ghazawneh and Henfridsson 2013)

Modularity Accessibility

11 Digital platforms and competition of ecosystems (Tiwana et al. 2010)

Granularity Adaptability

12 Collaborative, dialectic discourse on ideation platforms (Majchrzak and Malhotra 2013)

Granularity Transferability

13 Large-scale information infrastructure design aimed at cultivating a growing installed base (Hanseth and Lyytinen 2010)

Reflexive dynamics Accessibility

14 Wakes of innovation in temporary firm networks following the introduction of 3D visualization in one firm (Boland et al. 2007)

Reflexive dynamics Transferability

It is worth noticing that the provided examples and their suggested associations are

merely illustrative and thus incomplete. They were compiled to serve as a face-value

test of how useful the proposed attributes of digital artifacts and of generativity might

be to rationalize innovative outcomes across a wide range of scenarios. We argue that a

systematic investigation of each example would uncover additional relationships


between the attributes not shown in Table 10 yet. For instance, Boland et al. (2007)

mention that 3D visualization not only caused an innovation ripple effect, but

furthermore that the distributedness of digital construction models was leveraged to

increase knowledge exchange between the project partners. These seemingly strong ties

between digital artifacts and generative capacity might be a major reason why IS

scholars became interested in this field of research (Yoo 2013).

I.3 Research method

For answering the research question we turned to reviewing publications that discuss

generativity in IS research and to describing the identified themes in summary (cf. Rowe

2014). The review aimed (1) to uncover which different concepts were attached the

‘generativity’ label by the IS community, and more substantially, (2) to reveal to which

existing discourses within the IS discipline Zittrain’s generativity concept has been

commonly related to. We followed the recommendations of Cooper (1998) and

conducted research along the five logical stages of problem formulation, literature

search, evaluation, analysis, and presentation. In doing so, we greatly profited from

suggestions to increase process quality and to avoid potential errors associated with each

of these stages. This review is aimed at the IS research community, therefore we

confined literature search to the eight journals considered to be leading (AIS 2011). A

recent bibliographic analysis provided solid indication that articles published in these

outlets are good proxies for capturing the most visible conversations in the IS

community (Lowry et al. 2013). An exhaustive review of all IS publications was beyond

the objective of this paper, so we decided not to apply a cascading search strategy

involving backward and forward search or the gradual expansion of the target outlets.

We did however apply a very broad search strategy within the selected journals: we

searched the full text for any occurrence of the words generativity or generative, without

further specification. This search was carried out on February 13, 2015 and yielded 130

hits. We excluded all articles that carried the search term in their reference section only.

As an exception, we included the article from Selander et al. (2013), because the authors

clearly related their research to a particular conceptualization of generativity. We also

excluded all articles that employed the searched-for-term merely as an attribute. For

instance, Huang et al. (2014) call one approach to instill ambidexterity opportunity-

generative (i.e., to identify/create business opportunities), but do not further elaborate

on this term. In total, the search strategy resulted in a review sample of 52 publications

for further evaluation and analysis. The distribution among the individual outlets is

shown in Table 11.


Table 11. Literature search results

# Source Database Hits Relevant

1 EJIS (European Journal of Information Systems) Proquest 37 13

2 ISJ (Information Systems Journal) Wiley 28 6

3 ISR (Information Systems Research) Informs 3 3

4 JAIS (Journal of the Association for Information Systems) Aisel 14 7

5 JIT (Journal of Information Technology) Proquest 18 10

6 JMIS (Journal of Management Information Systems) Ebsco -- --

7 JSIS (Journal of Strategic Information Systems) Sciencedirect 23 6

8 MISQ (Management Information Systems Quarterly) Ebsco 7 7

130 52

Subsequently, we coded the retained articles according to the scholarly discourse from

which they borrowed their understanding of generativity. For example, Lambert and

Peppard (1993) draw from organizational learning literature, and specifically refer to

the work of Argyris (1976) when they conceptualize “generative or double loop

learning (as) (…) new ways of looking at the world, challenging assumptions, goals,

and norms” (Lambert and Peppard 1993: 192). When no discernible source was

apparent, we deducted the scholarly discourse directly from the analyzed paper. With

the aim to increase parsimony, we merged categories in a second round when the topics

of the papers appeared to be sufficiently closely related. For example, this led us to

collapse all articles discussing issues of organizational knowledge and organizational

learning into a single category. This procedure resulted in 12 remaining categories

related to scholarly discourses which developed one or more meanings of generativity.

Figure 4. Number of sources (y-axis) referenced in at least n papers (x-axis)

For complementing data collection pertaining the research question, we started with the

subset of 14 papers that follow Zittrain’s generativity concept. We grouped those articles

according to the type of digital artifact they focused on, resulting in 3 categories. We

also extracted the bibliographic information of the 14 papers, which yielded 959 entries,

of which 706 were unique. Figure 4 above depicts how many source papers (y-axis)

were referenced at least n times (x-axis). For example, two sources were mentioned at

1 2 3 4 6 9 13 22 53 140

706

0

500

1000

11 10 9 8 7 6 5 4 3 2 1


least 10 times (one source 11 times, and another one 10 times). We were interested in

the main sources from which the 14 papers collectively drew. We decided to include

those sources for further analysis which were mentioned by at least one quarter of all

papers, which we found to be a reasonable cut-off value. This led to 22 sources which

were referenced at least 4 times. For clarification and illustrative purposes, Figure 5

depicts the resultant citation network, with the directed edges denoting the referencing

direction and the color shading indicating the number of references.

Figure 5. The 14 sample papers and the 22 sources which were referenced at least 4 times

We discarded four referenced articles for further evaluation because these were cross-

references within our sample of 14 (Ghazawneh and Henfridsson 2013; Hanseth and

Lyytinen 2010; Tilson et al. 2010; Yoo et al. 2010). We also dropped Zittrain’s (2006)

article on generativity and Klein and Myers’ (1999) paper on how to conduct interpretive

case study research. Analogous to the approach towards answering the first research

question, we then coded the remaining 16 sources according to the scholarly discourse

in which they were mentioned in the referencing articles, and merged closely related

categories in a second round. This resulted in 6 distinct categories, showing which

research streams were most commonly discussed in the 14 papers and thus discussed in

conjunction with Zittrain’s generativity concept.

I.4 The many meanings of generativity in IS research

We stated in the introduction that whenever a discipline habitually borrows concepts

and theories from other areas, it is likely that the same terminology will be used to mean


different things, which might puzzle a researcher new to a particular topic. The results,

summarized in Table 12, show that indeed very heterogeneous concepts have been

labeled with the generativity term. The associated meanings span from the

philosophical–like the generative mechanisms central to critical realist research–all the

way down to a more literal understanding of generativity in design ethnography as

creating a particular solution. The first occurrence of generativity in our sample dates

from 1985 (Lyytinen 1985), but 23 (i.e., nearly half) of the articles were published since

2013. Mapping the generativity landscape therefore seems to be a useful contribution to

a timely scholarly conversation. Owing to the limited ambition of this article we refrain

from summarizing the different scholarly discourses in which the generativity term

plays a role and refer the interested reader to the reference literature collected below.

For the remainder of this paper we focus exclusively on Zittrain’s understanding of

generativity, i.e. the capacity of a technology or a system to be malleable by diverse

groups of actors in unanticipated ways.

Table 12. Same term, different things: meanings of generativity in identified articles

Scholarly discourse

Meaning(s) of generativity

Referenced literature (selection)

Articles applying referenced meaning of generativity #

Law discourses, in particular on free software and commons

Generativity of technologies and systems as capacity to be malleable by diverse groups of actors in unanticipated ways

Zittrain 2006, 2008 Bergvall-Kåreborn and Howcroft 2014; Eaton et al. 2015; Ghazawneh and Henfridsson 2013; Grisot et al. 2014; Hanseth and Lyytinen 2010; Henfridsson et al. 2014; Henningsson and Henriksen 2011; Nambisan 2013; Tilson et al. 2010; Yoo et al. 2010; Yoo 2013

14

Tilson et al. 2010; Yoo et al. 2010

Lusch and Nambisan 2015; Racherla and Mandviwalla 2013; Selander et al. 2013

Critical realism

Generative mechanisms as structures with enduring properties that are capable to cause observable events

Bhaskar 1978, 1998 Henfridsson and Bygstad 2013; Klecum et al. 2014; Lyytinen and Newman 2008; Volkoff and Strong 2013; Walsh 2014

11

Pentland 1999 Avgerou 2013; McLeod and Doolin 2012

Bourdieu 1973, 1998 Schultze and Boland 2000

Harré and Madden 1975 Chae and Poole 2005

Myers and Klein 2011 Cecez-Kecmanovic 2011

Roberts 2006 Chatterjee and Sarker 2013


Scholarly discourse

Meaning(s) of generativity

Referenced literature (selection)

Articles applying referenced meaning of generativity #

Organizational knowledge & organizational learning discourses

Generativity as basis for creative work; generative dance of knowledge and knowing; double-loop and generative learning; generative variation

Orlikowski 2006 7

Orlikowski 2006 Galliers 2006; Swan 2006

Cook and Brown 1999 Pozzebon and Pinsonneault 2012

Argyris 1976 Lambert and Peppard 1993

Senge 1992 Huysman et al. 1994

Zollo and Winter 2002 Prieto and Easterby-Smith 2006

Linguistics Generative grammar as a formal specification of language

Chomsky 1957, 1966, 1986

Gaskin et al. 2014; Lee et al. 2008; Lyytinen 1985; Shawe-Taylor 1987; Truex and Baskerville 1998

5

Organizational routines

Organizational routines as generative systems capable to produce a variety of performances depending on experiences and context

Feldman and Pentland 2003; Pentland and Feldman 2008; Pentland and Rueter 1994

Beynon-Davies 2010; Iannacci 2014; Kutsch et al. 2013; Robey et al. 2013

4

Complexity theories

Generative relationships; generativity of complex objects; generative process of self-organization

Lane and Maxfield 1996 McBride 2005 3

Law and Singleton 2005 Whitley and Darking 2006

Merali 2006

Institutional theories

Generative regime; institutionalization as generative process

Reimers 1996 2

Zucker and Darby 2005 Baptista 2009

Information systems design

Generative fit of IS designed to enhance generative capacity of humans

Avital and Te'eni 2009 2

Avital and Te'eni 2009 Majchrzak and Malhotra 2013

Design ethnography

Generative in the sense of creating a particular solution

Otto and Smith 2013 Baskerville and Myers 2015 1

Hermeneutic cycle

Generative structures as collective meaning

Klein and Myers 1999 Njenga and Brown 2012 1

Psychology Generative learning process

Wittrock 1974 Kwok et al. 2002 1

Public values Generative perspective Davis and West 2009 Pang et al. 2014 1


I.5 The generative capacity of digital artifacts

I.5.1 Approaching the scholarly discourses related to generativity from two

different angles

We already introduced Zittrain’s conceptualization of generativity as the capacity of a

technology or a system to be malleable by diverse groups of actors in unanticipated

ways. We posited that the particular characteristics of digital artifacts might enable or

facilitate generativity in myriad ways. In the reviewed literature sample, the majority of

articles focused on just two types of digital artifacts, namely digital platforms and digital

infrastructures, while three papers (or about 20%) addressed digital artifacts in general,

see also Table 13 below.

Table 13. Types of digital artifacts and themes discussed in the literature review sample

Digital artifact type Themes Approach Articles #

Digital infrastructures

Shared, open, heterogeneous, and evolving system consisting of digital artifacts and their user, operations, and design communities

Dynamic complexity; bootstrapping; installed base; paradox of change and control; procrastination principle; inscription and interpretation

Conceptual Hanseth and Lyytinen 2010; Tilson et al. 2010

2

Empirical Grisot et al. 2014; Henningsson and Henriksen 2011; Racherla and Mandviwalla 2013

3

Digital platforms

Extensible framework that addresses a family of generic functionalities meeting the needs of heterogeneous user communities

platform-controlling actor and third-party actors; ecosystem; boundary resources; emergent tensions and dialectic resolution; modular-layered architecture

Conceptual Lusch and Nambisan 2015; Yoo et al. 2010

2

Empirical Bergvall-Kåreborn and Howcroft 2014; Eaton et al. 2015; Ghazawneh and Henfridsson 2013; Selander et al. 2013

4

Digital artifacts (in general)

Object created by and composed of digital technology and the outcome of coordinated human action

Hierarchy of parts and network of patterns; digital artifacts as operant resources

Conceptual Nambisan 2013; Yoo 2013

2

Empirical Henfridsson et al. 2014 1

Before turning to a description of the identified artifact types and the ways in which they

have been suggested to unlock or foster innovation through their generative capacity, it

is worth exploring established themes to which IS researchers related the nascent

generativity concept. Following an evaluation of the 14 sample articles listed above,

Table 14 below provides an overview of the scholarly discourses with which the

generativity concept has been commonly associated, and of the literature base jointly


regarded as salient to these discussions. This gives us a slightly different angle for

reviewing the field of generativity.

As could be expected from the results summarized in Table 13 above, we find the two

discourses on digital infrastructures and network effects related to them, as well as on

digital platforms and the ecosystems those create. Furthermore, the role of digital

artifacts to enable and facilitate innovation has been commonly discussed, providing a

cue that the outcome of generativity–that is, unanticipated change by diverse groups of

actors–is thought to be innovation. The fourth dominating scholarly discourse deals with

decomposition of complex systems and modularity, suggesting that any discussion

about generativity should be conducted with consideration of the rich history of

modularity in IS research (cf. Yoo 2013). Finally, a few articles established links to

structuration theory and also to organizational agility. Because we do not aim for

exhaustive review of the literature base with this paper, we chose to disregard these two

themes. Nevertheless we acknowledge their relevance to generativity research, as

exemplified by investigations of Woodard and Clemons (2014) and Kretzer et al. (2014).

Hence, in what follows and drawing from the reviewed literature base, we describe the

four dominating strands which we identified and their association with generativity.


Table 14. Scholarly discourses commonly discussed in conjunction with generativity

Scholarly discourse Aspects discussed

Referenced literature

Sum of citations

Attributes and dynamics of digital infrastructures, network effects and installed base

Digital infrastructures are socio-technical systems; modern digital infrastructures span several organizations and may have global reach; evolution and innovation is non-linear, typically initiated to suit a specific local context and thus uncontrollable top-down; infrastructures might deviate from planned purpose over time; early standard-setting is central to fast growth; cumulative evolution leads to network effects; evolution of infrastructure is both enabled and constrained by installed base; intelligence at the endpoints, not the center of an infrastructure

Benkler 2006; Ciborra 2000; Shapiro and Varian 1999; Star and Ruhleder 1996

19

Attributes and dynamics of digital platforms and ecosystems

Ecosystems consist of a set of relatively stable components (digital platform) and another set of evolving components that allow variation and innovation (third-party contributions); interfaces set standards for how to interact with the platform; platforms lower barrier of entry and foster experimentation; digital platforms pose issues of power and autonomy

Baldwin and Woodard 2009; Boudreau 2012; Tiwana et al. 2010

18

Digital artifacts as enablers and facilitators of innovation

Democratization of innovation as consequence of pervasive digitization; recombination of existing resources drives innovation; heterogeneous groups of actors provide varied innovation capability and knowledge resources

Arthur 2009; Boland et al. 2007; von Hippel 2005; Yoo 2010

20

Decomposition of complex systems, logic of modularity

Isomorphism of product design logic and organizational structure; decomposition into design hierarchies; modularity as organizing logic for complex systems; stable interfaces enable concurrent design of sub-components and reduce dependency

Baldwin and Clark 2000; Clark 1985; Simon 1996

14

Structuration theory applied to technology

Flexible interpretation of digital artifacts by their users; digital artifacts are both shaped by their context and shape their context

Orlikowski 1992

4

IT-enabled organizational agility

Digital artifacts generate real options which facilitate change; malleability of digital artifacts enable organizational agility

Sambamurthy et al. 2003

4

I.5.2 Digital infrastructures

In the literature, digital infrastructures (or information infrastructures, as they are more

commonly called) are regarded as socio-technical systems, hence they consist of more

than technology components (Ciborra 2000; Star and Ruhleder 1996). They are

networked systems that span beyond and across individual organizations and may gain


global reach (Ciborra 2000). Hanseth and Lyytinen (2010: 4) define these

infrastructures, of which the internet is a prime exemplar, as follows:

“A shared, open (and unbounded), heterogeneous and evolving socio-technical system

(...) consisting of a set of IT capabilities and their user, operations and design

communities. (...) Structurally a (digital infrastructure) is recursively composed of other

infrastructures, platforms, application and IT capabilities. (...) Control is distributed

and episodic and an outcome of negotiation and shared agreements. (...) Episodic forms

of control determine which groups of designers control which parts or elements of the

(digital infrastructure). (...) There are no clear boundaries between those that can

design the (infrastructure) and those that may not. (...) The openness (...) implies that

during their lifetime the social and technical diversity and heterogeneity of (digital

infrastructures) will increase.”

Digital infrastructures depend on the active involvement of heterogeneous groups of

actors in using, operating, and designing them. Zittrain captures this dependency as

“invitation to outside contribution” (2008: 90) and sheds light to the self-reinforcing

dynamics of active infrastructures: the more actors contribute in using, operating, and

designing, the greater the generative capacity of this system becomes, which in turn will

lead to more, and more varied unanticipated evolution and innovation. Hanseth and

Lyytinen (2010) call this central aspect dynamic complexity, which the authors regard

to be conceptually close to generativity. Evolution and innovation of digital

infrastructures is non-linear, typically initiated to suit specific needs of a local group of

actors (Star and Ruhleder 1996), which is why infrastructures cannot be controlled top-

down, and over time their development paths might deviate from originally planned

purposes (Ciborra 2000). Given the importance of attracting actors to participate in a

digital infrastructure, scholars explored key mechanisms how to achieve this goal. One

such mechanism is bootstrapping, the tactic of attracting early users by making the

infrastructure immediately useful for their specific needs, while deliberately neglecting

long-term issues such as architectural robustness if so required (Grisot et al. 2014;

Hanseth and Aanestad 2003; Hanseth and Lyytinen 2010). Bootstrapping can be

regarded as an entrepreneurial approach to making the most out of limited available

resources and letting serendipitous design activity ultimately lead to creative innovation

(cf. Fisher 2012; Ries 2011).

The self-reinforcing dynamics of digital infrastructures can only unfold if there is a

sufficiently large and structurally stable installed base, that is all the individual elements

and their connections making up the digital infrastructure (Star and Ruhleder 1996;


Tilson et al. 2010). Without stability in the installed base extensions and additions are

not possible, hence setting a standard early on is indispensable for continued growth of

such systems and the generation of network effects (cf. Shapiro and Varian 1999).

Setting a standard, however, may dampen the possibility for emergent design, may

create unwanted path dependencies, and thus may limit creative innovation (Racherla

and Mandviwalla 2013; Star and Ruhleder 1996), a tension which Tilson et al. (2010)

call the paradoxes of change and control.

By discussing the design of the internet, in particular its enforcement of the internet

protocol (IP) as the sole standard for data transmission while leaving myriad options on

all other architectural layers, Zittrain (2008) suggests the procrastination principle as

one way to overcome this tension: the elements on which all actors rely should not solve

design problems that affect just some of them. This concept resembles Benkler’s (2006)

argument of instilling intelligence at the endpoints, not the center of an infrastructure.

Henningsson and Henriksen (2011) pick up these ideas to conclude that a digital

infrastructure should inscribe just a few commonly accepted regulations, but otherwise

be open to interpretation, that is ambiguity and emergent evolution in its design and

usage. This openness is consistently named a key attribute of digital infrastructures

(Grisot et al. 2014; Hanseth and Lyytinen 2010; Henningsson and Henriksen 2011;

Racherla and Mandviwalla 2013; Tilson et al. 2010) and can be related to earlier ideas

of tinkering/bricolage (Ciborra 1991) and improvisation (Orlikowski 1996).

I.5.3 Digital platforms

Again, we refer to Hanseth and Lyytinen (2010: 4) and conceptualize digital platforms

as follows:

“Platform designs (...) organize IT capabilities into frameworks allowing the software

to address a family of generic functional specifications that meet the needs of multiple,

heterogeneous and growing user communities. (...) Platforms typically grow in

complexity as designers take into account heterogeneous user needs while maintaining

backward compatibility and horizontal compatibility across different combinations of

capabilities. Therefore, many platforms, originally conceived as limited sets of IT

capabilities, obtain later emergent features; they start growing in seemingly unlimited

fashion and serve unexpected users (…) generating exponentially growing technical and

social complexity.”

The boundaries between a platform and an infrastructure are fluid. For instance, what

Grisot et al. (2014) call an infrastructure in their article might also fit above definition


of a platform. For the purpose of this paper, and in line with Hanseth and Lyytinen

(2010), we delineate the former from the latter by demanding a digital platform to be

controlled by one single actor, while control of a digital infrastructure is distributed

across many actors. Identifying a platform-controlling actor consequently introduces

third-party actors which leverage the platform for pursuing their own goals. A relatively

stable, centrally controlled digital platform in conjunction with the variety of evolving

applications and services provided by third-party actors compounds the overall platform

ecosystem (Baldwin and Woodard 2009; Boudreau 2012; Lusch and Nambisan 2015;

Tiwana et al. 2010).

Upon first thought it might seem odd to contemplate the generative capacity of a system

whose central element–the digital platform–is controlled by one single actor. However,

several researchers showed that nurturing unexpected change brought by heterogeneous

groups of actors ultimately serves the interest of all parties: the platform-controlling

actor needs an attractive ecosystem of applications and services in order to keep up

effectively with competing ecosystems; and third-party actors want to distribute their

applications and offer their services within a thriving ecosystem because they can reach

broader audiences than they otherwise could (Bergvall-Kåreborn and Howcroft 2014;

Eaton et al. 2015; Ghazawneh and Henfridsson 2013). Hence, the platform-controlling

actor has a strong incentive to design and keep its platform as open and malleable as

possible (Eaton et al. 2015; Ghazawneh and Henfridsson 2013), while third-party actors

engage in ecosystems from which they can expect to profit most (Bergvall-Kåreborn

and Howcroft 2014; Selander et al. 2013).

Still, the platform holder exerts considerable control, mainly through software tools and

accompanying regulations that expose the digital platform to third-party actors, that is

through platform boundary resources such as APIs and associated governance

(Ghazawneh and Henfridsson 2013; Tiwana et al. 2010). Designing, maintaining, and

evolving boundary resources confronts the platform-controlling actor with challenges

similar to the paradox of change and control mentioned in the discussion on digital

infrastructures (Baldwin and Woodard 2009; Tiwana et al. 2010). What is more, it is

likely that inventive third-party actors will utilize boundary resources in unanticipated

ways or that they will create their own, unauthorized boundary resources to exploit

platform features that were not intended to be accessible. For instance, when Apple

introduced its iPhone device and accompanying digital platform, third-party actors soon

found ways to jailbreak it, thus creating a way to open up the platform for outside

development (Eaton et al. 2015; Ghazawneh and Henfridsson 2013). Both Eaton et al.


(2015) and Ghazawneh and Henfridsson (2013) argue that permanent loops of emerging

tensions between platform-controlling actor and third-party actors and their dialectic

resolution in the interest of all parties keep a platform and the broader ecosystem

relevant and lead to unexpected, creative, and innovative outcomes.

Some researchers explored aspects of what von Hippel (2005) calls the democratization

of innovation. Digital platforms empower individuals or very small organizations to

innovate, in that they provide powerful means to come up with new things and offer a

large audience for experimentation (Boudreau 2012). Eaton et al. (2015) and Bergvall-

Kåreborn and Howcroft (2014) remark that such democratization has only become

possible because there are a few large and very resourceful companies that design and

nurture digital platforms, pointing to apparent imbalances on both ends of the

ecosystem, which might lead to unilateral abuse of power. The authors also observe that

even among third-party actors, discrepancies with regard to their size, capabilities, and

other facets are tremendous. This accentuates the heterogeneity of actors so central to

generativity, but also indicates that there will be winners and losers within each

ecosystem (Tiwana et al. 2010). Selander et al. (2013) emphasize that even large

organizations have good reasons to participate in digital platforms, mostly in order to

gain access to capabilities and other resources which they could hardly obtain otherwise.

Finally, Selander et al. (2013) and Yoo et al. (2010) discuss why large organizations

might be motivated to design and nurture digital platforms in the first place. They

suggest that the modular-layered architecture of digital technology unlocks vast

possibilities for resources recombination and innovation (cf. Arthur 2009; Lusch and

Nambisan 2015). Due to the separation of services from devices–through the possibility

to reprogram and repurpose digital artifacts–and the separation of content from the

transport medium–through digitization of data–the authors argue that it makes sense to

connect heterogeneous actors via a common platform, even accepting the tensions that

arise from conflicting agendas: the joint innovative outcomes within an ecosystem will

most likely be superior in quality and quantity than what an organization would be able

to achieve in isolation (cf. Chesbrough 2003).

I.5.4 Digital artifacts, innovation, and modularity

Interestingly, there is no widely accepted definition of a digital artifact, and indeed it

may be doubted that this term is useful at all (Alter 2015). Nevertheless and for the

limited purpose of providing an overview of past research on generativity we suggest

the following working definition, borrowed from Orlikowski’s (1992) discussion of the


technology concept and combined with the conceptualization of distinct digital artifact

attributes (Kallinikos et al. 2013):

A digital artifact is an object created by and composed of digital technology and the

outcome of coordinated human action. It is created and changed by human actors, but

it is also used by humans to accomplish some action. Digital artifacts fundamentally

differ from physical artifacts in that they are interactive, editable, reprogrammable,

distributed, modular, granular, and reflexive.

Researchers arguing why systems interspersed with digital artifacts have greater

generative capacity and produce more innovative outcomes than systems consisting

solely of physical artifacts highlight how the first differ from the latter. Henfridsson et

al. (2014) focus on the two aspects of reprogrammability and negligent marginal cost of

replication and suggest two architectural logics for the physical and digital realm,

respectively: a physical artifact is organized in a hierarchy of parts to cope with

complexity in design and production, but at the cost of fixed boundaries. Digital artifacts

however profit from a network of patterns approach, in which design patterns are

combined and applied to a specific context. By changing the context and adapting the

design accordingly, digital artifacts can be repurposed or their functionality can be

enhanced. Arthur (2009) reasons that the extent to which existing resources can be

readily recombined drives innovation, and that digital artifacts are catalysts of

innovation. Nambisan (2013) spins this thought further and argues that in principle

digital artifacts can be operant resources and trigger innovation on their own, through

seeking variation and recombination autonomously. This idea, however, will need

further reconciliation with the socio-technical tradition of IS research. For instance,

Boland et al. (2007) highlight that it is not technology which drives innovation, but

rather its accessibility to heterogeneous groups of actors, which employ their varied

capabilities and come up with innovative results. This train of thought is very close to

the reasoning of Zittrain (2008), who regards participation as input to generativity and

innovation as its output.

Following the isomorphism argument (Baldwin and Clark 2000), Henfridsson et al.

(2014) posit that an organization that designs and produces physical artifacts will be

structured along hierarchical components, whereas an organization dealing with digital

artifacts will be structured along functional patterns. On similar terms, Yoo (2013)

argues that organizations cannot leverage the full potential of digital artifacts by sticking

to the logic of modularity and the organizational implications it entails. The author

suggests that portraying digital artifacts as parts of generative systems accentuates their


capacity to enable emergent change through variation and recombination. Yoo’s

propositions are in their infancy and will require further clarification in light of the thick

research threads around modularity. Ever since the characterization of complex systems

as “ones made up of a large number of parts interacting in a nonsimple way” (Simon

1962: 468), their decomposition into design hierarchies has been discussed extensively

(Baldwin and Clark 2000; Clark 1985; Simon 1996). This discourse led to ideas of

modularity as organizing logic for complex systems and of stable interfaces enabling

concurrent design of sub-components (Baldwin and Clark 2000), and left myriad traces

in IS/IT research, for instance service-oriented architecture (Erl 2005) and the modular-

layered architecture proposed by Yoo et al. (2010).

I.6 Conclusion and further research

With this contribution we set out to describe how Zittrain’s generativity concept has

been brought into IS research, and in particular in relation to which established scholarly

discourses the emerging topic has been discussed. To this end, we first clarified

terminological issues and identified a broad range of generativity concepts from 52

articles published in leading IS journals. For the subsequent analysis we focused on

Zittrain’s conceptualization of generativity as the malleability of technologies and

systems by heterogeneous groups of actors with unanticipated outcomes. From an

initial, illustrative analysis of how the distinct attributes of digital artifacts may be

conducive to the generative capacity of socio-technical systems in which they operate,

we concluded that myriad types of artifacts may instill generativity. However, we found

that the reviewed sample of 14 articles concentrated on just two types of digital artifacts,

namely digital infrastructures and digital platforms. Besides these main strands we

detected that generative systems were regularly discussed as means to lead towards

innovation, brought by heterogeneous groups of actors. Finally, we identified that

discussions of generativity were commonly held in light of the rich intellectual tradition

pertaining complex systems and the logic of modularity. Overall, we regard this paper

as a contribution to generativity research as sketched by Tilson et al. (2010), Yoo et al.

(2010), and Yoo (2013). When a concept from a different scholarly field is adopted, it

is important that we first gain a clear grasp of the terminology, the phenomenon, and the

context in which it is discussed before setting out to conduct further research.

Before highlighting possible further research, we would like to acknowledge the main

limitation of this contribution, namely its lack of exhaustiveness with regard to the

reviewed literature. We confined ourselves to articles published in leading IS journals,


leaving out many interesting papers. For instance, Zhang et al. (2014) outlined how the

transferability characteristic of generative systems–which the authors call generative

diffusion–can be explored in the context of open source software development. There

are many more relevant articles available, as a database search will reveal. Nevertheless,

we believe that despite the self-imposed limitation this paper still yields considerable

value. First, articles published in the leading IS outlets are influential in shaping overall

research in the field (Lowry et al. 2013). And second, it is likely that scholars interested

in the state of the art of generativity research within the IS community will first seek to

identify exactly those articles which we selected and described in this review. Still, we

caution the reader to consider this major limitation.

Turning to possible avenues for further research, the results of this review article draw

attention to four topics in particular that could stimulate worthwhile research. First, we

do not understand sufficiently well the overlaps and differences of modularity and

generativity. Yoo (2013) put this challenge into the limelight, and prior research

regularly discussed both concepts in conjunction. What are the parts which we can

transfer to generativity, how does generativity really differ and how does it lead to

different results? These are all questions yet to be answered, but first results indicate

that there can be fruitful cross-pollination and that the logics of modularity and

generativity may lead to different organizational regimes (Henfridsson et al. 2014; Lee

and Berente 2012; Svahn and Henfridsson 2012).

Second, the variety of examined types of digital artifacts might be expanded. Prior

research discussed the generative capacity of digital infrastructures and digital platforms

only. In our introductory section we argued that it is the distinct characteristics of digital

artifacts (Kallinikos et al. 2013) that may instill generativity in a system, hence there

should be other types of digital artifacts besides infrastructures and platforms to which

the generativity concept can be valuably applied. One way forward might be the

examination of the attributes of digital artifacts (e.g., reprogrammability,

distributedness) and how they are favorable with regard to the attributes of generativity

(e.g., adaptability, transferability). We refer to the research of Henfridsson et al. (2014)

and Zhang et al. (2014) for exemplary steps onto this path.

Third, in light of the reviewed literature we were not very precise in delineating digital

artifacts from socio-technical systems. On the one hand, we posited that digital artifacts

as elements of a system may instill generativity. On the other hand, we gave definitions

of digital infrastructures, digital platforms, and digital artifacts in general that

highlighted their socio-technical nature. This lack of conceptual clarity might be tackled


by expressing digital artifacts and their generative capacity with the vocabulary of

sociomateriality (Leonardi 2012; Orlikowski 2000). For instance, recent research from

Woodard and Clemons (2014) already points into this direction.

And fourth, throughout this paper we discussed generativity as means towards

innovation. However, unanticipated outcomes do not have to be innovations, there

might be other worthwhile ends to be considered. For example, generative capacity

might be exploited for organizational agility purposes, that is to better sense

environmental change and respond to it (Goldman et al. 1995; Sambamurthy et al.

2003). Some traces towards alternative ends of generativity can already be found in IS

research (e.g., Kretzer et al. 2014; Svahn and Henfridsson 2012).

To conclude, IS research on generativity has just begun. From reviewing the literature

we concur with Yoo (2013) to believe that the intellectual tradition of IS research

combined with the novel lens of generativity can bring much to the table in providing

valuable insights to information and technology management, but also in advancing our

knowledge of socio-technical systems in a digitized world.

Part B: Untangling generativity: two perspectives on unanticipated change produced by diverse actors 47

II Untangling generativity: two perspectives on

unanticipated change produced by diverse actors

Table 15. Bibliographic information for Article II

Title Untangling Generativity: Two Perspectives on Unanticipated Change Produced by Diverse Actors


Outlet ECIS 2016 Proceedings

Year 2016

Status Published

Abstract. Digital platform ecosystems capitalize on the engagement of large groups of

actors with diverse skills that create unexpected services, find novel uses, and move the

ecosystem forward in unanticipated ways. The generativity concept captures this

phenomenon, which some scholars regard as fundamental to our understanding of how

digital innovation plays out. Despite such claims, it is unclear what generativity means

and how it manifests itself. After scrutinizing the foundational definition, we delineate

two perspectives on generativity. The ‘generative properties’ perspective asks which

properties of digital artifacts embedded in social structure invite actors to set free their

creativity and produce unanticipated outcomes. In contrast, the ‘generative patterns’

perspective asks which patterns of events lead to an evolutionary dynamic that produces

unanticipated change. We find cues for both perspectives in literature, which tentatively

validates them. We then formulate a socio-technical model of digital platform

ecosystems and describe ecosystem generativity by applying both perspectives. In this

context, the first perspective highlights the digital platform and its controlling

orchestrator, while the second perspective directs our attention to interactions among

actors and artifacts. We conclude this paper by discussing how generativity can help

explain digital innovation.

Keywords: generativity, digital innovation, digital platform, ecosystem, socio-

technical system

48 Part B: Untangling generativity: two perspectives on unanticipated change produced by diverse actors

II.1 Introduction

Digital innovation, or innovating with digital technologies, arguably becomes a

dominant force as physical artifacts are being augmented with or completely replaced

by digital counterparts (Lyytinen and Yoo 2002). For instance, about 80% of current

innovations in cars are credited to novel uses of digital technologies (Mossinger 2010).

Digital innovation seems to follow a different logic than traditional innovation regimes

(Svahn 2012; Yoo et al. 2010). For example, engaging with digital artifacts is usually

more flexible and more affordable than tinkering with physical artifacts, which has

‘democratized’ access to technology (von Hippel 2005). In addition, digital artifacts are

created, assembled, accessed, or otherwise manipulated exclusively through other

digital artifacts. Such ‘reflexivity’ (Kallinikos et al. 2013) creates a dynamic that sees

increased permeation of digital artifacts in any domain where digital technologies are

introduced (Boland et al. 2007). Democratization and reflexivity create a competitive

environment that relies on a torrent of combinative innovations (cf. Arthur 2009)

contributed by self-organizing and diverse groups of actors (cf. Benkler 2006). A

growing number of information systems (IS) scholars regard the ‘generativity’ concept

as suitable to capture essential aspects of digital innovation (Tilson et al. 2010; Yoo et

al. 2010), by formulating that a generative system is characterized by diverse groups of

actors producing unanticipated changes (Zittrain 2006, 2008).

Yoo (2013) went even so far as to put generativity on par with the concept of modularity

(Baldwin and Clark 2000; Parnas 1972; Simon 1996), which brought us product

hierarchies, loose coupling, and object-oriented programming, to name a few

accomplishments. Modularity is a powerful principle in the effective design and

production of very complex artifacts (Simon 1962). If generativity is to be compared

with modularity, it will need to provide similarly authoritative guidance on artifact

design and evolution. However, there is no consensus on what the nature of generativity

is and how it manifests itself. Is generativity a quality, i.e. can an artifact be more

generative than another one? Is it a relational construct, that is does it emerge in one

system through interactions of its elements, while it does not emerge in another? Or is

it something different altogether? If we cannot answer these questions clearly, there is

no point trying to design ‘for generativity’ or ‘towards generativity’ – because we do

not know which design objectives to aim for.

We believe this uncertainty is largely attributable to ambiguity in the generativity

concept itself. The purpose of this article is to clarify this conceptual fuzziness. To

facilitate the discussion, we frame it around digital platform ecosystems, or systems in


which various actors and artifacts are organized around a central digital platform. We

chose digital platform ecosystems because of two main reasons: first, many narratives

of digital innovation revolve around digital platforms and the ecosystems they create

(Baldwin and Woodard 2009; Iansiti and Levien 2004). Second, digital platform

ecosystems have arguably been at the heart of generativity research (e.g., Woodard and

Clemons 2014). Therefore, we ask:

How can we describe generativity in a digital platform ecosystem?

This question breaks down into two goals. The first goal is to gain a more precise

understanding of generativity. Through interpreting the foundational definition, we

propose two distinct perspectives on the concept: one in which a system is generative

due to its characteristics, and one in which a system is generative due to its evolution.

To test this proposition, we inquire salient literature on how prevalent the two

perspectives are. The second goal is to apply this newly gained understanding to digital

platform ecosystems. The ensuing discussion reveals how the first perspective on

generativity draws our attention to the central digital platform and its governance, while

the second perspective emphasizes interactions between individual ecosystem entities.

These results underscore the merits of distinguishing both perspectives. The article

concludes with directions for further research on generativity and digital innovation, and

a discussion of key limitations.

II.2 Two perspectives on generativity

In its broadest sense generativity refers to the ability or function of producing or giving

rise to something (Oxford English Dictionary 2009). Unsurprisingly, the term has been

attached to a large number of distinct concepts within and outside of IS research (Avital

and Te'eni 2009; Eck et al. 2015). The concept discussed here is that of Zittrain (2006,

2008), not necessarily because of its clarity – indeed it allows starkly different

interpretations as we will see – but because of its exposure among IS scholars sparked

by two calls for research (Tilson et al. 2010; Yoo et al. 2010). Zittrain (2008: 70) defines

generativity as “a system’s capacity to produce unanticipated change through unfiltered

contributions from broad and varied audiences”. Let us analyze this definition.

A ‘system’ refers to any meaningful set of interacting elemental parts. While Zittrain

had primarily technology in mind (cf. Zittrain 2006), to our reading a system comprises

both social actors and technical artifacts, it is socio-technical (Bostrom and Heinen

1977). The actors engaging with the system’s artifacts are ‘broad and varied audiences’.


This means that a priori undefined actors of varying capabilities, from the curious

individual all the way up to the mightiest organization, may participate in a generative

system (cf. Zittrain 2008). What is more, their participation is ‘unfiltered’, that is they

act autonomously, and their activities are not centrally controlled. A generative system

therefore is designed to let people engage creatively with technology, and this may lead

to unanticipated change. Or as Zittrain (2008) puts it, a generative system transforms

participatory input into unanticipated change.

‘Unanticipated change’ suggests that generativity is a relational construct: to qualify

change as ‘unanticipated’, we need to specify who is surprised by what (or by whom).

Literature usually identifies the original artifact designers as those being surprised (e.g.,

Woodard and Clemons 2014; Yoo 2013). The surprise element might be the outcome of

creative activity of individual actors (e.g., Elaluf-Calderwood et al. 2011), or it might

be the result of serendipitous collective behavior that is beyond any one actor’s control

(e.g., Woodard and Clemons 2014). In the first case, the system’s main achievement is

to catalyze human ingenuity; in the second, the system is capable to evolve into

directions that were initially unimaginable – it is perpetually in the making.

We argue that this distinction offers two perspectives on generativity, which lead to

different conclusions on how to describe a generative system. As Figure 6 schematically

illustrates, the first perspective regards generativity as consequence of system design,

while the second perspective regards generativity as consequence of system evolution.

In what follows, we strive to delineate each perspective in turn.

Figure 6. Generativity as consequence of system design, and as consequence of system evolution

II.2.1 Generative properties: generativity as consequence of system design

The first perspective on generativity regards a system’s capacity to churn out a variety

of novel products or to be put to a multitude of surprising uses. The system consists of

actors which contribute with their creativity and skills, and of suitable artifacts that help

those actors accomplish their goals, but largely remain unaffected. Equally important,

artifact owners impose few limitations on what third party actors can do, that is

governance arrangements invite outside contribution (cf. Tiwana et al. 2013). For


example, people engage with the Eclipse software development suite to create new

software, which subsequently causes change somewhere else. The Eclipse suite,

meanwhile, remains pristine and can be utilized for the next development project.

Another example is the Kuka Quantec, an industrial robot. It is programmable along six

degrees of freedom and can be equipped with a host of ancillary sensors and actuators.

The robot can be told to weld, paint, cut and do other useful things in a workshop. But

people have also made it dance, play table tennis, write with a calligraphy pen, and found

numerous other unanticipated use contexts. The robot itself does not change, as erasing

the machine instructions is all it takes to reset it to its original state. Yet another example

is OpenStreetMap whose application programming interfaces (APIs) allow third parties

to harness cartographic data for innumerable, and partially unexpected, purposes. For

example, there are interactive world maps which resemble watercolor paintings, or maps

that visualize all geocoded Wikipedia articles. However, these applications do not cause

any change in OpenStreetMap’s cartographic data nor in its APIs.

These examples fulfil the definition of generativity. There is a system in which a priori

unknown actors creatively engage with artifacts to produce results that the original

artifact designers did not have in mind initially. But these results, as unanticipated as

they may be, have no effect on the artifacts with which they have been produced.

Through artifact design and governance arrangement, the system is generative because

of some inherent qualities that it possesses – it is a toolbox for unanticipated change.

Our view then turns to uncovering which properties enable and foster generativity. For

example, we may note that Eclipse supports many different programming languages, or

that anybody can install it free of charge. These properties make Eclipse applicable in a

broad range of tasks and accessible to broad audiences (cf. Zittrain 2008).

Generative capacity then is a function of the inherent generative properties of a system

and the number and diversity of actors that can potentially engage with its artifacts.

Moreover, we can compare the relative degree of generative capacity between systems

by comparing their properties. For instance, we may plausibly claim that Eclipse is more

generative (or less restrictive) than Visual Studio, because this latter software

development suite supports fewer programming languages and usage must be paid for.

Through gradual systematization of such properties, we can aim to distill generative

properties which increase a system’s generative capacity across a broad range of

contexts: given artifact and governance properties X, the system is generative. This

knowledge may guide how to design artifacts and associated social structures for

generativity. For example, modularity might be identified as a generative property,


which may explain why the Eclipse suite is architected around a central platform to

which various modules can be easily attached.

II.2.2 Generative patterns: generativity as consequence of system evolution

The second perspective on a generative system greatly differs to the one discussed

before. This perspective regards how and to which extent a system develops elements,

structures and behaviors that surpass the imagination or ambition of its original

designers. Again taking Eclipse as example, we can observe how its design community

engages with the Eclipse software artifact to extend and improve it in myriad ways, with

an evolutionary dynamic that has likely eclipsed the originators’ expectation of some 15

years ago. Similarly, the lineage of the Kuka Quantec can be traced back to the early

1970s. Since then, industrial robots (and organizations around them) have co-evolved

with advances in areas such as electronics, computing, and operations management to a

level of sophistication which was unimaginable a few decades back. And

OpenStreetMap, founded by an individual some ten years ago to create openly

accessible maps of Great Britain, has since attracted volunteers that contribute with map

data from all over the world. Additionally, the technology with which its data is

collected, stored, maintained, and distributed has undergone numerous extensions,

replacements, and optimizations.

As before, these examples fulfill the definition of generativity, albeit in a remarkably

different way. Instead of inherent characteristics we regard time as evocative of

generativity. Here, we emphasize the evolutionary path a system has taken. The system

has evolved generatively because the many interactions between actors and artifacts in

aggregate have produced unanticipated changes. Our view then focuses on identifying

events that sparked and cultivated generativity in a specific context. For instance, we

may note that after governance of the Eclipse ecosystem was handed from IBM, which

started Eclipse, to Eclipse Foundation, a nonprofit organization, adoption of the Eclipse

suite soared. Or we might become interested into why development of the Eclipse

platform is clocked around one major platform release per year, and how this fixed

schedule affects change processes in peripheral parts of the ecosystem.

Yet, past accomplishments of system changes beyond the original designers’

imagination are no guarantee of generative evolution in the future. We cannot simply

assume that systems continue to evolve generatively on the sole basis of their generative

history. The evolutionary path can narrow into a state of lock-in (Sydow et al. 2009),

wherein all change would become predictable and thus the trajectory would turn towards


stagnation or even degeneration. Therefore, the true challenge is to detect generative

patterns that underlie empirical events and which contingently explain generative

evolution across a broader range of contexts: an order of events that reflects pattern X

leads the system to evolve generatively. This knowledge may guide how to increase the

probability of continued generative dynamics (cf. Yoo 2013). For instance, we might

discover that heterogeneity breeds conflict and tensions, whose resolution may lead to

unexpected, yet productive outcomes (Follett 2013; Lee and Cole 2003). This pattern

might explain why IBM released control over the Eclipse ecosystem, why the governing

Eclipse Foundation was set up as nonprofit organization supported by a consortium of

commercial firms, and why the foundation promotes numerous projects loosely linked

with the core Eclipse product.

II.3 Generative properties and generative patterns in literature

Derived from the definition of generativity as “a system’s capacity to produce

unanticipated change through unfiltered contributions from broad and varied

audiences” (Zittrain 2008) we argue that there are two distinct perspectives on a

generative system. The first perspective asks which system properties invite social

actors to set free their creativity and produce unanticipated outcomes. The second

perspective, in contrast, asks which patterns of interactions among actors and artifacts

lead to an evolutionary dynamic that produces unanticipated change.

With an interest to learn how prevalent each perspective is in salient research, and to

which extent both perspectives have been distinguished, we reviewed a selection of

literature. We started with a keyword search for ‘generativity OR generative’ in the

‘basket of eight’ journals (AIS 2011), Organization Science, ICIS and ECIS

Proceedings, carried out on November 01, 2015. Of 549 hits, 50 articles explicitly

mentioned Zittrain’s definition of generativity. Out of those, 20 articles were selected

for further analysis on the basis that generativity was not only briefly mentioned, but an

important part of the paper.

We distilled how the selected papers described generativity as follows: First, all

paragraphs in which any of the keywords occurred were extracted via software, which

yielded 466 data points. Next, we manually retained those text passages which discussed

an aspect of generativity. For example, Bygstad (2015) defines generativity as “ability

of technical and social elements to interact and recombine to produce new solutions”

(ibid.: 3), which we kept as data point, while we did not include the research question

“How is generativity different in heavyweight and lightweight IT?” (ibid.: 3). On


average, we retained 5.5 text excerpts per paper (σ = 1.9) for a total of 109 data points.

For each paper, we continued by summarizing the selected citations and condensing

them to a short statement. 2 papers shared the same description, which we collapsed

(Zhang et al. 2012; Zhang et al. 2014). We found that 5 papers assumed both

perspectives, which led us to split their descriptions along this line (Bygstad 2015; Eaton

et al. 2015; Tilson et al. 2010; Yoo et al. 2010; Yoo 2013). In total, the literature sample

provided 24 accounts on what generativity is and how it shows itself, which we ordered

into 8 thematic groups. Table 16 gives an overview of the review results, while the full

summary tables are to be found in the appendix to this paper.

Table 16. Descriptions of generativity in IS literature

# Aspect of ‘generative properties’ perspective Sources

1A Generativity is a consequence of combinatory capacity and reprogrammability of digital artifacts

Eaton et al. 2015; Tilson et al. 2010; Um et al. 2013; Yoo 2013

1B Generativity is a socio-technical phenomenon, wherein digital platforms attract external contributions

Bygstad 2015; Jain and Ramesh 2015; Nielsen and Hanseth 2010

1C Generative capacity of digital artifacts is leveraged through creativity and diverse skills of outside parties

Förderer et al. 2014; Jackson 2015; Vassilakopoulou and Grisot 2012

1D Generativity can and should be governed by the platform-controlling actor

Jarvenpaa and Tuunainen 2013; Thorseng and Jensen 2015; Wareham et al. 2014; Yoo et al. 2012

# Aspect of ‘generative patterns’ perspective Sources

2A Generativity is serendipitous system evolution through emergent change

Bygstad 2015; Woodard and Clemons 2014

2B Generative evolution is caused by high diversity in space and in skills among involved actors and artifacts

Lyytinen et al. 2016; Yoo 2013; Zhang et al. 2012; Zhang et al. 2014

2C Generativity thrives on knock-on effects wherein a local change event leads to change in other system parts

Yoo et al. 2010; Yoo et al. 2012

2D Generative dynamics cause tensions which must be dealt with

Eaton et al. 2015; Tilson et al. 2010; Venters et al. 2014

As for descriptions of generativity as consequence of system design, we identified four

thematic groups. The first group (1A) regards generativity from a technology viewpoint.

These descriptions emphasize combinatory capacity and reprogrammability of digital

artifacts, but neglect to explicate by whom these generative properties are actualized. In

contrast, the second group (1B) describes generativity as inherently socio-technical

phenomenon, in which social actors engage with technology. These descriptions identify

digital platform architectures as particularly useful to attract varied contributions from

outside parties. Meanwhile, the third group (1C) emphasizes the importance of social


actors’ local knowledge, diverse skills, and their creativity in producing unanticipated

outcomes. Finally, the fourth group (1D) brings in another aspect, namely that the

creation of unexpected changes can be steered, usually by the platform-controlling actor.

This highlights the governance dimension of the generative properties perspective.

With regard to descriptions of generativity as consequence of system evolution, we

identified another four groups. The first group (2A) captures the idea of emergent system

change and the serendipity of this evolution. The second group (2B) specifies two causes

for generative evolution. First, interactions among actors and artifacts are distributed

and uncoordinated. Second, the diversity of involved actors is assumed to be high.

Together, distribution in space and distribution in skills create a constellation in which

the system changes in surprising ways. The third group (2C) adds to this thought by

emphasizing possible knock-on effects of unanticipated change, wherein an episode of

innovative change in one part of the system can lead to follow-up innovations in other

parts of the system. Lastly, the fourth group (2D) highlights that change breeds tensions,

which require adequate resolution. Far from offering clear-cut advice, these descriptions

rather imply that tensions go hand in hand with generativity.

Five papers regard both generative properties and generative patterns, yet do not

distinguish them clearly. Take Eaton’s et al. (2015) excellent study on how tensions play

out in digital platform ecosystems as example. The authors describe how a digital

platform that is inviting to third-party developers is generative (ibid.: 220). At the same

time, they conclude that the resolution of emerging tensions in the ecosystem leads to

generative evolution (ibid.: 238). In the first instance generativity is portrayed as

inherent system quality, while in the second instance generativity is presented as

temporal phenomenon. This difference makes sense in light of the two perspectives on

generativity we propose here, but might confuse a reader otherwise.

In conclusion, we found support for both the generative properties and generative

patterns perspectives. Crucially, we were able to categorize all descriptions of

generativity in the selected literature with these two perspectives, a result which we

regard as preliminary validation of their appropriateness. We also identified five papers

that did not present the generativity concept consistently. This result indicates that

generativity is not yet well understood, or at least that its foundational definition leaves

ample room for varied interpretations. The generative properties and generative patterns

perspectives developed here make two such interpretations explicit. The following

section shows how each perspective lends itself to a different approach of studying

generativity in the context of digital platform ecosystems.


II.4 Generativity in digital platform ecosystems

It is the main argument of this paper that generativity can be studied in terms of

generative properties and in terms of generative patterns. Regardless of the perspective

taken, the implicit assumption has been that a generative system is one in which many

elements interact in non-trivial ways, that is it is a complex system (Simon 1962).

Indeed, the sampled literature regards generativity in complex systems in general (6

articles), in information infrastructures (4 articles) which are immensely complex

systems (Hanseth and Lyytinen 2010), and in digital platform ecosystems (10 articles)

which are also complex systems (Tiwana et al. 2010). Despite its complexity this last

category features a simple generic topology: there is a central digital platform around

which the whole system is arranged. This particular topology allows us to bring some

order in complexity, which in turn lets us discuss generativity with higher granularity

than before. In particular, we find that the generative properties perspective highlights

the central role of the digital platform and its controlling actor, while the generative

patterns perspective emphasizes individual behavior and interactions which in aggregate

lead to generative evolution.

II.4.1 Modeling a digital platform ecosystem

The term ‘ecosystem’ carries the notion of a breathing, living system, in which a large

number of heterogeneous elements are delicately balanced through their mutual

dependencies and interactions (Iansiti and Levien 2004). In IS discourses the ecosystem

analogy has been attached preferably to complex systems whose nucleus is a digital

platform, a rather stable bundle of commonly used digital functionality that can be easily

coupled with complementary functionalities, or modules. Operating systems and the

many artifacts and actors around them (e.g., Apple iOS, Microsoft Windows) come

readily to mind as apt examples of digital platform ecosystems. Apart of the term applied

here, such a topology is called digital ecosystem (Um et al. 2013), platform ecosystem

(Tiwana et al. 2010), software ecosystem (Manikas and Hansen 2013), or technology

ecosystem (Wareham et al. 2014) in literature.

There are many possible ways to model a digital platform ecosystem (Manikas and

Hansen 2013). In elaborating the model presented in Figure 7, it was important to us to

consider both artifacts and actors, because we regard generativity to be a socio-technical

phenomenon. To this end, we took the widely accepted model of Tiwana et al. (2010)

as solid foundation and extended it with social actors. Furthermore, we noticed that it is

helpful to distinguish between those adding functionality and those putting it to


productive use in their local context. Consequently, we model a digital platform

ecosystem as socio-technical system consisting of six elements: digital platform,

module, composition, orchestrator, developer, and user. These six elements can be

partitioned into three pairs of corresponding actor-artifact combinations, or three types

of socio-technical subsystems: the orchestrator-platform subsystem, the developer-

module subsystem, and the user-composition subsystem.

Figure 7. Socio-technical model of a digital platform ecosystem

A digital platform organizes commonly used functionality into a coherent and relatively

stable software framework (Baldwin and Woodard 2009). It is designed to be extended

by modules that perform complementary or more specialized tasks (Tiwana et al. 2010).

For example, the Chrome internet browser is a digital platform that can be extended with

a module that implements file transfer protocol (FTP). This module complements the

core functionality of browsing the web with the ability to upload documents to a remote

server. A major duty of any digital platform is to expose its core functionality through

interfaces over which modules can interact with the platform (Tiwana et al. 2010).

Hence a digital platform defines a compatibility standard (David and Greenstein 1990)

on which all modules can rely on. More often than not a user will want to customize a

digital platform they adopt with a specific combination of modules (Yoo 2013). For

example, most users install apps after buying a smartphone. We call any specific

configuration of digital platform and modules which jointly address a specific set of user

requirements a composition (Bosch and Bosch-Sijtsema 2010). While a digital platform

and its modules provide functionality, the myriad compositions and the utilization of

these compositions create value in specific use contexts (cf. Lusch and Nambisan 2015).


This distinction shows that digital platforms and their modules cannot be discussed

without considering the various social actors involved.

The first actor to consider is the orchestrator (Jansen et al. 2009), which designs and

maintains the digital platform and seeks to improve the overall ecosystem fitness.

Crucially, the ecosystem is unlikely to survive without continued participation of the

orchestrator (Selander et al. 2013). Generally, the orchestrator is assumed to be a single

actor, that is control of the digital platform is not distributed among a number of

orchestrators (cf. Hanseth and Lyytinen 2010). We chose the ‘orchestrator’ label instead

of the more commonly used term ‘platform owner’ (e.g., Tiwana et al. 2010; Wareham

et al. 2014) to underline that this actor’s duties reach beyond merely providing the digital

platform (Iansiti and Levien 2004). For example, the orchestrator balances and resolves

tensions that emerge in the digital platform ecosystem (Eaton et al. 2015). The next actor

to consider is the developer, which designs and maintains one or more modules. We

assume that an ecosystem attracts a multitude of developers, which through their

modules contribute to its variety and choice (Boudreau 2012). Lastly, the user assembles

available modules on a platform to fit their specific needs and employs the resulting

composition profitably. Again, we assume that many users engage in an ecosystem and

that they create a multitude of compositions. In our understanding users are not passive

recipients of a finished product; instead, they add to the variety and choice of an

ecosystem through their compositions (Lusch and Nambisan 2015).

At this point it is worth emphasizing this model’s ambition. We do not claim that it

allows describing an arbitrary digital platform ecosystem with sufficient detail for any

purpose of investigation – indeed, the concept of an ecosystem defies drawing precise

boundaries (Iansiti and Levien 2004). But we believe that the model is useful to discuss

the generativity phenomenon for three reasons: first, it gives us a common vocabulary

to structure widely different empirically observable systems, such as Apple iOS,

Eclipse, or OpenStreetMap. Second, it describes a sufficiently complex system. Third,

we can analyze both generative properties and generative patterns within this model. In

the following, we discuss selected aspects to consider when studying generativity in

digital platform ecosystems.

II.4.2 Generative properties in digital platform ecosystems

The idea of generative properties rests on the insight that distinct system properties let

actors engage with artifacts in ways that lead to unanticipated outcomes. For example,

we can assume that the very concept of an extensible digital platform is generative.


When authors allude to the “generativity in Apple’s platform” (Eaton 2012: 173) or

“generativity of business intelligence platforms” (Kretzer and Maedche 2014: 207) they

arguably make exactly this assumption. We could now productively continue on this

path and seek to identify additional properties which invite developers and users to

unboundedly engage with system artifacts – for example APIs that expose platform

functionality to third parties (Ghazawneh and Henfridsson 2013), compatibility

standards which ensure that modules can be easily assembled to compositions

(Wareham et al. 2014), or versatility, that is how much can be done with the digital

platform (Zittrain 2008). Notably, these properties are concerned with the digital

platform, and – just as importantly but sometimes neglected (cf. Boland et al. 2007) –

how developers and users engage with the platform to produce surprising outcomes.

This has two interesting implications. First, the orchestrator is assumed not to produce

any unanticipated change. After all, the orchestrator designs and maintains the digital

platform, which is the low-variety part in an ecosystem (Baldwin and Woodard 2009).

The ‘raison d’être’ of a digital platform is to provide a set of commonly used

functionality via stable interfaces, which does not lend itself to ‘unanticipated change’.

Furthermore, we assume a singular orchestrator, which is incompatible with the ‘broad

and varied audiences’ part of the generativity concept. Relaxing this assumption so that

there are many orchestrators – as can be observed on several open-source platforms like

the Linux kernel (Lee and Cole 2003) – does not change the premise of few unexpected

changes. We can rather expect extensive coordination among the orchestrators, as it has

to be decided which functionality shall become part of the digital platform and how it is

to be incorporated (Lee and Cole 2003).

Nevertheless, the orchestrator is crucial to increase generativity in a digital platform

ecosystem. First, the orchestrator designs and maintains the digital platform. Second,

even the most versatile and accessible digital platform does not produce any surprising

change on its own – it needs the participation of broad and varied audiences. Therefore,

we can state the contribution of the orchestrator-platform subsystem to generativity as

being the part in an ecosystem which attracts a large number of diverse actors and gives

them the means to engage creatively with technology.

Another implication of identifying particular attributes of the digital platform and its

governance as evocative of generativity is that neither modules nor compositions seem

to matter for assessing the generativity of a digital platform ecosystem. The creation of

unanticipated modules and compositions is the result of generative capacity of the

central digital platform, not condition for further generativity. Here, a major limitation


of the generative properties perspective becomes apparent. We have no means to

incorporate the everyday observations that any change is cumulative, that innovation

breeds further innovation, or that activities in one part of the ecosystem have

repercussions in other parts (cf. Boland et al. 2007). Consider a developer that creates a

surprising module. Let us assume that this module alleviates a perceived constraint of

the ecosystem (cf. Leonardi 2011). For example, when the first virtual smartphone

keyboard was introduced which allowed fast typing with continuous finger motion

(Biggs 2008), the constraint of ‘slow individual-character tapping’ was alleviated.

Similarly, but more controversially, several developers regularly seek ways to

‘jailbreak’ Apple iOS with specialized modules, which alleviates constraints to access

platform functionalities that the orchestrator did not intend to expose (Eaton et al. 2015).

Such modules create opportunities for developers which can now incorporate new

functionalities into their modules. Effectively, ecosystem generativity changes as its

modules change.

Also, the compositions affect ecosystem generativity, as users might find surprising

application contexts or creative assemblies. One example is how Facebook, which was

conceived for connecting with friends, was perceived by some users as formidable

means for organizing political rallies (Beaumont 2011). Another example is how

Wordpress, a digital platform originally designed to host blogs, was combined with a

selection of modules to host initially unanticipated forms of web publishing such as

large news portals or product catalogs (Boudreaux 2012). As users gradually discover

and exploit an ecosystem’s various opportunities, developers will likely seek to support

them with new modules (cf. Arthur 2009), which in turn is likely to create new

opportunities to be discovered (cf. Woodard et al. 2013).

To conclude, the generative properties perspective is useful to identify and analyze how

developers and users are empowered to engage with a digital platform to produce

unanticipated outcomes. But the perspective is less suitable to assess how generativity

plays out throughout all elements in a digital platform ecosystem. Unanticipated

modules and compositions lead to repercussions throughout the system, they do not

leave it unaffected. Hence, if we want to capture generativity in ecosystems, we should

not only analyze generative properties, but also how the ecosystem evolves over time.

II.4.3 Generative patterns in digital platform ecosystems

Generative patterns are contingent explanations for unanticipated ecosystem evolution

distilled from actual events. We can think of them as generalized chains of events which


explain how individual interactions play out over time to change the ecosystem in

surprising ways. Take this illustrative pattern: users join an ecosystem when they expect

to tap into its resources profitably (Selander et al. 2013). In selectively interacting with

ecosystem elements they assemble compositions for use in their local contexts. When

different users create similar compositions, in aggregate they may solidify one aspect of

the ecosystem (Hanseth and Lyytinen 2010). In turn, this affects how new users perceive

the ecosystem and also how developers respond with modules (Grisot et al. 2014). This

pattern might explain such different outcomes as the appropriation of the bitcoin

blockchain, the technology behind the well-known digital currency, for trading bonds

and other assets (The Economist 2015a), or the widespread use of smartphones for

shooting and sharing ‘selfies’ in all kinds of circumstances, a practice further solidified

with the addition of front-facing cameras to smartphone devices.

A distinct feature of complex systems, and thus of digital platform ecosystems, is the

possibility of emergent evolution, that is system change as consequence of autonomous,

yet mutually adaptive, behavior of individuals (Woodard and Clemons 2014). We may

assume that such behavior tends to lead to unanticipated consequences, otherwise it

would be centrally planned (cf. Mintzberg and Waters 1985). The ecosystem model

hence highlights a crucial aspect of generativity: it manifests in unanticipated change on

system level, but it is caused in locally bound contexts where individual actors create

and appropriate artifacts and adapt to other actors’ activities. The orchestrator-platform

subsystem constantly engages in interactions with the other ecosystem entities. In

addition, developers and users are attracted to or drop out of the focal ecosystem, which

establishes new links, releases historic ties, introduces new artifacts or modifies existing

ones. These interactions change the ecosystem in various ways, and some changes will

be unanticipated, or generative. Hence, individual interactions lie at the heart of

generative evolution. It is therefore appropriate to ask how locally bound subsystems

are influenced by the ecosystem they engage with, how they behave given these

influences, and how their collective behaviors change the ecosystem. Answering these

questions leads us to gradually formulate generative patterns whose explanatory power

we can test on future events.


Figure 8. A macro-micro-macro model of digital platform ecosystem change

Tracing emergent evolution forces our perspective to alternate between macro and micro

levels. This comes natural to complex systems scholars (Maguire et al. 2006) as well as

to the philosophy of critical realists (Bhaskar 2008). A useful analytical tool from social

sciences literature, pictured in Figure 8, is the macro-micro-macro model (Coleman

1986; Hedström and Swedberg 1998), which critical realist IS research recently adopted

through contributions in an MIS Quarterly special issue (Dobson et al. 2013;

Henfridsson and Bygstad 2013; Volkoff and Strong 2013; Williams and Karahanna

2013). The purpose of this model is to explain how a system produces change on the

macro level via micro-level actions of individual entities. In the context of a digital

platform ecosystem we can describe the model as follows (cf. Coleman 1986): The

ecosystem in one point in time influences individuals (step one, situational macro-to-

micro transition), which act in light of the current ecosystem condition (step two, micro-

level action formation). In aggregate, their behavior transforms the ecosystem into a

new state, be it intended or unintended (step three, transformational micro-to-macro

transition). Step one explains how the ecosystem attracts, empowers, and constrains its

actors. For example, the Google Android ecosystem attracts many developers due to its

sheer size. Step two explains how individuals act in their local contexts, usually

harnessing ecosystem resources in their reach. We assume that actors carry out activities

that are manifested in artifacts. For example, a perceived limitation of individual-

character tapping on smartphones led developers (actors) to create modules (artifacts)

that, among others, allow text entry via faster continuous finger motion; that suggest

words and sentences based on past entries; and that allow text entry via speech. Finally,

step three explains how actions of individual actors combine to produce emergent

change. For example, the addition of a large variety of text entry modules now available

on the Android ecosystem created innovation in the unlikely field of virtual keyboards.

In addition, ecosystem heterogeneity increased as developers added a growing number

of modules.


This macro-micro-macro model gives generative patterns a common format, namely as

explanations wherein individual behavior connects interesting aspects in the current

ecosystem state with unanticipated changes in a future state. Furthermore, a generative

pattern describes a tendency, that is chains of events that would be characteristic were

they to unfold, but do not necessarily need to happen (Wynn and Williams 2012). For

example, even if a given ecosystem consists of widely heterogeneous actor-artifact

combinations, we cannot predict that it is going to evolve generatively; but neither

would we be surprised if this happened.

Structurally, there is no need to treat the orchestrator-platform subsystem differently to

any other socio-technical entity in an ecosystem. They all contribute to system evolution

through individual behavior and thus in essentially the same way. Still, there is a large

difference to be made with regard to the impact which the orchestrator’s behavior has

on the overall course of system evolution. Because all entities interact with the

orchestrator-platform subsystem, the orchestrator naturally ‘cultivates’ (Itami and

Numagami 1992) the route which the whole ecosystem takes. This central role shows

itself in particular when the orchestrator strives to resolve erupting tensions. Through

the way it governs tensions the orchestrator recalibrates the overall ecosystem’s

positions on control vs. autonomy, standardization vs. variety, stability vs. change, and

other paradoxes which define the character of an ecosystem (Tilson et al. 2010;

Wareham et al. 2014). The powerful position of the orchestrator is not a given, however.

As a key difference between any actor and the orchestrator is the latter’s influence on

the ecosystem’s trajectory, we can think of circumstances in which a group of actors

jointly contest the orchestrator’s decisions and its authority (Eaton et al. 2015).

To conclude, the generative patterns perspective is useful to identify and analyze how

the activities of individual ecosystem entities may coalesce to a trajectory that over time

lets an ecosystem evolve in unanticipated ways. Furthermore, it draws our attention to

differences between actors with regard to how impactful their particular activities are.

While the first perspective is mostly concerned with artifacts and social structure, this

second perspective emphasizes the role of actors and of localized events. In particular,

it helps us to explain, not predict, how unanticipated ecosystem evolution happens.

II.5 The promise of generativity for explaining digital innovation

Let us return to the broader promise of generativity that we stated at the beginning of

this article: the concept of diverse groups of actors producing unanticipated change

arguably captures the essence of what digital innovation is about (cf. Yoo 2013). As


Post (2010) formulated it, generativity is “a nice shorthand label for a complicated

thought” (ibid.: 2578). This might be true, yet it does not mean that generativity is well

understood, let alone helpful for the IS discipline to study digital innovation. Zittrain

(2008) provided an elegant definition, which proved useful for advocating specific

policy measures targeted to counter a trend of increased vertical integration of Internet-

based businesses (cf. Best 2014; Hallingby et al. 2016). But to an IS scholar daring to

explore digital innovation, crucial elements of the generativity concept are left under-

specified. In the following, we discuss how this paper contributes to untangling the

generativity concept with the goal to make it more amenable to IS research.

Furthermore, we outline a few avenues for follow-up research.

First, there is much to gain from a socio-technical account of the generativity

phenomenon. Arguably, Zittrain regards interconnected general-purpose computers as

a prime exemplar of generativity (Zittrain 2006). This focus on technology left its traces

in IS research in qualifications such as “generativity of iOS” (Eaton et al. 2015: 234),

“a platform’s generativity” (Förderer et al. 2014: 2), or “generativity of digital

technologies” (Yoo 2013: 230). Yet, the long socio-technical tradition in IS research

has sharpened our understanding on how technical and social entities relate to each

other. For example, Boland et al. (2007) make the point that technology per se does not

herald innovation in complex networks. Rather, diverse groups of actors with access to

technology are those putting their capabilities to innovative uses.

In line with this tradition, the generative properties perspective is concerned with both

digital artifact and governance properties. We should pursue two goals with equal

emphasis: (1) systematically identify artifact designs with a large generative capacity;

and (2) detect which social structures are suitable for letting broad audiences harness

such designs and collaborate effectively. Large and long-lived open source software

communities provide particularly fertile grounds for studying such arrangements,

because they proved the viability of socio-technical systems that rely on contributions

of broad and varied audiences (cf. von Krogh et al. 2012). The open source arena also

shows us that generativity does not imply an absence of coordination and collaboration,

but rather the establishment of novel social practices (Lindberg et al. 2014; Raymond

1999). Epistemologically, we regard affordance theory (Markus and Silver 2008) as

markedly useful to model generative socio-technical systems. It provides a context-

bound relational explanation of how particular material properties of artifacts combine

with goals and skills of social actors to create value potential (Strong et al. 2014). Recent

research of Bygstad (2015) and of Yoo et al. (2012) already follow this path.


Second, as digitalization progresses, the unique attributes of digital artifacts (cf.

Kallinikos et al. 2013) increasingly affect the processes of creating new products,

services, and experiences. Perhaps most profoundly, the indefinite malleability of digital

artifacts – in form of code and content – cause organizational structures to become more

malleable as well (Henfridsson et al. 2014; Lyytinen et al. 2016). Even where these

changes seem chaotic, they likely follow some regularities which can be uncovered (Yoo

2013). For example, large information infrastructures arguably evolve along discernable

stages of bootstrapping, adoption, scaling, and demise (Hanseth and Lyytinen 2010;

Henfridsson and Bygstad 2013).

The generative patterns perspective is concerned with such contingent explanations for

socio-technical evolution. Again, open source software communities are particularly

amenable for generativity research. For one, open source software development happens

in the public domain, which leaves comprehensive evolutionary traces to study (e.g.,

Lungu 2009). This is hardly a new observation. Nevertheless, we see a surge of open

source projects, including a push from software powerhouses such as Microsoft (The

Economist 2015b). What is more, infrastructure for organizing open source

collaborative work is maturing, as services like GitHub exemplify. Hence, here is a large

area to be explored with regard to generative ecosystem evolution. While this paper

outlined how a critical realist might want to identify generative patterns, a much broader

arsenal is available to us to explain how digital platform ecosystems and other complex

systems change over time. For example, longitudinal empirical data points can be

analyzed to trace how sequences of events combine to generative evolution (Um et al.

2013), or how distributed networks evolve over time (Zhang et al. 2014). We can also

turn to simulations to model unanticipated change in complex systems (Woodard and

Clemons 2014). Finally, there is still more conceptual work required to grasp the

generative implications of pervasive digitalization (Lyytinen et al. 2016).

Lastly, there is tremendous potential in studying how generative properties and

generative patterns cross-pollinate each other. For example, we may ask under which

circumstances a generative digital platform increases the propensity of an ecosystem to

evolve generatively. In return, we may also ask to which extent generative evolution

leads to the emergence of generative properties in a digital platform. We refer to recent

research of Woodard and Clemons (2014) that points into these directions.


II.6 Conclusion

With this paper we develop a description of generativity in digital platform ecosystems.

To this end, we delineate two perspectives on generativity. We view the elaboration of

these two perspectives as this paper’s main theoretical contribution. The first perspective

sees generativity as consequence of system design, in particular design of the central

digital platform around which the broader ecosystem is organized. It assumes that

certain properties, for example open APIs, make a system generative while others make

it restrictive: given artifact and governance properties X, the system is generative. This

perspective views generativity mainly as a design issue, albeit with a strong governance

aspect. After all, there must be social actors which tap into the generative potential of

the digital platform. Therefore, the orchestrator must not only design a generative digital

platform, but also has the duty to attract diverse actors that can creatively engage with

it.

In contrast, the second perspective regards generativity as consequence of ecosystem

evolution. It assumes that certain event patterns where actors and artifacts interact

produce generative change: an order of events that reflects pattern X lead the system to

evolve generatively. This perspective views generativity as non-deterministic

phenomenon, in which past accomplishments are no assurance for continued generative

evolution. It is a more intricate depiction of digital platform ecosystems, wherein events

of socio-technical interactions are scattered across the ecosystem, and where highly

diverse ambitions and skills meet. The orchestrator is constantly confronted with

tensions that require resolution. Instead of a grand designer, the orchestrator is more like

a humble curator, and the way it cultivates the ecosystem heralds either generative or

de-generative evolution.

We believe that a clear distinction between generative properties and generative

patterns helps researchers to describe, understand, and explain the generativity

phenomenon more precisely. While we discussed some implications for research largely

on the example of digital platform ecosystems, the two proposed perspectives might

prove useful in studying other forms of complex socio-technical systems as well. Indeed,

reviewing salient literature showed that both perspectives have been discussed across a

number of different types of complex systems, including digital platform ecosystems

and information infrastructures.

A second result of this paper is the formulation of an explicitly socio-technical model

of digital platform ecosystems. Expanding existing ecosystem models, we introduce


three types of interacting socio-technical subsystems, namely orchestrator-platform,

developer-module, and user-composition. This topology let us discuss generative

properties and generative patterns on common grounds. Specifically, we find that the

first perspective, generativity as consequence of system design, focuses primarily on the

orchestrator-platform subsystem. Meanwhile, the second perspective, generativity as

consequence of system evolution, highlights that any macro-level change is caused by

micro-level behavior.

This study certainly has its limitations. First, and most apparently, we did not anchor the

proposed concepts in empirical material. Instead, we delineated the two perspectives on

generativity from analyzing a widely accepted definition of this phenomenon. This

approach was chosen in order to clear some of the prevailing uncertainty around the

meaning of generativity. By surveying salient literature, we were able to locate these

differing interpretations in past research, which we regard as ‘prima facie’ validation of

their appropriateness. Second, we were light in connecting our argumentation to related

theories. To mention just two possibilities, the discussion on generative properties might

have benefited from linking it with affordance theory, and the generative pattern

perspective lends itself to association with complex system theories. It was our intention

to develop this paper’s arguments as diligently as possible, and we felt that establishing

many theoretical links would have stretched them too thinly.


II.7 Appendix: descriptions of generativity in extant literature

Table 17. Descriptions of generativity as consequence of system design

# Thematically grouped summary descriptions of ‘generative properties’ perspective as presented in literature

1A Generativity is a consequence of combinatory capacity and reprogrammability of digital artifacts

1 Boundary resources make digital platforms externally accessible and thus generative. (Eaton et al. 2015)

2 XML is generative as it allows encoding local practices yet adheres to global standards. (Tilson et al. 2010)

3 Generativity is the capacity of an ecosystem to recombine existing software modules in novel ways, which produces unprompted change. It explains how platforms can offer many innovations. (Um et al. 2013)

4 Through their properties such as reprogrammability, a finite number of digital artifacts can be recombined nearly arbitrarily, which makes them generative. (Yoo 2013)

1B Generativity is a socio-technical phenomenon wherein digital platforms attract external contributions

5 Generativity is the ability of social and technical elements to recombine for producing new solutions. It is a characteristic of an information infrastructure, not that of an isolated solution. (Bygstad 2015)

6 Generativity is the creation of new functionalities, brought about outside development entities and enabled by digital platform architecture. (Jain and Ramesh 2015)

7 Generativity refers to how easily external actors can use a digital platform to develop new services. It is not about creation as activity, but about how technology properties invite creation. (Nielsen and Hanseth 2010)

1C Generative capacity of digital artifacts is leveraged through creativity and diverse skills of outside parties

8 Digital platforms are designed for generativity, which reflects the ability to generate unforeseen outcomes by mobilizing outside actors’ local expertise. (Förderer et al. 2014)

9 Loose coupling makes complex digital artifacts generative, which brings in heterogeneous actors for design and production of combinatorial innovations. (Jackson 2015)

10 Generativity is the potential of a digital platform to support the creation a wide range of unforeseen services. It considers IT characteristics and how they relate to actors’ creativity. (Vassilakopoulou and Grisot 2012)

1D Generativity can and should be governed by the platform-controlling actor 11 Generativity refers to the engagement of diverse user communities with open and flexible IT

artifacts to produce new content. The IT artifact owner can influence this process. (Jarvenpaa and Tuunainen 2013)

12 Information infrastructures are generative due to their open interfaces which let external actors create new content. Generative capacity is reduced when outside innovation is controlled. (Thorseng and Jensen 2015)

13 Generativity is the ability of a system to create something new without the originator’s input. Successful ERP systems provide a standard core and attract third parties that produce complements. Ecosystems must harness autonomous behavior for generativity, but they must also give it direction. (Wareham et al. 2014)

14 A digital platform accomplishes generativity, as it allows adding complements by external actors after the fact. It must not be controlled too heavily, as driving out third parties reduces generativity. (Yoo et al. 2012)


Table 18. Descriptions of generativity as consequence of system evolution

# Thematically grouped summary descriptions of ‘generative patterns’ perspective as presented in literature

2A Generativity is serendipitous system evolution through emergent change

1 Infrastructure evolves over time, which is captured by the generativity concept. (Bygstad 2015)

2 Generativity is an instance of the phenomenon of emergent outcomes arising from decentralized behavior. It explains how digital artifacts support the emergence of ecosystems. (Woodard and Clemons 2014)

2B Generative evolution is caused by high diversity in space and in skills among involved actors and artifacts

3 Generativity results from combining heterogeneous skills to stimulate co-production of new outputs. As an idea moves towards product, each actor adds its own expertise. (Lyytinen et al. 2016)

4 Digital artifacts are highly evolving and thus generative. They emerge through distributed interactions among heterogeneous actors. (Yoo 2013)

5 Generative diffusion refers to the ability of products to change over time through the participation of uncoordinated actors with different skills. One class of examples are IT artifacts which are created in open source communities, because they mutate as they are being adopted. (Zhang et al. 2012; Zhang et al. 2014)

2C Generativity thrives on knock-on effects wherein a local change event leads to change in other system parts

6 In a generative system, innovations can spring up independently at any layer, which in turn triggers innovations in other layers. (Yoo et al. 2010)

7 Generativity manifests through processes in which innovations in one part of the ecosystem trigger follow-up innovations in other parts. (Yoo et al. 2012)

2D Generative dynamics cause tensions which must be dealt with

8 Through generatively evolving boundary resources, ecosystem tensions are resolved. (Eaton et al. 2015)

9 Reflexivity leads to emergence of generative dynamics. These dynamics lead to tensions between stability/ change and control/ autonomy. (Tilson et al. 2010)

10 Generativity reduces control over infrastructure by leaving it accessible to third parties. It shows how software agency intertwined with human agency shapes future possibilities and is influenced by coordination tensions. Such an evolution of infrastructure challenges principles of planned change. (Venters et al. 2014)

Part B: Reconstructing open source software ecosystems: finding structure in digital traces 71

III Reconstructing open source software ecosystems: finding

structure in digital traces

Table 19. Bibliographic information for Article III

Title Reconstructing Open Source Software Ecosystems: Finding Structure in Digital Traces


Outlet ICIS 2016 Proceedings

Year 2016

Status Published

Abstract. We report on the computational reconstruction of 273 open source software

ecosystems, consisting of 41,388 artifacts and couplings between them, extracted from

digital traces of 34.4 million software artifacts. We argue that digital traces are a new

kind of data source, and propose ‘exploratory data loops’ to exploit the benefits of digital

trace data in early stages of a research program. We apply this schema to systematically

assess data quality, inform sample selection, and detect patterns. Empirically, we show

that highly distributed networks are unlikely to follow a hierarchically modular

structure, despite popular belief. As is shown visually with two examples, very distinct

structures can emerge from autonomous behavior. The results indicate that different, yet

similarly effective, strategies may exist to organize for distributed innovation in digital

ecosystems. The paper is concluded by outlining how follow-up work will harness the

reconstructed ecosystems for detecting behavioral patterns in distributed networks.

Keywords: doubly distributed networks, digital ecosystems, open source software,

digital trace data, computational social science, data-driven research

72 Part B: Reconstructing open source software ecosystems: finding structure in digital traces

III.1 Introduction

Software developers like to quip that their best programming skills include cleverly

googling code snippets and stitching together frameworks of various sorts (e.g., Ali

2016; Haney 2016). As with most jokes, there is truth to be found. Driven by widespread

success of the open source software model, designers of digital artifacts can freely draw

upon a wealth of useful and diverse resources. As a result, individual artifacts are

coupled with many other artifacts, thus creating complex digital ecosystems (Selander

et al. 2013). It has been argued that digitized economies will increasingly produce

innovations within highly distributed networks (von Hippel 2005; Yoo et al. 2010). Yet,

empirical research that shows how distributed innovation plays out is scarce, as is

research that investigates several ecosystems simultaneously (Manikas 2016). Arguably

this is because digital ecosystems are heterogeneous and complex (Lyytinen et al. 2016),

and thus their elements, structures, and behaviors are hard to identify and to describe.

Therefore, to enable empirical research on distributed innovation, a systematic approach

for identifying digital ecosystems is beneficial. This paper lays the foundations for an

endeavor that aims to uncover innovation dynamics in distributed networks by asking:

How can we reconstruct open source software ecosystems from digital trace data?

Regarded individually, digital artifacts, or artifacts that are created, assembled, accessed

or otherwise manipulated through digital technology (Kallinikos et al. 2013; Leonardi

2010), are already complex objects that are hard to manage and to study (cf. Brooks

1986). Investigating ecosystems of artifacts, and social actors related to them, pushes

the limits of what can be achieved with traditional approaches (Ma and Muelder 2013).

But with challenges come opportunities. By definition, open source software ecosystems

– digital ecosystems whose constituting artifacts are open source software artifacts –

emerge and evolve in the public domain, because the source code of these artifacts is

available to the public (Raymond 1999; von Hippel and von Krogh 2003). They also

generate troves of highly granular data that information systems (IS) researchers can tap

into (Agarwal et al. 2008). Consequently, computational approaches to make sense of

digital trace data started to emerge in the IS discipline under the label of computational

social science (e.g., Berente and Seidel 2015; Gaskin et al. 2014; Howison et al. 2011).

They share the principles of moving between multiple levels of aggregation, and of

iterating between exploratory and confirmatory phases.

Here we present the results of an exploratory phase, structured as follows. First, an

overview of open source software ecosystems and distributed innovation is given. We


identify a lack of research that investigates digital ecosystems of real-world scale, taking

into consideration how the individual elements influence each other and how their

interdependencies shape the respective ecosystems they inhabit. Next, by discussing the

characteristics of digital trace data, we propose three ‘exploratory data loops’ for

appropriating digital traces systematically in early stages of research. We then present a

dataset of 144,311 open source software artifacts and relationships between them, which

were reconstructed from over 58 million data points. We apply the proposed exploratory

data loops to this dataset, which yields three main results. First, we find data quality to

be sufficient for the intended use. Second, we select 273 open source software

ecosystems out of the available population. Third, we present empirical evidence that

highly distributed ecosystems are unlikely to have small-world structure (Watts and

Strogatz 1998), despite the prevalence of said structure throughout many socially

constructed complex networks (Albert and Barabási 2002). We briefly discuss the

significance of these results, and conclude the paper by outlining how we plan to harness

the presented dataset in further research.

III.2 Distributed innovation in open source software ecosystems

III.2.1 Open source software ecosystems

Rare is the software developer who sets out to implement her design with nothing more

than a text editor and a compiler. Digital artifacts seldom come into existence and

operate in isolation. Rather, they rely and depend on other digital artifacts for their

functioning. For example, to implement a modern web application, designers marshal a

multitude of resources over which they have no direct control, such as scripting

languages and external data sources (Jazayeri 2007). Couplings among digital artifacts

induce interference: as they are indefinitely malleable and thus amenable to change (cf.

Kallinikos et al. 2013), digital artifacts co-evolve with their environment, causing

further need for changes (Boland et al. 2007). Such coupled elements share the same

ecosystem (Iansiti and Levien 2004). IS researchers regard an ecosystem as a socio-

technical system comprising of interdependent digital artifacts and social actors (e.g.,

Selander et al. 2013). Hence, we define a digital ecosystem as a collection of digital

artifacts that co-evolve through mutual interference, and the social actors related to these

artifacts that are linked by a common interest (Eck and Uebernickel 2016b; Selander et

al. 2013).

The ecosystem metaphor carries notions of complexity and emergence, wherein

distributed social actors and digital artifacts interact in non-trivial ways (e.g., El Sawy


et al. 2010). Studying digital ecosystems raises the question of how to detect and trace

structure and dynamics in light of a large number of constituting elements. Case study

approaches (e.g., Boland et al. 2007) risk concerns of generalizability, whereas

simulations (e.g., Woodard and Clemons 2014) might be perceived as overly stylized or

reductionist. Digital trace data, or data generated by actors interacting with digital

artifacts, are promising data sources for investigations that aim to match the complexity

and size of the systems under study (Watts 2007). The widespread diffusion of open

source software (Raymond 1999) and of open source development activities happening

in the public domain (von Krogh and von Hippel 2006) has generated – and continues

to generate – large amounts of digital trace data. Despite valid concerns not to jump on

large data pools simply on the merit of accessibility (Rai 2016), open source software

ecosystems are destined for studying complex digital ecosystems, precisely because they

are so accessible to outsiders, and thus enable distributed innovation (Kogut and Metiu

2001).

In order to study open source software ecosystems, we must identify them first, that is

reconstruct their elemental parts and the couplings between them from empirical data.

While the majority of research has been confined to studying individual open source

software artifacts (Crowston et al. 2012), there is some existing IS research which

pursued ecosystem reconstruction. Two distinct strategies emerged from this research.

First, in the tradition of social network analysis (Wasserman et al. 2005), it is possible

to reconstruct ecosystems based on activity of social actors. Two open source software

artifacts are assumed to be related if at least one actor contributes to both artifacts, that

is links between digital artifacts are established via joint affiliation of social actors (cf.

Faust 2005). This approach lends itself to studying how people collaborate, whereas the

co-evolution of digital artifacts is less of a concern (e.g., Grewal et al. 2006; Singh et al.

2011; Zhang et al. 2014). The second strategy captures known relationships between

artifacts, based on a priori knowledge of the ecosystem structure. More specifically, a

focal digital platform and its various complementing modules (cf. Tiwana et al. 2010)

are selected for further investigation, like the ecosystem around Wordpress, a content

management system (Um et al. 2015). While this approach is suitable to reconstruct an

open source software ecosystem, it is restricted to the digital platform kind, which

excludes more heterogeneous topographies (cf. Hanseth and Lyytinen 2010).

Furthermore, it is a top-down approach, which is somewhat at odds with the complex

and emergent nature of ecosystems (cf. Woodard and Clemons 2014). Recently, a

bottom-up strategy has been established in discourses of software evolution (Mens

2008): ecosystem reconstruction based on design interdependencies, called artifact


couplings in the context of this paper. Through inspection of source code (e.g., Lungu

2009) or of metadata associated with a digital artifact (e.g., Blincoe et al. 2015;

Gonzalez-Barahona et al. 2009; Santana and Werner 2013), artifact couplings are

systematically detected. In the research reported here, we apply this strategy to

reconstruct open source software ecosystems.

III.2.2 Distributed innovation

Yoo et al. (2008; 2010) and Lyytinen et al. (2016) theorize that most innovation in the

digital realm will emanate from doubly distributed networks characterized by

distribution of resources and distribution of control. The first dimension denotes that

digital artifacts stem from disparate development trajectories, conceived by actors with

varying capabilities. The second dimension refers to the absence of centralized control

and authority. Thus, in doubly distributed networks, resources and activities are

scattered across the ecosystem heterogeneously. A dynamic ensues wherein ecosystem

components co-evolve, which creates opportunities for unbounded, distributed

innovation. This seems to be a fair description of at least some open source software

ecosystems, which makes them a reasonable fit for inquiries of distributed innovation

(von Krogh and von Hippel 2006).

Despite some promising empirical results (e.g., Berente et al. 2008), the idea of doubly

distributed networks remains underexplored. Instead, the concept of generativity has

seen wider adoption among IS researchers keen to study distributed innovation in

complex networks, or processes of creating innovative outcomes from collective

behavior that is distributed along key dimensions (von Hippel 1988). While a number

of different definitions for generativity exist, arguably the most salient one was

formulated by Zittrain (2008: 70): “a system’s capacity to produce unanticipated

change through unfiltered contributions from broad and varied audiences”. It is worth

noticing how close this definition is to the notion of doubly distributed networks, as it

outlines the very same characteristics of distributed innovation: first, an invitation to

participate irrespective of actors’ individual capabilities (broad and varied audiences);

second, a lack of central control or oversight (unfiltered contributions); and third,

innovative outcomes from collective behavior (unanticipated change).

Generativity research in IS has been mostly conceptual to date (Eck and Uebernickel

2016b; Tilson et al. 2010; Yoo et al. 2010; Yoo 2013). While a number of empirical

contributions have adopted the generativity concept, it has usually not been their primary

concern (e.g., Eaton et al. 2015; Ghazawneh and Henfridsson 2013; Wareham et al.


2014). To the best of our knowledge, Um et al. (2015) are the first to empirically

examine generativity in an open source software ecosystem. The authors draw three

main conclusions: First, they infer that a large ecosystem tends to evolve generatively.

Second, they observe that ecosystem expansion does not predict innovation within the

ecosystem. Third, they detect widely varying activity between the individual

components; while a selected few are highly influential, many others are rather esoteric.

These are fascinating results, yet they are derived from examining merely one open

source software ecosystem. Clearly, an examination across many such ecosystems is

required to acquire generalizable knowledge on how the unfiltered contributions of

broad and varied audiences produce innovations. The research results presented here

report on preparations for such a larger-scale empirical study, which draws upon digital

trace data for empirical grounding.

III.3 Digital trace data as new kind of data source

III.3.1 Digital trace data in information systems research

Most events of people interacting with digital artifacts leave digital trace data (Agarwal

et al. 2008; Howison et al. 2011). Data points usually cover intended outcomes (e.g.,

email message) as well as by-products of the performance (e.g., message size). Some

traces are amenable to computational harvesting, which results in an extracted dataset.

This dataset may contain traces of activity both in pristine form (e.g., email recipient)

and in post-processed form (e.g., detected hyperlinks per message). We can transform

this dataset into structured entities following a pre-defined model. For instance, we

might want to create a graph of email senders and recipients, and the hyperlinks which

were sent around. We can use this populated model to obtain measures, which are

indicators for higher-level constructs. Following an approach of this kind, we might

answer the question of how internet memes spread virally (e.g., Bauckhage 2011). An

illustration of this process (Howison et al. 2011) is depicted with Figure 9. The bi-

directional arrows indicate that the process can be either construct-driven or data-

driven. In the first case, the avid researcher starts with a construct in mind, working her

way backwards to specify which kind of digital trace data is required. In the second case,

she starts exploring a given set of digital trace data in search for both expected and

unexpected patterns (Grover and Lyytinen 2015).


Figure 9. Generic process to harvest and analyze digital trace data

Digital trace data is attractive to researchers, because it differs from traditionally

collected data such as observational or survey data in several ways. First, it is found

data, that is it was not painstakingly created with a specific inquiry in mind (Howison

et al. 2011). Second, it captures actual behavior which is not susceptible to respondent

bias (Freelon 2014). Third, digital trace data is granular, which lets us extract,

aggregate, and interpret it in varied novel ways (Latour 2010). Forth, access sets digital

trace data apart from preceding attempts subsumed under the knowledge discovery label

(cf. Chung and Gray 1999). From Github to Twitter and Wikipedia, the open source and

social media movements produced large data pools outside of organizational walled

gardens. It is not surprising, then, to see appropriation of digital trace data in IS research

for studying varied topics, such as creation of coordination mechanisms in open source

software development (Howison and Crowston 2014); emergence of routine structures

in collaborative work (e.g., Lindberg 2015); and social network formation (e.g., Johnson

et al. 2014) in online communities, to name a few. In line with Hedman et al. (2013) we

expect empirical research based on digital trace data to gain further momentum in the

years to come.

However, even with digital trace data there is no free lunch. Here, we highlight two

fundamental issues that need to be dealt with. First, researchers are confronted with

ethical issues of hitherto unknown magnitude (cf. Zimmer 2010). Due to the very nature

of digital trace data, we can safely assume that people did not consent to any analyses

unrelated to their immediate interaction with the digital artifact. Furthermore, due to the

amount and granularity of available data, any compromise of personal rights will likely

incur high damages. Second, digital trace data typically is incomplete along at least two

dimensions (Bird et al. 2009). Digital artifacts are susceptible to change and unintended

use, which makes full coverage, consistent data extraction, and correct interpretation

challenging (cf. Howison and Crowston 2014). Moreover, the extraction procedure

might not cover all potentially available digital trace data, for example due to technical

limitations designed into the digital artifact that the researcher harnesses for extracting

data. In light of both the promises and perils of digital trace data, we discuss the

implications of incorporating such data in early stages of empirical research in the

following.


III.3.2 Implication for IS: in need of ‘exploratory data loops’

Despite the opportunity to perform both construct-driven and data-driven research, IS

scholars seemingly appropriated digital trace data largely as ‘yet another data source’

(cf. Hedman et al. 2013). With a few notable exceptions (e.g., Müller et al. 2016; Um et

al. 2015), data-driven research is widely lacking, in the sense that it explores the entirety

of an accessible body of digital trace data before narrowing down to a small sample.

Instead, it is common to filter out the majority of available traces a priori, typically based

on theoretical considerations. Sampling follows concept-driven approaches, which in

traditional empirical work have proven to serve us well. For example, in their study of

how knowledge flows within open source software ecosystems, Singh et al. (2011)

select a specific ecosystem a priori, thus reducing the number of individual software

artifacts under consideration from potentially over 100,000 down to about 2,400 (Singh

et al. 2011: 818). Sure enough, the authors are diligent in their methodology, and test

whether their main design decisions are sensible, for instance by presenting some

empirical support for their rationales.

We do not mean to imply that a concept-driven approach needs to lessen the

methodological soundness, original contribution, and overall quality of such research.

Our argument is rather that a substantial opportunity is missed if we do not reap the main

advantage of digital trace data over traditionally collected data forms: namely the chance

to engage with the empirical context, before diving into the thick of investigation (cf.

Tukey 1980). To describe it more vividly, traditional approaches trained us to first

decide on the frame and then fill the canvas, whereas the advent of digital traces asks us

to explore the pre-filled canvas, challenges us to settle on the most interesting frame,

and to continue from there. Digital trace data, in combination with data processing

capabilities of typical computers available to researchers, afford a new way to interact

with empirical material, particularly in the early stages of a research endeavor. In the

following, we discuss three considerations, which leads us to propose three distinct

exploratory data loops. Very much in the tradition of exploratory data analysis (Tukey

1977), the proposed exploratory data loops are intended to make the researcher familiar

with an initially unfamiliar dataset. Their outcomes are descriptive, with subsequent

theory building research designs (e.g., Berente and Seidel 2014; Gaskin et al. 2014;

Howison and Crowston 2014) required to seek explanations for the phenomenon under

study.

First, we should take notice of the possibility to assess data quality and gain a macro

picture of the domain under study in the process (Naumann 2014). Obtaining an idea on


how the entities of interest are distributed; which dimensions in the data correlate and

which do not; how complete the dataset is; and generally getting a feeling for the data

are crucial steps to assess data quality – after all, it is found data that shall be approached

with due skepticism (Freelon 2014).

Second, we can explore data before settling on a specific sample, which informs

decision-making. In empirical research, samples are drawn if studying the full

population is not feasible. If the selection criteria are sound, a sample is representative

of the whole and thus we can derive valid conclusions from it. Unlike survey or case

study data, digital trace data exists before our research starts. Thus, there is little reason

not to examine the general shape of it before settling on a sensible extraction protocol.

This approach is also in the spirit of Bayesian statistics, which tells us that decisions

tend to become better when we take more information into account that is available at a

certain point in time (Taroni et al. 2014).

Third, and possibly most intricately linked to a researcher’s curiosity, digital trace data

let us explore and detect regularities, trends, and surprises (Grover and Lyytinen 2015).

For example, Basole (2016) examines the topology of twelve inter-firm networks and

detects systematic similarities among high-performing networks. Arguably, such data-

driven, predominantly descriptive approaches are particularly useful when investigating

very complex systems, whose many, multi-layered interactions we do not (yet) grasp.

Grolemund and Wickham (2014) make the point that any researcher who has obtained

a large collection of (digital trace) data is likely to follow a sense-making process that

iteratively switches between exploratory and confirmatory procedures. Even more

radically, Berente and Seidel (2015) recently questioned the integrity of research results

based on digital trace data that make the impression to have been the outcome of a purely

confirmatory research design.

III.4 Reconstructing open source software ecosystems from digital

traces

This article is part of a larger research endeavor that aims to uncover how large open

source software ecosystems emerge and change through co-evolution of their

constituting elements, and how these dynamics lead to innovative outcomes. Open

source software is an apt field for studying distributed innovation. It is known for

promoting individual initiative and distributed governance, which in aggregate has been

leading to outstanding outcomes (von Hippel and von Krogh 2003). Open source has

seen widespread use, which makes the population of all open source software artifacts


arguably going into many millions. To capture the unbounded nature of distributed

innovation, we find it crucial to be as inclusive as possible, while still acquiring and

processing empirical data efficiently. For data collection, we turned to Github, which is

not only a leading collaboration platform and repository for open source software

(GitHub 2017b), but also grants access to digital traces via an API, or an application

programming interface (GitHub 2017c). As endpoints can call the API only a limited

number of times per hour, we mostly relied on the Ghtorrent dataset, which makes a

large part of API-provided data available to researchers (Gousios 2013).

We regard an open source software ecosystem to be a set of interacting, coupled, and

co-evolving open source software artifacts and associated social actors. In order to limit

conceptual complexity in light of the vast amount of data we expected to collect, and in

line with prior work (e.g., Carlile 2002; Um et al. 2015), we chose to identify ecosystem

boundaries based on couplings between the artifacts, instead of couplings between both

artifacts and actors. A common bottom-up approach to detect artifact interdependencies

is through source code inspection (e.g., Lungu 2009). However, source code inspection

does not scale well across a broad range of software artifacts. For example, we would

need to devise different techniques for every programming language that we encounter

(cf. Baldwin et al. 2014). To overcome this hurdle, we traded completeness on artifact

level for coverage on population level. Specifically, we harnessed the Github discussion

boards. On these boards developers discuss issues, exchange ideas for improvement,

and so on. As part of their discussions, developers may reference external software

artifacts following a certain syntax (GitHub 2017e). We exploited this feature to detect

couplings uniformly across all open source software artifacts hosted on Github.

For example, on 05 May 2015 a developer commented with regard to a previously

identified issue with ‘msysgit/git’7: “It is fixed in curl master (not released yet), commit

curl/curl@59f3f92” (GitHub 2015). This developer referenced ‘curl/curl’, and the

context of the discussion clearly shows that there is a coupling between the two artifacts.

Blincoe et al. (2015) demonstrated that cross-references on Github discussion boards

indeed reveal couplings between artifacts, which makes this approach methodologically

sound. Conceptually, we argue that a discussion in which developers reference external

software indicates that an artifact co-evolves with its environment. Therefore, only

extracting artifacts for which discussion entries exist filters out those objects published

7 Artifacts on Github are uniquely identified with the syntax ‘artifact_owner/artifact_name’.


on Github for purely archival purposes (cf. Kalliamvakou et al. 2016). Moreover, this

approach also filters out couplings between artifacts which co-exist but do not co-

evolve, that is artifacts that are connected technologically, but do not change over time

or that change without mutual influence. For example, while a weather forecasting app

might rely on an external data provision service, it might well be that the forecasting

algorithm improves over time without the interface between the two artifacts ever

changing. Lastly, given the well-defined syntax of how to reference external artifacts on

Github discussion boards, couplings can be detected via automated pattern matching.

This makes the chosen approach efficient, albeit not exhaustive. The data available to

us spanned 8 years, from 10 Apr 2008 (the day Github became available to the public)

until 31 Mar 2016, and encompassed 58,260,994 discussion board entries. The

extraction technique yielded 544,979 couplings across 144,311 different open source

software artifacts.

From this sample we created a network graph, with the software artifacts as its nodes

and couplings between artifacts as its edges. It would be inaccurate to regard this graph

as ‘the one’ open source software ecosystem, because of the variety of the open source

software field: for instance, the ‘xbmc/xbmc’ media player hardly co-evolves with the

data visualization library ‘mbostock/d3’. We assume that an individual ecosystem can

be detected by localizing groups of artifacts which share many couplings within their

group but few couplings outside of their group. For example, ‘xbmc/xbmc’ is coupled

with 210 other artifacts in our sample and ‘mbostock/d3’ is coupled with 155 artifacts,

but not with each other. The problem of detecting groups within a network is known as

community detection in graph theory (Girvan and Newman 2002). We selected the well-

performing infomap algorithm (Lancichinetti and Fortunato 2009; Rosvall and

Bergstrom 2008) to partition the graph into groups. In line with methodical

recommendations (Bohlin et al. 2014) the non-deterministic algorithm was repeated 500

times to increase the chance of obtaining a global optimum instead of a local one, which

resulted in 22,609 individual groups. The largest group had 941 nodes, 14 groups had

500 or more nodes, 360 groups had 50 or more nodes, and 20,480 groups had 10 or

fewer nodes. We interpret each group detected via the infomap algorithm as being an

individual open source software ecosystem (cf. Blincoe et al. 2015).

III.5 From data hairball to surprising pattern

This paper documents steps towards an empirical study on distributed innovation in

open source software ecosystems. As discussed above, we argue that when drawing


upon digital trace data, three exploratory data loops are mandated in early stages of

research: to assess data quality; to inform sample selection; and to detect intriguing

regularities. Following this sequence, we show how an initially overwhelming ‘data

hairball’ (cf. Nocaj et al. 2014) was systematically analyzed to detect an unexpected

regularity among doubly distributed ecosystems, which is illustrated in Table 20 and

explained in the following.

Table 20. Ecosystems ‘rails/rails’ and ‘minecraftforge/minecraftforge’

Software artifacts

Couplings

Highly coupled artifacts

Total forks

Scale-free correlation (r2)

899

1,458

12

84,857

0.73

Software artifacts

Couplings

Highly coupled artifacts

Total forks

Scale-free correlation (r2)

653

1,732

25

16,464

0.88

Graph layout ARF layout algorithm (Geipel 2007) Node/label color Number of forks (logarithmically scaled gradient from green to purple) Node/label size Number of couplings (logarithmic scaling) Edge width Number of times coupling between two nodes was detected in digital trace data

(logarithmic scaling) Label Name of artifact (shown if highly coupled, i.e. coupled with >5% of all artifacts

in ecosystem)

III.5.1 Exploratory data loop I: assessing data quality

We assessed data quality with a number of metrics and approaches, of which we

highlight three. First, we verified that the available trace data was sufficiently

exhaustive. For example, we counted over 34.4 million different software artifacts in the

dataset available to us, which is close to the 35 million reported by Github (GitHub

2017b). Second, we must be confident in the extraction technique which we


implemented with custom-written software. Following a comparable technique, Blincoe

et al. (2015) report to have detected 18,533 artifacts coupled via 89,784 links in a dataset

comprising 2.4 million as of May 2014; that is, 0.8% of all artifacts were coupled. In

contrast, we extracted couplings between 0.4% of all artifacts in the dataset. This is not

surprising, given that socially constructed networks tend to grow faster in number of

nodes than in number of edges between nodes (Johnson et al. 2014). In this context, it

means that the population of artifacts increased faster than couplings between artifacts.

Third, we could confirm that the constructed network graph topology was as expected.

For example, we found evidence that the constructed graph is scale-free, which – among

others – indicates that it reflects the outcome of complex social activity and is not purely

random (Albert and Barabási 2002), which would have been the case if the underlying

dataset were systematically flawed. Scale-freeness is defined as the degree distribution

(in our case: couplings between artifacts) fitting a power-law distribution (Barabási and

Albert 1999). We tested for scale-freeness as described in Kasthurirathna and

Piraveenan (2015), and calculated r2=0.86 for the whole graph, denoting a very strong

correlation between a fitted power-law function and the empirical data.

III.5.2 Exploratory data loop II: informing sample selection

In prior studies of open source software ecosystems, sample selection relied on rather

coarse, a priori criteria such as identical programming language (Howison and Crowston

2014) or affiliation to a particular digital platform (Um et al. 2015). In contrast, owing

to the many ecosystems reconstructed from digital trace data, we were able to apply

more precise selection criteria. This research is guided by an interest to study distributed

innovation. Thus, it is essential to select those ecosystems for inquiry that are likely to

satisfy the conditions of a doubly distributed network, meaning that they are highly

distributed both in resources and in control dimensions.

We operationalized distribution of resources as the number of artifacts in an open source

software ecosystem. Hence, we assume that an ecosystem consisting of many coupled

artifacts exhibits a heterogeneous distribution of resources overall, including knowledge

(cf. Lyytinen et al. 2016). Moreover, we operationalized distribution of control as the

sum over the number of forks per artifact in an ecosystem. On Github, anybody can

’fork’ the source code of an open source software artifact, change this copy at will, and

propose any change to be integrated with the master version of the source code (GitHub

2017a). Although the original artifact owner may choose to ignore change proposals,

previous research identified forking as a mechanism that effectively diverts control into

the periphery (e.g., Biazzini and Baudry 2014; Dabbish et al. 2012). Hence, we assume


that control is distributed in an ecosystem whose constituting artifacts have many forks,

and thus many prospects for outside contributions exist.

The 98th percentiles of each dimension were set as thresholds: We considered an

ecosystem to be doubly distributed, if and only if it was composed of at least 42 software

artifacts and 3,097 total forks. Applying the selection criteria yielded a sample of 273

open source software ecosystems, representing 41,388 software artifacts (median: 94

artifacts per ecosystem) and 3,596,202 total forks (median: 7,749 forks per ecosystem).

III.5.3 Exploratory data loop III: detecting intriguing regularities

A large number of distributed networks, such as the world wide web and power grids

(Solé et al. 2002), have been found to be small-world networks (Watts and Strogatz

1998). They are sparsely connected, yet information (or any other signal) spreads very

fast due to short-cut nodes that sit in between otherwise unconnected groups of nodes;

furthermore, many nodes are part of well-connected local clusters (Albert and Barabási

2002). Small-world networks can be thought as hierarchically modularized systems,

wherein short-cut nodes (interfaces) connect local clusters (modules) to form other

modules. Indeed, small-world structure has been found to be common in complex open

source software systems (e.g., Myers 2003).

Table 21. Test results for small-world and scale-free structure

doubly distributed

moderate very high ∑

smal

l wor

ld

yes N = 10 r2 = .74

N = 4 r2 = .75

14

no N = 27 r2 = .71

N = 38 r2 = .69

65

∑ 37 42 79

Consequently, we expected to find many instances of small-world networks in our

sample of 273 open source software ecosystems. We applied a test procedure similar to

Telesford et al. (2011), and found only 67 small-world networks, or 25% of the sample.

Even more unexpected, the sample shows that the most highly distributed networks are

likely not small-world, as shown in Table 21. Out of those 42 ecosystems that lie within

the 4th quartile of both resources distribution (≥163 artifacts per ecosystem) and control


distribution (≥14,378 forks per ecosystem), only 4 (10%) passed the small-world test. In

contrast, out of the 37 ecosystems occupying the 1st quartile of each distribution

dimension, 10 (27%) ecosystems can be considered small-world. Because small-world

networks tend to be scale-free (Valverde et al. 2002), we tested each ecosystem also for

this property. The average values for r2 are given in Table 21, and they corroborate the

finding that hierarchically modular structure is not a common property of highly

distributed ecosystems.

An illustration of the difference in structure is given with Table 20 above. Overall, this

example hints at the possible existence of different patterns of how doubly distributed

networks produce innovative outcomes. On the left-hand side, the ecosystem we call

‘rails/rails’ is shown, which is not small-world. It is clearly dominated by a single

artifact, ‘rails/rails’, which is the Ruby on Rails framework for web development.

Judging by the depicted network graph, we may assume that this artifact is a digital

platform on which other artifacts rely (cf. Tiwana et al. 2010). For instance, we find an

artifact that extends Ruby on Rails with authentication capabilities (devise); one that

adds a data administration interface (activeadmin); and an artifact which enables file

compression (sprockets). The most highly coupled artifact in this ecosystem maintains

links to 435 other artifacts, or to 48% of the ecosystem population. In follow-up research

it will be of interest to investigate to which extent the central artifact unilaterally

influences its environment, or whether such influence is mutual, that is how co-evolution

of a digital platform ecosystem topology plays out between the central digital platform

and its surrounding modules. Research on digital platform ecosystem innovation

dynamics is in its infancy, but suggests that both governance (Eaton et al. 2015) and

functionality (Um et al. 2015) of the digital platform jointly shape these dynamics.

In contrast, the ‘MinecraftForge/MinecraftForge’ ecosystem, presented on the right-

hand side of Table 20, is a small-world network. This ecosystem mostly consists of

software artifacts that modify the popular video game Minecraft. Among others, we find

an artifact (in jargon: a mod) that adds virtual oil pipelines (BuildCraft); an artifact that

facilitates installation of other mods (MinecraftForge); and an artifact which connects

separate instances of Minecraft servers (BungeeCord). The most highly coupled artifact

maintains links to 105 other artifacts, or to 16% of the ecosystem population. Visually,

this ecosystem resembles a network which might have emerged through wakes of

innovation, wherein autonomous activities in one part of the network lead to

repercussions in other parts (cf. Boland et al. 2007). The absence of a digital platform

as central reference point increases ecosystem complexity (cf. Hanseth and Lyytinen


2010), raising the question how social actors coordinate their disparate activities.

Research on individual open source software artifacts suggests that coordination is

managed by means of modularization (MacCormack et al. 2006), partitioning of tasks

until they can be handled by a single person (Howison and Crowston 2014), and

incremental as opposed to punctuated changes to the source code (Scacchi 2004). It will

be of interest to investigate to which extent these results scale up from the artifact to the

ecosystem level.

III.6 Instead of a conclusion: plans for further research

This research is guided by an interest to identify and explain the dynamics of

autonomous behavior leading to distributed innovation in open source software

ecosystems. This paper reports how a dataset of 273 ecosystems, amassing 41,388

software artifacts between them, was constructed from digital trace data captured on

Github. We contribute to the body of literature on distributed innovation in digital

ecosystems (Yoo et al. 2010) in two ways. Empirically, we show that highly distributed

open source software ecosystems cannot be expected to be small-world; a surprising

conclusion given previous results (Albert and Barabási 2002). Methodologically, we

outline three exploratory data loops, which provide useful guidance on how to

systematically approach an initially overwhelming set of digital trace data in early stages

of research.

Going forward, we plan advances in four major areas. First, we intend to analyze

sequences of activities in doubly distributed ecosystems, similar to the approach

proposed by Lindberg et al. (2013), but on a considerably larger scale. We expect to find

significant differences in complexity and variety of action sequences, depending on

varying coordination requirements in ecosystems of different topology. Furthermore,

we expect to discern diverse strategies that lead to comparable outcomes with regard to

relevant success measures (cf. Gresov and Drazin 1997). Second, we intend to

considerably refine the ecosystem graph model. Crucially, we plan to add the notions of

growth and decay, for instance by modelling the addition/removal of software artifacts

due to activity/inactivity (cf. Hanseth and Lyytinen 2010). Third, we aim to explicitly

consider social actors and their interdependencies as part of an ecosystem (Woodard and

Clemons 2014). In particular, we expect to find explanations for how social and

technical subsystems co-evolve (Henfridsson et al. 2014). Fourth, we plan to make the

dataset, as well as accompanying tools, available to the public at a later stage of the

research program. In summary, we expect that further empirical analyses and more


substantiated theorizing, supported by the dataset presented in this paper, will lead to

considerably better-founded and more fine-grained conclusions than those we were able

to present here.

Part B: Coordination across open source software communities: findings from the Rails ecosystem 89

IV Coordination across open source software communities:

findings from the Rails ecosystem

Table 22. Bibliographic information for Article IV

Title Coordination Across Open Source Software Communities: Findings from the Rails Ecosystem

Authors Alexander Eck

Outlet MKWI 2018 Proceedings

Year 2018

Status Published

Abstract. While coordination of work within open source software (OSS) communities

is well-researched, it is virtually unknown how they coordinate across community

boundaries. However, as OSS projects are often part of a larger digital ecosystem of

interdependent artifacts and communities, cross-community coordination is a pertinent

topic. We turn to the ecosystem around Ruby on Rails to empirically explore this

research gap. To this end, we scrutinize 96 coordination episodes among five

interrelated OSS projects and identify four cross-community coordination mechanisms:

adaptation, upgrading, positioning, and departure. Each mechanism describes a distinct

and stable arrangement to integrate contributions across community borders. After

presenting our findings, we reason about the significance of the results on explaining

generative change in digital ecosystems.

Keywords: cross-community coordination, open source software, digital ecosystems,

generative change, case study

90 Part B: Coordination across open source software communities: findings from the Rails ecosystem

IV.1 Introduction

The open source software (OSS) model of designing and changing complex artifacts

based on a publicly accessible codebase (Raymond 1999) has created new forms of

coordinating which have been investigated thoroughly during the last two decades

(Crowston et al. 2012). Despite this work, little is known about coordination across

communities of interdependent artifacts, as opposed to coordination within the

community around an individual OSS artifact (Faraj et al. 2016). Cross-community

coordination is of practical importance, because many OSS artifacts are coupled with

other artifacts to form larger digital ecosystems and as such their communities have to

interact (Selander et al. 2013). One example is the ecosystem around Ruby on Rails, a

web application framework. Thousands of add-on modules extend utility of this artifact

(RubyGems 2017). Cross-community coordination is also of theoretical relevance, as it

provides an apt setting to study the phenomenon of generative change, that is

unanticipated change produced by contributions from broad and varied audiences

(Zittrain 2008), which is believed to drive innovation in the digital age (Yoo 2013).

Therefore, we ask:

How do open source software communities coordinate with other communities to

change their respective artifacts within a shared ecosystem?

OSS communities usually introduce change through granular, incremental steps

(Howison and Crowston 2014). We harness this mode of working to empirically inquire

coordination episodes. By recognizing patterns among these granular episodes, we

identify mechanisms with which actors from different communities integrate their

respective contributions – that is, we identify cross-community coordination

mechanisms that lead to change, or potential for change, in OSS artifacts.

Our findings from five communities that are part of the Ruby on Rails ecosystem suggest

four mechanisms. First, cross-community coordination reflects the hierarchy between

artifacts. If one artifact serves as the basis of the second artifact, it is likely that the

community of the latter readily seeks to adapt to any relevant change of the outside

artifact. Second, communities anticipate fundamental changes to an artifact they depend

on by contemplating possible consequences and upgrading their artifact accordingly.

Third, cross-community coordination often revolves around locating and eliminating a

design flaw that is identified through artifact coupling. The communities work together

to position the appropriate location for fixing the design flaw. Fourth, members of an


outside community may argue for a departure from existing artifact design, based on

experiences within the context of their own artifact.

The remainder of this paper is organized as follows. We review the relevant literature

on coordination across open source software communities. We then present the research

design and findings of the empirical study, followed by a discussion how the findings

fit into the broader topic of generative change in digital ecosystems. The paper

concludes with a summary of main limitations and directions for further research.

IV.2 Background

IV.2.1 Coordination, coordination episodes, and coordination mechanisms

Coordination refers to the extent to which actors that need to integrate their respective

contributions due to interdependencies do so consistently and coherently (Faraj and

Xiao 2006). A major purpose of an OSS project, i.e. an OSS artifact and the community

supporting it, is to advance artifact design via source code contributions (Grewal et al.

2006). Therefore, a large part of coordination consists of managing change in software

artifacts through resolving interdependencies that emanate from source code

contributions (Crowston et al. 2007). Consequently, this paper investigates coordination

episodes during which possible source code contributions are discussed and integrated.

A coordination episode is a logically connected series of activities with a trigger-

activities-resolution structure (Annabi et al. 2008): a trigger that prompts coordination

need; sequential coordination activities between actors during which knowledge is

exchanged and source code contributions are discussed; and an eventual resolution

whether and what kind of change is implemented. We can expect that over time, OSS

communities develop stable arrangements to coordinate, or distinct coordination

mechanisms (Okhuysen and Bechky 2009). It is reasonable to believe that coordination

mechanisms exist also when two OSS communities repeatedly interact to integrate

contributions across community boundaries, which makes it possible to identify

mechanisms for cross-community coordination.

IV.2.2 Coordination mechanisms across communities of open source software

Often, a software artifact is coupled with other artifacts to enable and augment its own

capabilities (e.g., Haefliger et al. 2008). As such, it is typically part of a larger digital

ecosystem, or a set of interdependent, co-evolving artifacts and the communities

supporting them (Selander et al. 2013). Artifacts belonging to an active ecosystem

commonly change, which induces the regular need to integrate contributions originating


from outside communities. Such cross-community coordination is difficult, because the

OSS projects making up a digital ecosystem are heterogeneous and distributed, both in

terms of resources and in terms of control (cf. Lyytinen et al. 2016). For example, social

practices how artifacts are changed may differ between communities, and whether an

outside artifact changes so that depending artifacts are forced to be adapted is ultimately

beyond the control the focal community.

Extant research on cross-community coordination mainly highlights the role of

boundary spanners, i.e. actors who participate in multiple projects and who are capable

to integrate heterogeneous contributions and reconcile dependencies between

distributed artifacts (Daniel and Stewart 2016; Singh 2010; Vasilescu et al. 2016; Weiss

et al. 2006). Hence, boundary spanners coordinate by leveraging their personal

knowledge, not through processes that span actors in disparate communities.

Beyond the attention given to the boundary spanning role, there is little research on

cross-community coordination. This might be attributed to the property of software

artifacts to be loosely coupled via defined interfaces (cf. Baldwin and Clark 2000).

Because interfaces codify the rules and protocols of exchange, they commonly avoid the

need for direct interaction (Ghazawneh and Henfridsson 2013). Yet, interfaces neither

resolve all interdependencies nor do they remain unchanged over time (cf. Bogart et al.

2016). For those situations in which coordination is necessary, we found weak cues for

two distinct coordination mechanisms in the literature, which we label adaptation and

departure. First, a community may adapt its artifact when an interdependent outside

artifact changes, which is a rather transactional mechanism: a community accepts the

change produced by another community and adapts accordingly (Bogart et al. 2016;

Decan et al. 2016; Gonzalez-Barahona et al. 2009). Second, a more involved process is

to advocate a departure from existing artifact design. This mechanism exploits the

affordance of OSS communities to receive contributions from outsiders: a community

influences another one to deviate from its existing artifact design and implement a

change, usually one that helps the outsiders themselves (Bogart et al. 2016). Overall,

our knowledge of cross-community coordination is limited. In the following, we report

on an exploratory case study, addressing this research gap.

IV.3 Research design and methods

We conducted an explanatory multiple case study of cross-community coordination in

the digital ecosystem around Ruby on Rails (or simply Rails), an artifact that simplifies

development of web applications written in the Ruby programming language. The goal


of this examination was to describe and explain how disparate communities coordinate

to change their OSS artifacts in light of interdependencies. Our choice of the Rails

ecosystem was based on three main considerations. First, Rails was designed to be

coupled with other artifacts written in the Ruby programming language – Ruby gems in

jargon – which created the potential for cross-community coordination needs. Second,

with its history reaching back to 2004 (Hansson 2017), both the codebase and the

community of Rails, and in extension many related projects, can be assumed to be

mature. This was important to us, because we did not intend to trace the dynamics of

how coordination mechanisms emerge, but rather identify established coordination

mechanisms. Third, through reconstructing many digital ecosystems, including the Rails

ecosystem (Eck and Uebernickel 2016a), we noticed that the Rails artifact maintained

an unusually large number of couplings with other artifacts. This made Rails an ideal

starting point to investigate cross-community coordination in OSS settings.

IV.3.1 Data collection

We selected a sample of five OSS projects that are part of the Rails ecosystem, listed in

Table 23, based on four major considerations. First, the project had to be hosted on

Github, a popular service for collaborative OSS development, as we collected empirical

data from there. Second, we wanted to ensure that the projects had many contributors,

measured by Github forks (Biazzini and Baudry 2014), to increase the chances of

observing coordination. Third, we chose popular projects, measured by Github stars

accumulated (Borges et al. 2016). As an added benefit, these criteria ensured that the

selected projects were relatively mature, because it takes time to accumulate a sizable

number of forks and stars.


Table 23. Artifacts included in the case study

Artifact Description Forks Stars Ep First activity Last activity

Ruby on Rails (04 Nov 2008)

Framework for web application development

13,073 32,195 74 24 Aug 2009 06 Jul 2016

Devise (16 Sep 2009)

Provides advanced authentication functionality

3,499 15,576 38 05 Apr 2010 29 Jun 2016

Activeadmin (15 Apr 2010)

Simplifies creation of admin interfaces for web apps

2,497 6,763 41 01 Aug 2011 11 Jul 2016

Kaminari (06 Feb 2011)

Provides pagination functionality for web apps

779 5,909 17 04 Mar 2011 08 Jan 2016

Formtastic (07 Apr 2008)

Domain-specific language for designing web-based forms

604 4,894 22 24 Aug 2009 11 Jul 2016

Artifact Artifact name (in brackets: day artifact became available on Github) Forks Number of forks, a measure of contributing community size (as of 15 Jul 2016) Stars Number of stars, a measure of artifact popularity (as of 15 Jul 2016) Ep Total number of episodes with artifact participation (after exclusion criteria) First/last activity Start day of first episode and end day of last episode with artifact participation

Fourth, based on our knowledge of projects belonging to the Rails ecosystem (Eck and

Uebernickel 2016a) and employing self-written software, we systematically searched

the discussion threads of projects in the Rails ecosystem, with data ranging from 29 Oct

2007 to 01 Jul 2016. Specifically, we captured discussions during which an outside

project was referenced, and took this referencing as cue that two artifacts were coupled,

which potentially led to the need of repeated cross-community coordination (Faraj et al.

2016). Consider this example: In June 2011, an issue on the Devise discussion board

triggered a coordination episode: “I have just updated devise 1.4.0 => 1.4.1 in my

project, and now it throws an exception”. After some causal theorizing, the root cause

was identified: “it’s broken by this commit in Rails: rails@0ca69ca”. Notably, a

reference to Rails (that follows a well-defined syntax) was included. Less than two hours

after the initial issue description, Devise was adapted to accommodate the change made

by the Rails community, which resolved the episode.

We counted the number of outside references, and retained pairs of projects that shared

many references. For example, we counted 109 instances in which either Rails or Devise

referenced the other project. In total, the five OSS projects that we included in the case

study shared 394 references between them. We then manually inspected the full

discussion threads that contained those references. A discussion thread on Github

typically revolves around a specific and usually self-contained issue for which

coordination is needed to sort it out (Lindberg et al. 2016). In our data sample, we


observed that an issue discussion may spill over to another project, or that it may span

multiple threads within the same project. Therefore, we followed the activities

pertaining an issue even if it spanned multiple discussion threads and grouped them

together. Overall, we collected 245 grouped discussion threads, which we regarded as

tentative empirical evidence for cross-community coordination episodes, our unit of

analysis. To ensure that the data indeed represented episodes of cross-community

coordination, we inductively elaborated a list of exclusion criteria. For example, we

excluded an episode if it was still ongoing at time of data collection (see Table 248).

After applying the exclusion criteria, we were left with 96 episodes of cross-community

coordination.

Table 24. Exclusion criteria and number of excluded episodes

Exclusion criterium Description Matches

No influence Outside artifact was changed independent of this episode, or root cause was artifact-internal design flaw.

64

User mistake User mistake identified as root cause, or reported issue could not be reproduced.

44

Third artifact Dismissed initial causal theory and identified third artifact coupling as root cause for issue.

24

No resolution Episode was ongoing at time of data collection (July 15, 2016). 12

Other Just version number updated, or only one actor involved in episode, or episode did not aim to introduce any change.

5

Initial number of episodes 245

Episodes after exclusion criteria were applied 96

IV.3.2 Construct operationalization

The selected episodes served as basis for our analysis of how cross-community

coordination is carried out in the Rails ecosystem and what effect it has on the artifacts

involved. Our unit of analysis were individual coordination episodes. Therefore, we

needed to operationalize the trigger of an episode, individual coordination activities, and

the eventual episode resolution, as summarized in Table 25.

To operationalize the trigger construct we relied on open coding and axial grouping

(Corbin and Strauss 2008), eliciting five possible triggers from our data.

8 This table is not part of the published paper due to space limitations but included here for increased clarity.


As for coordination activities, we drew upon the coding book of Lindberg et al. (2016),

which distinguishes between knowledge integration and direct implementation.

Knowledge integration comprises activities needed to compile and integrate knowledge

among the involved actors. Direct implementation covers activities needed to create and

evaluate source code that may change the artifact. Just as in Lindberg et al. (2016) our

dataset showed other activities as well, such as acknowledgment (e.g., “You, sir, are

awesome.”) which we did not analyze as they were not material to our research question.

Each coordination activity was performed by an actor, whose actor status we classified

as either insider or outsider following a simple heuristic: We counted how often an actor

contributed to each community across our full dataset and made her member of the

community to which she contributed most. Then we compared assigned membership of

this actor (e.g., Rails) with the community in which she performed a coordination

activity (e.g., Devise). Matching pairs were coded as insider, mismatches were coded as

outsider.

Table 25. Operationalization of constructs and number of data points per construct

Construct Operationalization Count

Trigger Identified trigger: issue, not offering a possible root cause (N=16); issue, offering a possible root cause (49); anticipated outside change may require change in own artifact (18); outside artifact offers new functionality (3); change proposal (10).

96

Coordination activity Discussion entry that was coded as either knowledge integration (709) or direct implementation (599).

1,308

Actor status Actor was either insider (1,178) or outsider (130) of the community in which she performed the coordination activity.

1,308

Resolution Identified resolution: business rules change (47); software properties change (29); workaround found (5); no change (15).

96

Changed artifact Episode was resolved by changing: …upstream artifact (16); …downstream artifact (53); …both artifacts (7); none (20).

96

Episode length Number of activities in a coordination episode. 96

Episode duration Duration of coordination episode. 96

Actors involved Number of different actors who participated in an episode. 96

Files involved Number of files for which source code changes were discussed. 96

KI ratio Share of knowledge integration activities per episode. 96

OA ratio Share of activities performed by outsiders per episode. 96

KI: knowledge integration; OA: outsider activity


A coordination episode in our context resolves by (not) implementing a source code

change. To describe the resolution in more detail we were interested to learn also about

the kind of (non-)change that resolved the episode. Therefore, we drew upon Chapin et

al. (2001), who distinguish two types of source code change: change of business rules

are changes to the behavior of a software artifact such as fixing a design flaw, whereas

change of software properties alters nonfunctional metrics of a software artifact such as

increasing the readability of source code. Furthermore, we knew from prior literature

that agreeing on a workaround, i.e. a temporary and localized solution that does not

change the artifact, was yet another way to resolve a coordination episode (Ma et al.

2017).

In addition, we captured which of the two artifacts involved in a coordination episode

were changed: upstream, downstream, both, or none, following common terms in

software engineering (e.g., Bogart et al. 2016). An upstream artifact is one on which

others depend, while the depending artifact is called downstream. From manual

inspection of the five artifacts in our case study we established the upstream/downstream

relationships, depicted in Figure 10. For example, Rails is an upstream artifact to Devise.

By analogy, the Rails community is upstream to the Devise community.

Figure 10. Artifact dependencies and number of episodes per artifact pair

We elicited distinct cross-community coordination mechanisms through identifying

patterns across coordination episodes. To support this process we selected six additional

constructs that characterized the episodes as a whole (cf. Annabi et al. 2008). Episode

length captures the number of coded activities in a coordination episode. Episode

duration measures how much time passed between the first and the last activity in a

coordination episode. Actors involved counts how many different actors were involved

in an episode. Files involved counts the number of files for which source code changes

were discussed during the episode. Knowledge integration ratio sets the number of

knowledge integration activities in relation to the number of all activities in an episode.

Finally, outsider activity ratio measures the percentage of activities that outsiders

performed.


IV.3.3 Data analysis

Data analysis consisted of two logical stages of coordination episode coding and

coordination mechanism identification, through which we moved back and forth as we

learnt more about the empirical setting. First, we coded coordination episodes according

to the scheme summarized in Table 25. It struck us how straightforward the coding

scheme could be applied. For example, when source code was changed it was easy to

see whether functionality was added or fixed, readability was increased, etc. We

attribute this to two reasons: In OSS communities it is customary to carry out fine-

grained activities so that other community members can easily comprehend them

(Howison and Crowston 2014). In addition, the coordination episodes were usually

focused on solving the specific issue at hand and did not divert into other directions.

This demonstrated a certain level of discipline across the studied communities, which

was an indicator that stable arrangements to coordinate, i.e. coordination mechanisms,

existed.

Second, we identified coordination mechanisms in cross-community settings by seeking

commonalities and differences between the individual episodes and matching them with

preliminary coordination mechanism descriptions. We would switch to this stage of

analysis whenever we identified tentative patterns or peculiarities that were potentially

helpful to meet our research objective. As prior research mentioned adaptation and

departure mechanisms, these were our natural starting points for candidate mechanisms,

for which we wrote initial descriptions. We found evidence for these mechanisms and

discovered two additional coordination mechanisms, which we eventually labeled

upgrading and positioning.

IV.4 Findings: mechanisms of cross-community coordination

Through analyzing 96 episodes of cross-community coordination, our empirical analysis

unearthed four mechanisms that could comprehensively explain the observed

coordination work. Summary statistics for each mechanism is provided with Table 26.

In what follows, we show how each mechanism describes a different arrangement of

cross-community coordination.


Table 26. Cross-community coordination mechanisms

Trigger Activities Resolution

Mechanism N I Length Duration Actors Files KI ratio OA ratio C B U

Adaptation 43 72% 91% 42% 0%

Upgrading 7 14% 86% 43% 0%

Positioning 33 100% 91% 76% 67%

Departure 13 0% 8% 8% 8%

N Number of episodes attributed to mechanism I Percentage of episodes that were triggered by an issue (with or without root cause analysis) Length Number of activities per episode (histogram categories: 2-5; 6-20; 21-50; 51-92) Duration Duration per episode in days (<1; 1-10; 10-100; 100-1,297) Actors Number of different actors involved per episode (2-3; 4-8; 9-16; 17-24) Files Number of source code files involved per episode (<2; 2-4; 5-9; 10-87) KI ratio Share of knowledge integration activities per episode (<25%; 25%-50%; 50%-75%, >75%) OA ratio Share of outsider activities per episode (<5%; 5%-50%; 50%-95%, >95%) C Percentage of episodes that resolved with an artifact change B Percentage of episodes that resolved with a change of business rules U Percentage of episodes with change in upstream artifact

IV.4.1 The adaptation mechanism

Our data analysis shows how one community seeks to adapt its artifact to changes in an

upstream artifact as to restore functionality or utilize new functionality, which we define

as adaptation mechanism. Typically, a coordination episode of this type is triggered by

a malfunction, and its root cause is found to be a recent change in an upstream artifact.

This instigates change in the downstream artifact, which resolves the initially reported

issue. The adaptation mechanism leverages hierarchy of artifact couplings as its logic to

coordinate work across communities. The upstream community introduces changes

without reaching out to downstream communities. Instead, it assumes that downstream

artifacts will be adapted to restore compatibility. This is an efficient arrangement for the

upstream community, because it mitigates the costs of verbal exchange with outside

communities. It is also efficient for the downstream communities, given that most

adaptation episodes resolve after few activities and involve few actors. Furthermore, the

low number of files involved in an adaptation episode indicates that the necessary

adaptations are rather small, which in turn implies that the changes to the upstream

artifact cannot be radical as well.


IV.4.2 The upgrading mechanism

Our data analysis reveals how one community seeks to understand major changes in an

upstream artifact and how it upgrades the downstream artifact to appropriate these

changes, which we define as upgrading mechanism. Typically, a coordination episode

of this type is triggered by a collection of upcoming major changes (e.g., changes to

interfaces) that the community of an upstream artifact announced. After discussing the

implications for its own artifact, the downstream community systematically changes its

artifact to account for these implications. The upgrading mechanism resembles the

adaptation mechanism, as it follows the same logic of artifact hierarchy. However, it is

distinct in that coordination activities typically commence in anticipation of change to

the upstream artifact, as opposed to a usually reactive adaptation episode. This

mechanism is efficient for the upstream community, as indicated by the very low share

of outsider activities. Thus, the coordination burden is mostly on the downstream

community. Not only must it implement many changes to its artifact, it also has to spend

effort on collating and integrating knowledge, as evidenced by the comparatively high

knowledge integration ratio of most upgrading episodes.

IV.4.3 The positioning mechanism

Our data analysis surfaces how two communities seek to track down an issue observed

from coupling their respective artifacts and introduce change at the position they jointly

agree on, which we define as positioning mechanism. Typically, a coordination episode

of this type is triggered by an issue that is observed from the combination of upstream

and downstream artifacts. Both communities then investigate why the issue happens,

debate which artifact should be changed and implement the changes deemed

appropriate. The positioning mechanism is distinct as it always sees one community

reaching out to another due to an issue it cannot solve on its own. Creating a shared

understanding between the two communities is a major concern, suggested by the

comparatively high knowledge integration ratio combined with a high outsider activities

ratio. A positioning episode mainly increases artifact quality, because a previously

unknown issue is identified and resolved: 76% of all changes were to the functionality

of the artifact, usually fixing a design flaw. Remarkably, the main beneficiary is the

upstream artifact, as 67% of all positioning episodes in our data resolved with changes

to the upstream artifact. This suggests that a growing pool of downstream artifacts

effectively increases the number of collaborators who contribute with technical

expertise to the upstream artifact beyond community borders.


IV.4.4 The departure mechanism

Our data analysis exposes how an actor, based on experiences made with a coupled

artifact, proposes a change and the focal community decides whether and how to

implement the change despite it being a departure from existing design, which we define

as departure mechanism. Typically, a coordination activity of this type has few

activities, involves just a couple of actors, and no or just one source code file. Usually,

the initially proposed departure of artifact design is not accepted by the focal

community. The departure mechanism incorporates what is commonly considered a

hallmark of open source software: anybody may modify an OSS artifact and see this

change become part of the ‘official release’ if the community deems it beneficial (Faraj

et al. 2016). However, in our case study of five rather mature OSS projects, fewer than

15% of all episodes could be attributed to the departure mechanism. What is more, all

episodes but one resolved with a rejection of the proposed change by the focal

community, and typically did so after few coordination activities. This suggests that

OSS communities managing mature artifacts are reluctant to depart from existing

artifact design.

IV.5 Discussion: coordination as means of generative change

By analyzing how the communities of five interdependent OSS projects in the Rails

ecosystem resolve coordination needs stemming from changes in coupled artifacts, we

identified four cross-community coordination mechanisms, namely adaptation,

upgrading, positioning, and departure. Each mechanism describes a mode of integrating

contributions from coupled OSS projects that instigate change in the focal artifact. Thus,

cross-community coordination organizes the co-evolutionary production of change

caused by broad and varied audiences absent of central control, that is cross-community

coordination is a means of generative change in a digital ecosystem (Zittrain 2008).

The strength of a generative system lies in its capacity to allow contributions from

heterogeneous actors, and turn these contributions into productive changes. Because

they are the result of a creative dialogue between insiders and outsiders, some of these

changes are unanticipated and innovative (Zittrain 2008). Due to perpetual

incompleteness of digital artifacts (Garud et al. 2008) and their ability to be coupled

with other incomplete artifacts to form new combinations and variations, digital

ecosystems produce constant co-evolutionary changes. Arguably, this dynamic is a main

driver of digital innovation (Yoo 2013).


To this end, the identified cross-community coordination mechanisms help us better

rationalize and explain digital innovation processes in complex digital ecosystems,

particularly in an OSS setting. The adaptation and upgrading mechanisms show how

upstream OSS projects effectively influence the evolutionary trajectory of downstream

projects. In turn, the positioning mechanism demonstrates how autonomous, yet

interdependent OSS communities create synchronization opportunities to sort out design

flaws for mutual benefit. Finally, the departure mechanism suggests that mature OSS

projects have a strong sense of direction, which reduces the propensity of radical design

changes. Overall, the existence of coordination mechanisms that operate across OSS

communities are further proof of the self-organizing capability of heterogeneous

systems characterized by distributed resources and distributed control (Lyytinen et al.

2016).

IV.6 Limitations and conclusion

The contributions presented here are naturally limited by our research approach and the

scope of our data, which points at opportunities for future research. First, we cannot

claim generalizability of the findings beyond the studied context. Therefore, additional

research that studies different empirical contexts might complement our results. Second,

stemming from our selection of mature communities, we did not regard path

dependencies between the individual episodes. Future research might identify dynamic

aspects of cross-community coordination. Third, we did not observe coordination

activities directly, but derived them from verbalized contributions on discussion boards.

Hence, future research could employ ethnographic methods to draw a richer picture of

the micro-foundations of cross-community coordination. Fourth, we obtained data only

from the discussion boards on Github. Other tools to coordinate work are available, such

as email, chat, and (virtual) whiteboards. As such, further examinations could tap into

alternative data sources to identify coordination mechanisms that we did not observe.

In summary, much of the existing research on cross-community coordination assumes

that boundary spanners and loose coupling are sufficient to resolve interdependencies

between OSS projects. In this paper, we employed an explanatory case study to elicit

four mechanisms for coordinating work across interdependent OSS communities, which

is the main accomplishment of this paper. In addition, we argued that cross-community

coordination mechanisms are means of generative change in digital ecosystems, and

thus help us better explain processes of digital innovation.

References 103

References

Agarwal, R., Gupta, A. K., and Kraut, R. 2008. “The Interplay Between Digital and

Social Networks,” Information Systems Research (19:3), pp. 243–252.

AIS. 2011. Senior Scholars' Basket of Journals.

http://aisnet.org/?page=SeniorScholarBasket. Accessed 15 December 2017.

Albert, R., and Barabási, A.-L. 2002. “Statistical Mechanics of Complex Networks,”

Reviews of Modern Physics (74:1), pp. 47–97.

Ali, T. 2016. Essential Copying and Pasting from Stack Overflow.

https://www.gitbook.com/download/pdf/book/tra38/essential-copying-and-pasting-

from-stack-overflow. Accessed 15 December 2017.

Alter, S. 2015. “The concept of ‘IT artifact’ has outlived its usefulness and should be

retired now,” Information Systems Journal (25:1), pp. 47–60.

Annabi, H., Crowston, K., and Heckman, R. 2008. “Depicting What Really Matters:

Using Episodes to Study Latent Phenomenon,” ICIS 2008 Proceedings.

Archer, M. S. 1998. “Introduction: Realism in the Social Sciences,” in Critical

Realism: Essential Readings, M. S. Archer, R. Bhaskar, A. Collier, T. Lawson and

A. Norrie (eds.), London, New York: Routledge, pp. 189–205.

Argyris, C. 1976. “Single-Loop and Double-Loop Models in Research on Decision

Making,” Administrative Science Quarterly (21:3), pp. 363–375.

Arthur, W. B. 2009. The Nature of Technology: What It Is and How It Evolves, New

York: Free Press.

Avgerou, C. 2013. “Social Mechanisms for Causal Explanation in Social Theory

Based IS Research,” Journal of the Association for Information Systems (14:8),

399-419.

Avital, M., and Te'eni, D. 2009. “From Generative Fit to Generative Capacity:

Exploring an Emerging Dimension of Information Systems Design and Task

Performance,” Information Systems Journal (19:4), pp. 345–367.

Baldwin, C., MacCormack, A., and Rusnak, J. 2014. “Hidden Structure: Using

Network Methods to Map System Architecture,” Research Policy (43:8), pp. 1381–

1397.

Baldwin, C. Y., and Clark, K. B. 2000. Design Rules: The Power of Modularity,

Cambridge, Mass.: MIT Press.

Baldwin, C. Y., and Woodard, C. J. 2009. “The Architecture of Platforms: A Unified

View,” in Platforms, Markets, and Innovation, A. Gawer (ed.), Cheltenham, UK:

Edward Elgar, pp. 19–34.

104 References

Baptista, J. 2009. “Institutionalisation as a Process of Interplay Between Technology

and its Organisational Context of Use,” Journal of Information Technology (24:4),

pp. 305–319.

Barabási, A.-L., and Albert, R. 1999. “Emergence of Scaling in Random Networks,”

Science (286:5439), pp. 509–512.

Baskerville, R. L., and Myers, M. D. 2002. “Information Systems as a Reference

Discipline,” MIS Quarterly (26:1), pp. 1–14.

Baskerville, R. L., and Myers, M. D. 2015. “Design Ethnography in Information

Systems,” Information Systems Journal (25:1), pp. 23–46.

Basole, R. C. 2016. “Topological Analysis and Visualization of Interfirm collaboration

networks in the electronics industry,” Decision Support Systems (83), pp. 22–31.

Bauckhage, C. 2011. “Insights into Internet Memes,” (AAAI Conference on Weblogs

and Social Media 2011 Proceedings), pp. 42–49.

Beaumont, P. 2011. The Truth about Twitter, Facebook and the Uprisings in the Arab

World. http://www.theguardian.com/world/2011/feb/25/twitter-facebook-uprisings-

arab-libya. Accessed 15 December 2017.

Benkler, Y. 2006. The Wealth of Networks: How Social Production Transforms

Markets and Freedom, New Haven, CT: Yale University Press.

Berente, N., and Seidel, S. 2014. “Big Data & Inductive Theory Development:

Towards Computational Grounded Theory?” AMCIS 2014 Proceedings.

Berente, N., and Seidel, S. 2015. “Don't Pretend to Test Hypotheses: Lexicon and the

Legitimization of Empirically-Grounded Computational Theory Development,”

University of Georgia, University of Liechtenstein.

Berente, N., Srinivasan, N., Lyytinen, K., and Yoo, Y. 2008. “Design Principles for IT

in Doubly Distributed Design Networks,” ICIS 2008 Proceedings.

Bergvall-Kåreborn, B., and Howcroft, D. 2014. “Persistent Problems and Practices in

Information Systems Development: A Study of Mobile Applications Development

and Distribution,” Information Systems Journal (24:5), pp. 425–444.

Best, M. L. 2014. “The Internet that Facebook Built,” Communications of the ACM

(57:12), pp. 21–23.

Beynon-Davies, P. 2010. “The Enactment of Significance: A Unified Conception of

Information, Systems and Technology,” European Journal of Information Systems

(19:4), pp. 389–408.

Bharadwaj, A., El Sawy, O. A., Pavlou, P. A., and Venkatraman, N. 2013. “Digital

Business Strategy: Toward a Next Generation of Insights,” MIS Quarterly (37:2),

pp. 471–482.

References 105

Bhaskar, R. 1978. A Realist Theory of Science, Hassocks, Sussex: Harvester Press.

Bhaskar, R. 1998. “General Introduction,” in Critical Realism: Essential Readings, M.

S. Archer, R. Bhaskar, A. Collier, T. Lawson and A. Norrie (eds.), London, New

York: Routledge, pp. IX–XXIV.

Bhaskar, R. 2008. A Realist Theory of Science, London, New York: Routledge.

Biazzini, M., and Baudry, B. 2014. “May the Fork Be with You: Novel Metrics to

Analyze Collaboration on GitHub,” WETSoM 2014 Proceedings, pp. 37–43.

Biggs, J. 2008. Swype, Truly Gesture-Based Data Entry.

http://techcrunch.com/2008/09/09/tc50-swype-truly-gesture-based-data-entry/.

Accessed 15 December 2017.

Bird, C., Bachmann, A., Aune, E., Duffy, J., Bernstein, A., Filkov, V., and Devanbu,

P. 2009. “Fair and Balanced? Bias in Bug-Fix Datasets,” ESEC/FSE 2009

Proceedings, pp. 121–130.

Blincoe, K., Harrison, F., and Damian, D. 2015. “Ecosystems in GitHub and a Method

for Ecosystem Identification using Reference Coupling,” MSR 2015 Proceedings,

pp. 202–211.

Bogart, C., Kästner, C., Herbsleb, J., and Thung, F. 2016. “How to Break an API: Cost

Negotiation and Community Values in Three Software Ecosystems,” FSE 2016


Bohlin, L., Edler, D., Lancichinetti, A., and Rosvall, M. 2014. “Community Detection

and Visualization of Networks with the Map Equation Framework,” in Measuring

Scholarly Impact: Methods and Practice, Y. Ding, R. Rousseau and D. Wolfram

(eds.), Cham: Springer International Publishing, pp. 3–34.

Boland, R. J., Lyytinen, K., and Yoo, Y. 2007. “Wakes of Innovation in Project

Networks: The Case of Digital 3-D Representations in Architecture, Engineering,

and Construction,” Organization Science (18:4), pp. 631–647.

Borges, H., Hora, A., and Valente, M. T. 2016. “Understanding the Factors That

Impact the Popularity of GitHub Repositories,” ICSME 2016 Proceedings, pp.

334–344.

Bosch, J., and Bosch-Sijtsema, P. 2010. “From Integration to Composition: On the

Impact of Software Product Lines, Global Development and Ecosystems,” The

Journal of Systems and Software (83:1), pp. 67–76.

Bostrom, R. P., and Heinen, J. S. 1977. “MIS Problems and Failures: A Socio-

Technical Perspective. Part I: The Causes,” MIS Quarterly (1:3), pp. 17–32.

106 References

Boudreau, K. J. 2012. “Let a Thousand Flowers Bloom? An Early Look at Large

Numbers of Software App Developers and Patterns of Innovation,” Organization

Science (23:5), pp. 1409–1427.

Boudreaux, R. 2012. WordPress Platform for More than Blogging.

http://www.techrepublic.com/blog/web-designer/wordpress-platform-for-more-

than-blogging/. Accessed 15 December 2017.

Bourdieu, P. 1973. “The Three Forms of Theoretical Knowledge,” Social Science

Information (12:1), pp. 53–80.

Bourdieu, P. 1998. Practical Reason: On the Theory of Action, Stanford, Calif.:

Stanford University Press.

Brooks, F. P. 1986. “No Silver Bullet: Essence and Accidents of Software

Engineering,” Information Processing 1986 Proceedings, pp. 1069–1076.

Bygstad, B. 2015. “The Coming of Lightweight IT,” ECIS 2015 Proceedings.

Carlile, P. R. 2002. “A Pragmatic View of Knowledge and Boundaries: Boundary

Objects in New Product Development,” Organization Science (13:4), pp. 442–455.

Cecez-Kecmanovic, D. 2011. “Doing Critical Information Systems Research -

Arguments for a Critical Research Methodology,” European Journal of

Information Systems (20:4), pp. 440–455.

Chae, B., and Poole, M. S. 2005. “The Surface of Emergence in Systems

Development: Agency, Institutions, and Large-scale Information Systems,”

European Journal of Information Systems (14:1), pp. 19–36.

Chapin, N., Hale, J. E., Khan, K. M., Ramil, J. F., and Tan, W.‐G. 2001. “Types of

Software Evolution and Software Maintenance,” Journal of Software Maintenance

and Evolution: Research and Practice (13:1), pp. 3–30.

Chatterjee, S., and Sarker, S. 2013. “Infusing Ethical Considerations in Knowledge

Management Scholarship: Toward a Research Agenda,” Journal of the Association

for Information Systems (14:8), pp. 452–481.

Chesbrough, H. W. 2003. Open Innovation: The New Imperative for Creating and

Profiting from Technology, Boston, Mass.: Harvard Business School Press.

Chomsky, N. 1957. Syntactic Structures: Mouton & Co.

Chomsky, N. 1966. Topics in the Theory of Generative Grammar, The Hague:

Mouton.

Chomsky, N. 1986. Knowledge of Language: Its Nature, Origin, and Use, New York:

Praeger.

Chung, H. M., and Gray, P. 1999. “Special Section: Data Mining,” Journal of

Management Information Systems (16:1), pp. 11–16.

References 107

Ciborra, C. U. 1991. “From Thinking to Tinkering: The Grassroots of Strategic

Information Systems,” ICIS 1991 Proceedings, pp. 283–291.

Ciborra, C. U. (ed.). 2000. From Control to Drift: The Dynamics of Corporate

Information Infrastructures, Oxford: Oxford Univ. Press.

Clark, K. B. 1985. “The Interaction of Design Hierarchies and Market Concepts in

Technological Evolution,” Research Policy (14:5), pp. 235–251.

Coleman, J. S. 1986. “Social Theory, Social Research, and a Theory of Action,”

American Journal of Sociology (91:6), pp. 1309–1335.

Cook, S. D. N., and Brown, J. S. 1999. “Bridging Epistemologies: The Generative

Dance Between Organizational Knowledge and Organizational Knowing,”

Organization Science (10:4), pp. 381–400.

Cooper, H. M. 1998. Synthesizing Research: A Guide for Literature Reviews,

Thousand Oaks, Calif: Sage Publications.

Corbin, J. M., and Strauss, A. L. 2008. Basics of Qualitative Research: Techniques

and Procedures for Developing Grounded Theory, Los Angeles, Calif: Sage

Publications.

Crowston, K., Li, Q., Wei, K., Eseryel, U. Y., and Howison, J. 2007. “Self-

organization of Teams for Free/Libre Open Source Software Development,”

Information and Software Technology (49:6), pp. 564–575.

Crowston, K., Wei, K., Howison, J., and Wiggins, A. 2012. “Free/Libre Open-Source

Software Development: What We Know and What We Do Not Know,” ACM

Computing Surveys (44:2), Article 7.

Dabbish, L., Stuart, C., Tsay, J., and Herbsleb, J. 2012. “Social Coding in GitHub:

Transparency and Collaboration in an Open Software Repository,” CSCW 2012


Daniel, S., and Stewart, K. 2016. “Open Source Project Success: Resource Access,

Flow, and Integration,” The Journal of Strategic Information Systems (25:3), pp.

159–176.

David, P. A., and Greenstein, S. 1990. “The Economics of Compatibility

Standards: An Introduction to Recent Research,” Economics of Innovation and New

Technology (1:1-2), pp. 3–41.

Davis, P., and West, K. 2009. “What Do Public Values Mean for Public Action?

Putting Public Values in Their Plural Place,” The American Review of Public

Administration (39:6), pp. 602–618.

108 References

Decan, A., Mens, T., Claes, M., and Grosjean, P. 2016. “When GitHub Meets CRAN:

An Analysis of Inter-Repository Package Dependency Problems,” SANER 2016


Dobson, P., Jackson, P., and Gengatharen, D. 2013. “Explaining Broadband Adoption

in Rural Australia: Modes of Reflexivity and the Morphogenetic Approach,” MIS

Quarterly (37:3), pp. 965–991.

Doz, Y. L. 1996. “The Evolution of Cooperation in Strategic Alliances: Initial

Conditions or Learning Processes?” Strategic Management Journal (17:S1), pp.

55–83.

Eaton, B. 2012. The Dynamics of Digital Platform Innovation: Unfolding the Paradox

of Control and Generativity in Apple's iOS. PhD Thesis, London, UK.

Eaton, B., Elaluf-Calderwood, S., Sørensen, C., and Yoo, Y. 2015. “Distributed

Tuning of Boundary Resources: The Case of Apple's iOS Service System,” MIS

Quarterly (39:1), pp. 217–243.

Eck, A. 2018. “Coordination Across Open Source Software Communities: Findings

from the Rails Ecosystem,” MKWI 2018 Proceedings.

Eck, A., and Uebernickel, F. 2016a. “Reconstructing Open Source Software

Ecosystems: Finding Structure in Digital Traces,” ICIS 2016 Proceedings.

Eck, A., and Uebernickel, F. 2016b. “Untangling Generativity: Two Perspectives on

Unanticipated Change Produced by Diverse Actors,” ECIS 2016 Proceedings.

Eck, A., Uebernickel, F., and Brenner, W. 2015. “The Generative Capacity of Digital

Artifacts: A Mapping of the Field,” PACIS 2015 Proceedings.

El Sawy, O. A., Malhotra, A., Park, Y., and Pavlou, P. A. 2010. “Seeking the

Configurations of Digital Ecodynamics: It Takes Three to Tango,” Information

Systems Research (21:4), pp. 835–848.

Elaluf-Calderwood, S., Herzhoff, J., Sørensen, C., and Eaton, B. 2011. “Mobile Digital

Infrastructure Innovation: Towards a Tussle and Control Framework,” ECIS 2011

Proceedings.

Emery, F. E., and Trist, E. L. 1965. “The Causal Texture of Organizational

Environments,” Human Relations (18:1), pp. 21–32.

Erl, T. 2005. Service-Oriented Architecture: Concepts, Technology, and Design,

Upper Saddle River, NJ: Prentice Hall Professional Technical Reference.

Faraj, S., von Krogh, G., Monteiro, E., and Lakhani, K. R. 2016. “Special Section

Introduction—Online Community as Space for Knowledge Flows,” Information


References 109

Faraj, S., and Xiao, Y. 2006. “Coordination in Fast-Response Organizations,”

Management Science (52:8), pp. 1155–1169.

Faust, K. 2005. “Using Correspondence Analysis for Joint Displays of Affiliation

Networks,” in Models and Methods in Social Network Analysis, P. J. Carrington, J.

Scott and S. Wasserman (eds.), Cambridge: Cambridge University Press, pp. 117–

147.

Feldman, M. S., and Pentland, B. T. 2003. “Reconceptualizing Organizational

Routines as a Source of Flexibility and Change,” Administrative Science Quarterly

(48:1), pp. 94–118.

Fisher, G. 2012. “Effectuation, Causation, and Bricolage: A Behavioral Comparison of

Emerging Theories in Entrepreneurship Research,” Entrepreneurship Theory and

Practice (36:5), pp. 1019–1051.

Follett, M. P. 2013. “Power,” in Dynamic Administration: The Collected Papers of

Mary Parker Follett, H. C. Metcalf and L. F. Urwick (eds.), Mansfield Centre, CT.:

Martino Publishing, pp. 72–95.

Förderer, J., Kude, T., Schütz, S., and Heinzl, A. 2014. “Control versus Generativity:

A Complex Adaptive Systems Perspective on Platforms,” ICIS 2014 Proceedings.

Freelon, D. 2014. “On the Interpretation of Digital Trace Data in Communication and

Social Computing Research,” Journal of Broadcasting & Electronic Media (58:1),

pp. 59–75.

Galliers, R. D. 2006. “Commentary on Wanda Orlikowski's 'Material Knowing: The

Scaffolding of Human Knowledgeability',” European Journal of Information

Systems (15:5), pp. 470–472.

Garud, R., Jain, S., and Tuertscher, P. 2008. “Incomplete by Design and Designing for

Incompleteness,” Organization Studies (29:3), pp. 351–371.

Gaskin, J., Berente, N., Lyytinen, K., and Yoo, Y. 2014. “Toward Generalizable

Sociomaterial Inquiry: A Computational Approach for Zooming In and Out of

Sociomaterial Routines,” MIS Quarterly (38:3), pp. 849–A12.

Geipel, M. M. 2007. “Self-Organization Applied to Dynamic Network Layout,”

International Journal of Modern Physics C (18:10), pp. 1537–1549.

Gerring, J. 2012. Social Science Methodology: A Unified Framework, Cambridge,

New York: Cambridge University Press.

Ghazawneh, A., and Henfridsson, O. 2013. “Balancing Platform Control and External

Contribution in Third-party Development: The Boundary Resources Model,”

Information Systems Journal (23:2), pp. 173–192.

110 References

Girvan, M., and Newman, M. E.J. 2002. “Community Structure in Social and

Biological Networks,” PNAS (99:12), pp. 7821–7826.

GitHub. 2015. Msysgit/Git: Push to Gerrit via Https is Broken #332.

https://github.com/msysgit/git/issues/332#issuecomment-99059176. Accessed 15

December 2017.

GitHub. 2017a. Bootcamp: Fork a Repo. https://help.github.com/articles/fork-a-repo/.


GitHub. 2017b. Press Highlights. https://github.com/about/press. Accessed 15

December 2017.

GitHub. 2017c. REST API v3. https://developer.github.com/v3/. Accessed 15

December 2017.

GitHub. 2017d. The State of the Octoverse 2017. https://octoverse.github.com/.


GitHub. 2017e. Writing on GitHub: Autolinked References and URLs.

https://help.github.com/articles/autolinked-references-and-urls/. Accessed 15

December 2017.

Goldman, S. L., Nagel, R. N., and Preiss, K. 1995. Agile Competitors and Virtual

Organizations: Strategies for Enriching the Customer, New York: Van Nostrand

Reinhold.

Gonzalez-Barahona, J. M., Robles, G., Michlmayr, M., Amor, J. J., and German, D.

M. 2009. “Macro-level Software Evolution: A Case Study of a Large Software

Compilation,” Empirical Software Engineering (14:3), pp. 262–285.

Gousios, G. 2013. “The GHTorent Dataset and Tool Suite,” MSR 2013 Proceedings,

pp. 233–236.

Gresov, C., and Drazin, R. 1997. “Equifinality: Functional Equivalence in

Organization Design,” Academy of Management Review (22:2), pp. 403–428.

Grewal, R., Lilien, G. L., and Mallapragada, G. 2006. “Location, Location,

Location: How Network Embeddedness Affects Project Success in Open Source

Systems,” Management Science (52:7), pp. 1043–1056.

Grisot, M., Hanseth, O., and Thorseng, A. A. 2014. “Innovation of, in, on

Infrastructures: Articulating the Role of Architecture in Information Infrastructure

Evolution,” Journal of the Association for Information Systems (15:4), pp. 197–

219.

Grolemund, G., and Wickham, H. 2014. “A Cognitive Interpretation of Data

Analysis,” International Statistical Review (82:2), pp. 184–204.

References 111

Grover, V., and Lyytinen, K. J. 2015. “New State of Play in Information Systems

Research: The Push to the Edges,” MIS Quarterly (39:2), pp. 271–296.

Haefliger, S., von Krogh, G., and Spaeth, S. 2008. “Code Reuse in Open Source

Software,” Management Science (54:1), pp. 180–193.

Hallingby, H. K., Hartviksen, G., Elaluf-Calderwood, S., and Sørensen, C. 2016.

“Convergence in Action: A Case Study of the Norwegian Internet,” Telematics and

Informatics (33:2), pp. 641–649.

Haney, D. 2016. NPM & Left-Pad: Have We Forgotten How to Program?

http://www.haneycodes.net/npm-left-pad-have-we-forgotten-how-to-

program/?utm_source=hackernewsletter&utm_medium=email&utm_term=featured

. Accessed 15 December 2017.

Hanseth, O., and Aanestad, M. 2003. “Design as Bootstrapping. On the Evolution of

ICT Networks in Health Care,” Methods Archive (42:4), pp. 384–391.

Hanseth, O., and Lyytinen, K. 2010. “Design Theory for Dynamic Complexity in

Information Infrastructures: The Case of Building Internet,” Journal of Information

Technology (25:1), pp. 1–19.

Hansson, D. H. 2017. Ruby on Rails. http://rubyonrails.org/. Accessed 15 December

2017.

Harré, R., and Madden, E. H. 1975. Causal Powers: A Theory of Natural Necessity,

Totowa, N.J.: Rowman and Littlefield.

Hedman, J., Srinivasan, N., and Lindgren, R. 2013. “Digital Traces of Information

Systems: Sociomateriality Made Researchable,” ICIS 2013 Proceedings.

Hedström, P., and Swedberg, R. 1998. “Social Mechanisms: An Introductory Essay,”

in Social Mechanisms: An Analytical Approach to Social Theory, P. Hedström and

R. Swedberg (eds.), Cambridge: Cambridge Univ. Press, pp. 1–31.

Henfridsson, O., and Bygstad, B. 2013. “The Generative Mechanisms of Digital

Infrastructure Evolution,” MIS Quarterly (37:3), pp. 907–A5.

Henfridsson, O., Mathiassen, L., and Svahn, F. 2014. “Managing Technological

Change in the Digital Age: The Role of Architectural Frames,” Journal of

Information Technology (29:1), pp. 27–43.

Henningsson, S., and Henriksen, H. Z. 2011. “Inscription of Behaviour and Flexible

Interpretation in Information Infrastructures: The Case of European e-Customs,”

The Journal of Strategic Information Systems (20:4), pp. 355–372.

Howison, J., and Crowston, K. 2014. “Collaboration Through Open Superposition: A

Theory of the Open Source Way,” MIS Quarterly (38:1), pp. 29–50.

112 References

Howison, J., Wiggins, A., and Crowston, K. 2011. “Validity Issues in the Use of

Social Network Analysis with Digital Trace Data,” Journal of the Association for


Huang, J., Newell, S., Huang, J., and Pan, S.-L. 2014. “Site-shifting as the Source of

Ambidexterity: Empirical Insights from the Field of Ticketing,” Information

Systems Strategy-as-Practice (23:1), pp. 29–44.

Huysman, M. H., Fischer, S. J., and Heng, Michael S. H. 1994. “An Organizational

Learning Perspective on Information Systems Planning,” The Journal of Strategic


Iannacci, F. 2014. “Routines, Artefacts and Technological Change: Investigating the

Transformation of Criminal Justice in England and Wales,” Journal of Information

Technology (29:4), pp. 294–311.

Iansiti, M., and Levien, R. 2004. “Strategy as Ecology,” Harvard Business Review

(82:3), pp. 68–78.

Itami, H., and Numagami, T. 1992. “Dynamic Interaction between Strategy and

Technology,” Strategic Management Journal (13:S2), pp. 119–135.

Jackson, P. 2015. “Networks in a Digital World: A Cybernetics Perspective,” ECIS

2015 Proceedings.

Jain, R. P., and Ramesh, B. 2015. “The Roles of Contextual Elements in Post-Merger

Common Platform Development: An Empirical Investigation,” European Journal

of Information Systems (24:2), pp. 159–177.

Jansen, S., Finkelstein, A., and Brinkkemper, S. 2009. “A Sense of Community: A

Research Agenda for Software Ecosystems,” ICSE 2009 Proceedings, pp. 187–190.

Jarvenpaa, S., and Tuunainen, V. 2013. “Theoretical Elaboration of IT Enablement

Model in the Era of Customer and Community Digital Innovation,” ECIS 2013

Proceedings.

Jazayeri, M. 2007. “Some Trends in Web Application Development,” FOSE 2007


Johnson, S., Faraj, S., and Kudaravalli, S. 2014. “Emergence of Power Laws in Online

Communities: The Role of Social Mechanisms and Preferential Attachment,” MIS

Quarterly (38:3), pp. 795–808.

Kalliamvakou, E., Gousios, G., Blincoe, K., Singer, L., German, D. M., and Damian,

D. 2016. “An In-depth Study of the Promises and Perils of Mining GitHub,”

Empirical Software Engineering (21:5), pp. 2035–2071.

Kallinikos, J., Aaltonen, A., and Marton, A. 2013. “The Ambivalent Ontology of

Digital Artifacts,” MIS Quarterly (37:2), pp. 357–370.

References 113

Kasthurirathna, D., and Piraveenan, M. 2015. “Emergence of Scale-Free

Characteristics in Socio-Ecological Systems with Bounded Rationality,” Scientific

Reports (5), 10448.

Klecum, E., Lichtner, V., Cornford, T., and Petrakaki, D. 2014. “Evaluation as a

Multi-Ontological Endeavour: A Case from the English National Program for IT in

Healthcare,” Journal of the Association for Information Systems (15:3), 147-176.

Klein, H. K., and Myers, M. D. 1999. “A Set of Principles for Conducting and

Evaluating Interpretive Field Studies in Information Systems,” MIS Quarterly

(23:1), pp. 67–93.

Kogut, B., and Metiu, A. 2001. “Open‐Source Software Development and Distributed

Innovation,” Oxford Review of Economic Policy (17:2), pp. 248–264.

Krainin, J., and Lawrence, M. R. 1990. Memory & Imagination: New Pathways to the

Library of Congress.

Kretzer, M., and Maedche, A. 2014. “Generativity of Business Intelligence

Platforms: A Research Agenda Guided by Lessons from Shadow IT,” MKWI 2014


Kretzer, M., Maedche, A., and Gass, O. 2014. “Barriers to BI&A Generativity: Which

Factors impede Stable BI&A Platforms from Enabling Organizational Agility?”

AMCIS 2014 Proceedings.

Kutsch, E., Denyer, D., Hall, M., and Lee-Kelley, E. 2013. “Does Risk Matter?

Disengagement from Risk Management Practices in Information Systems

Projects,” European Journal of Information Systems (22:6), pp. 637–649.

Kwok, R. W., Lee, J. N., Huynh, M. Q., and Pi, S. M. 2002. “Role of GSS on

Collaborative Problem-based Learning: A Study on Knowledge Externalisation,”

European Journal of Information Systems (11:2), pp. 98–107.

Lambert, R., and Peppard, J. 1993. “Information Technology and New Organizational

Forms: Destination But no Road Map?” The Journal of Strategic Information

Systems (2:3), pp. 180–206.

Lancichinetti, A., and Fortunato, S. 2009. “Community Detection Algorithms: A

Comparative Analysis,” Physical Review E (80:5), 56117.

Landwerth, I. 2014. .NET Core is Open Source.

https://blogs.msdn.microsoft.com/dotnet/2014/11/12/net-core-is-open-source/.


Lane, D., and Maxfield, R. 1996. “Strategy under Complexity: Fostering Generative

Relationships,” Long Range Planning (29:2), pp. 215–231.

114 References

Langley, A. 1999. “Strategies for Theorizing from Process Data,” Academy of

Management Review (24:4), pp. 691–710.

Latour, B. 2010. “Tarde’s Idea of Quantification,” in The Social after Gabriel Tarde:

Debates and Assessments, M. Candea (ed.), London, New York: Routledge, pp.

145–162.

Law, J., and Singleton, V. 2005. “Object Lessons,” Organization (12:3), pp. 331–355.

Lee, G. K., and Cole, R. E. 2003. “From a Firm-Based to a Community-Based Model

of Knowledge Creation: The Case of the Linux Kernel Development,”

Organization Science (14:6), pp. 633–649.

Lee, J., and Berente, N. 2012. “Digital Innovation and the Division of Innovative

Labor: Digital Controls in the Automotive Industry,” Organization Science (23:5),

pp. 1428–1447.

Lee, J., Wyner, G. M., and Pentland, B. T. 2008. “Process Grammar as a Tool for

Business Process Design,” MIS Quarterly (32:4), pp. 757–778.

Leen, G., and Heffernan, D. 2002. “Expanding Automotive Electronic Systems,”

Computer (35:1), pp. 88–93.

Leonardi, P. M. 2010. “Digital Materiality? How Artifacts Without Matter, Matter,”

First Monday.

Leonardi, P. M. 2011. “When Flexible Routines Meet Flexible Technologies:

Affordance, Constraint, and the Imbrication of Human and Material Agencies,”

MIS Quarterly (35:1), pp. 147–168.

Leonardi, P. M. 2012. “Materiality, Sociomateriality, and Socio-Technical Systems:

What Do These Terms Mean? How Are They Different? Do We Need Them?” in

Materiality and Organizing: Social Interaction in a Technological World, P. M.

Leonardi, B. A. Nardi and J. Kallinikos (eds.), Oxford: Oxford University Press,

pp. 25–48.

Lindberg, A. 2015. The Origin, Evolution, and Variation of Routine Structures in

Open Source Software Development: Three Mixed Computational-Qualitative

Studies, Cleveland, Ohio.

Lindberg, A., Berente, N., Gaskin, J., and Lyytinen, K. 2016. “Coordinating

Interdependencies in Online Communities: A Study of an Open Source Software

Project,” Information Systems Research (27:4), pp. 751–772.

Lindberg, A., Berente, N., Gaskin, J., Lyytinen, K., and Yoo, Y. 2013. “Computational

Approaches for Analyzing Latent Social Structures in Open Source Organizing,”

ICIS 2013 Proceedings.

References 115

Lindberg, A., Xiao, X., and Lyytinen, K. 2014. “Theorizing Modes of Open Source

Software Development,” HICSS 2014 Proceedings.

Lowry, P. B., Moody, G. D., Gaskin, J., Galletta, D. F., Humpherys, S., Barlow, J. B.,

and Wilson, D. W. 2013. “Evaluating Journal Quality and the Association for

Information Systems (AIS) Senior Scholars’ Journal Basket via Bibliometric

Measures: Do Expert Journal Assessments Add Value?” MIS Quarterly (37:4), pp.

993–1012.

Lungu, M. F. 2009. Reverse Engineering Software Ecosystems, Lugano.

Lusch, R. F., and Nambisan, S. 2015. “Service Innovation: A Service-Dominant Logic

Perspective,” MIS Quarterly (39:1), pp. 155–176.

Lyytinen, K., and Newman, M. 2008. “Explaining Information Systems Change: A

Punctuated Socio-technical Change Model,” European Journal of Information

Systems (17:6), pp. 589–613.

Lyytinen, K., and Yoo, Y. 2002. “Research Commentary: The Next Wave of Nomadic

Computing,” Information Systems Research (13:4), pp. 377–388.

Lyytinen, K., Yoo, Y., and Boland, R. J. 2016. “Digital Product Innovation Within

Four Classes of Innovation Networks,” Information Systems Journal (25:1), 47–75.

Lyytinen, K. J. 1985. “Implications of Theories of Language for Information

Systems,” MIS Quarterly (9:1), pp. 61–74.

Ma, K.-L., and Muelder, C. W. 2013. “Large-Scale Graph Visualization and

Analytics,” Computer (46:7), pp. 39–46.

Ma, W., Chen, L., Zhang, X., Zhou, Y., and Xu, B. 2017. “How Do Developers Fix

Cross-project Correlated Bugs? A Case Study on the GitHub Scientific Python

Ecosystem,” ICSE 2017 Proceedings, pp. 381–392.

MacCormack, A., Rusnak, J., and Baldwin, C. Y. 2006. “Exploring the Structure of

Complex Software Designs: An Empirical Study of Open Source and Proprietary

Code,” Management Science (52:7), pp. 1015–1030.

Maguire, S., McKelvey, B., Mirabeau, L., and Oztas, N. 2006. “Complexity Science

and Organization Studies,” in The SAGE Handbook of Organization Studies, S. R.

Clegg, C. Hardy, T. B. Lawrence and W. R. Nord (eds.), London: Sage

Publications, pp. 165–215.

Majchrzak, A., and Malhotra, A. 2013. “Towards an Information Systems Perspective

and Research Agenda on Crowdsourcing for Innovation,” The Journal of Strategic


Manikas, K. 2016. “Revisiting Software Ecosystems Research: A Longitudinal

Literature Study,” Journal of Systems and Software (117), pp. 84–103.

116 References

Manikas, K., and Hansen, K. M. 2013. “Software Ecosystems – A Systematic

Literature Review,” Journal of Systems and Software (86:5), pp. 1294–1306.

Markus, M. L., and Silver, M. S. 2008. “A Foundation for the Study of IT Effects: A

New Look at DeSanctis and Poole's Concepts of Structural Features and Spirit,”

Journal of the Association for Information Systems (9:10/11), pp. 609–632.

McBride, N. 2005. “Chaos Theory as a Model for Interpreting Information Systems in

Organizations,” Information Systems Journal (15:3), pp. 233–254.

McLeod, L., and Doolin, B. 2012. “Information Systems Development as Situated

Socio-technical Change: A Process Approach,” European Journal of Information

Systems (21:2), pp. 176–191.

Mens, T. 2008. “Introduction and Roadmap: History and Challenges of Software

Evolution,” in Software Evolution, T. Mens and S. Demeyer (eds.), Berlin,

Heidelberg: Springer-Verlag, pp. 1–11.

Merali, Y. 2006. “Complexity and Information Systems: the emergent domain,”

Journal of Information Technology (21:4), pp. 216–228.

Meuter, M. L., Bitner, M. J., Ostrom, A. L., and Brown, S. W. 2005. “Choosing

Among Alternative Service Delivery Modes: An Investigation of Customer Trial of

Self-service Technologies,” Journal of Marketing (69:2), pp. 61–83.

Mingers, J., Mutch, A., and Willcocks, L. 2013. “Critical Realism in Information

Systems Research,” MIS Quarterly (37:3), pp. 795–802.

Mintzberg, H., and Waters, J. A. 1985. “Of Strategies, Deliberate and Emergent,”

Strategic Management Journal (6:3), pp. 257–272.

Mossinger, J. 2010. “Software in Automotive Systems,” IEEE Software (27:2), pp.

92–94.

Müller, O., Junglas, I., vom Brocke, J., and Debortoli, J. 2016. “Utilizing Big Data

Analytics for Information Systems Research: Challenges, Promises and

Guidelines,” European Journal of Information Systems (25:4), pp. 289–302.

Myers, C. R. 2003. “Software Systems as Complex Networks: Structure, Function, and

Evolvability of Software Collaboration Graphs,” Physical Review E (68:4), 46116.

Myers, M. D., and Klein, H. K. 2011. “A Set of Principles for Conducting Critical

Research in Information Systems,” MIS Quarterly (35:1), pp. 17–36.

Nambisan, S. 2013. “Information Technology and Product/Service Innovation: A Brief

Assessment and Some Suggestions for Future Research,” Journal of the

Association for Information Systems (14:4), pp. 215–226.

Naumann, F. 2014. “Data Profiling Revisited,” ACM SIGMOD Record (42:4), pp. 40–

49.

References 117

Nielsen, P., and Hanseth, O. 2010. “Towards a Design Theory of Usability and

Generativity,” ECIS 2010 Proceedings.

Njenga, K., and Brown, I. 2012. “Conceptualising Improvisation in Information

Systems Security,” European Journal of Information Systems (21:6), pp. 592–607.

Nocaj, A., Ortmann, M., and Brandes, U. 2014. “Untangling Hairballs: From 3 to 14

Degrees of Separation,” Symposium on Graph Drawing 2014 Proceedings, pp.

101–112.

Oestreicher-Singer, G., and Zalmanson, L. 2013. “Content or Community? A Digital

Business Strategy for Content Providers in the Social Age,” MIS Quarterly (37:2),

pp. 591–616.

Okhuysen, G. A., and Bechky, B. A. 2009. “Coordination in Organizations: An

Integrative Perspective,” The Academy of Management Annals (3:1), pp. 463–502.

Orlikowski, W. J. 1992. “The Duality of Technology: Rethinking the Concept of

Technology in Organizations,” Organization Science (3:3), pp. 398–427.

Orlikowski, W. J. 1996. “Improvising Organizational Transformation Over Time: A

Situated Change Perspective,” Information Systems Research (7:1), pp. 63–92.

Orlikowski, W. J. 2000. “Using Technology and Constituting Structures: A Practice

Lens for Studying Technology in Organizations,” Organization Science (11:4), pp.

404–428.

Orlikowski, W. J. 2006. “Material Knowing: The Scaffolding of Human

Knowledgeability,” European Journal of Information Systems (15:5), pp. 460–466.

Otto, T., and Smith, R. C. 2013. “Design Anthropology: A Distinct Style of

Knowing,” in Design Anthropology, W. Gunn, T. Otto and R. C. Smith (eds.),

London, New York: Bloomsbury Academic, pp. 1–32.

Oxford English Dictionary. 2009. Generative, adj.

http://www.oed.com/viewdictionaryentry/Entry/77523. Accessed 15 December

2017.

Pang, M.-S., Lee, G., and Delone, W. H. 2014. “IT Resources, Organizational

Capabilities, and Value Creation in Public-Sector Organizations: A Public-Value

Management Perspective,” Journal of Information Technology (29:3), pp. 187–205.

Parnas, D. L. 1972. “On the Criteria to Be Used in Decomposing Systems into

Modules,” Communications of the ACM (15:12), pp. 1053–1058.

Pentland, B. T. 1999. “Building Process Theory with Narrative: From Description to

Explanation,” The Academy of Management Review (24:4), pp. 711–724.

118 References

Pentland, B. T., and Feldman, M. S. 2008. “Designing Routines: On the Folly of

Designing Artifacts, While Hoping for Patterns of Action,” Information and

Organization (18:4), pp. 235–250.

Pentland, B. T., and Rueter, H. H. 1994. “Organizational Routines as Grammars of

Action,” Administrative Science Quarterly (39:3), pp. 484–510.

Post, D. G. 2010. “The Theory of Generativity,” Fordham Law Review (78:6), pp.

2755–2766.

Pozzebon, M., and Pinsonneault, A. 2012. “The Dynamics of Client-Consultant

Relationships: Exploring the Interplay of Power and Knowledge,” Journal of

Information Technology (27:1), pp. 35–56.

Prieto, I. M., and Easterby-Smith, M. 2006. “Dynamic Capabilities and the Role of

Organizational Knowledge: An Exploration,” European Journal of Information

Systems (15:5), pp. 500–510.

Racherla, P., and Mandviwalla, M. 2013. “Moving from Access to Use of the

Information Infrastructure: A Multilevel Sociotechnical Framework,” Information


Rai, A. 2016. “Editor's Comments: Synergies Between Big Data and Theory,” MIS

Quarterly (40:2), pp. iii–ix.

Raymond, E. 1999. “The Cathedral and the Bazaar,” Knowledge, Technology & Policy

(12:3), pp. 23–49.

Reimers, K. 1996. “The Non-Market Preconditions of Electronic Markets:

Implications for their Evolution and Applicability,” European Journal of


Reuver, M. de, Sørensen, C., and Basole, R. C. 2017. “The Digital Platform: A

Research Agenda,” Journal of Information Technology.

Ries, E. 2011. The Lean Startup: How Today's Entrepreneurs Use Continuous

Innovation to Create Radically Successful Businesses, New York: Crown Business.

Roberts, L. M. 2006. “Response: Shifting the Lens on Organizational Life: The Added

Value of Positive Scholarship,” The Academy of Management Review (31:2), pp.

292–305.

Robey, D., Anderson, C., and Raymond, B. 2013. “Information

Technology, Materiality, and Organizational Change: a Professional Odyssey,”

Journal of the Association for Information Systems (14:7), pp. 379–398.

Rosvall, M., and Bergstrom, C. T. 2008. “Maps of Random Walks on Complex

Networks Reveal Community Structure,” PNAS (105:4), pp. 1118–1123.

References 119

Rowe, F. 2014. “What Literature Review is not: Diversity, Boundaries and

Recommendations,” European Journal of Information Systems (23:3), pp. 241–

255.

RubyGems. 2017. Gems. https://rubygems.org/gems. Accessed 15 December 2017.

Sambamurthy, V., Bharadwaj, A., and Grover, V. 2003. “Shaping Agility through

Digital Options: Reconceptualizing the Role of Information Technology in

Contemporary Firms,” MIS Quarterly (27:2), pp. 237–263.

Sambamurthy, V., and Zmud, R. W. 2000. “Research Commentary: The Organizing

Logic for an Enterprise's IT Activities in the Digital Era—A Prognosis of Practice

and a Call for Research,” Information Systems Research (11:2), pp. 105–114.

Santana, F. W., and Werner, C. M. L. 2013. “Towards the Analysis of Software

Projects Dependencies: An Exploratory Visual Study of Software Ecosystems,”

IWSECO 2013 Proceedings, pp. 1–12.

Scacchi, W. 2004. “Free and Open Source Development Practices in the Game

Community,” IEEE Software (21:1), pp. 59–66.

Schultze, U., and Boland, R. J. 2000. “Knowledge Management Technology and the

Reproduction of Knowledge Work Practices,” The Journal of Strategic Information

Systems (9:2-3), pp. 193–212.

Selander, L., Henfridsson, O., and Svahn, F. 2013. “Capability Search and Redeem

Across Digital Ecosystems,” Journal of Information Technology (28:3), pp. 183–

197.

Senge, P. M. 1992. “You Can't Get There from Here: Why Systems Thinking is

Inseparable from Learning Organizations,” Massachusetts Institute of Technology

(MIT).

Shapiro, C., and Varian, H. R. 1999. Information Rules: A Strategic Guide to the

Network Economy, Boston, MA: Harvard Business School Press.

Shawe-Taylor, J. 1987. “Information and its Relation to Formalisms for the

Complexities of the Real World,” Journal of Information Technology (2:3), pp.

151–155.

Simon, H. A. 1962. “The Architecture of Complexity,” Proceedings of the American

Philosophical Society (106:6), pp. 467–482.

Simon, H. A. 1996. The Sciences of the Artificial, Cambridge, Mass.: MIT Press.

Singh, P. V. 2010. “The Small-World Effect: The Influence of Macro-Level Properties

of Developer Collaboration Networks on Open-Source Project Success,” ACM

Transactions on Software Engineering and Methodology (TOSEM) (20:2), Article

6.

120 References

Singh, P. V., Tan, Y., and Mookerjee, V. 2011. “Network Effects: The Influence of

Structural Capital on Open Source Project Success,” MIS Quarterly (35:4), pp.

813–830.

Solé, R. V., Ferrer‐Cancho, R., Montoya, J. M., and Valverde, S. 2002. “Selection,

Tinkering, and Emergence in Complex Networks,” Complexity (8:1), pp. 20–33.

Star, S. L., and Ruhleder, K. 1996. “Steps Toward an Ecology of Infrastructure:

Design and Access for Large Information Spaces,” Information Systems Research

(7:1), pp. 111–134.

Strong, D., Volkoff, O., Johnson, S., Pelletier, L., Tulu, B., Bar-On, I., Trudel, J., and

Garber, L. 2014. “A Theory of Organization-EHR Affordance Actualization,”

Journal of the Association for Information Systems (15:2), pp. 53–85.

Svahn, F. 2012. Digital Product Innovation: Building Generative Capability through

Architectural Frames. Dissertation, Umeå.

Svahn, F., and Henfridsson, O. 2012. “The Dual Regimes of Digital Innovation

Management,” HICSS 2012 Proceedings, pp. 3347–3356.

Swan, J. 2006. “Commentary on Wanda Orlikowski's 'Material Knowing: The

Scaffolding of Human Knowledgeability',” European Journal of Information

Systems (15:5), pp. 467–469.

Sydow, J., Schreyögg, G., and Koch, J. 2009. “Organizational Path

Dependence: Opening the Black Box,” Academy of Management Review (34:4), pp.

689–709.

Taroni, F., Biedermann, A., Bozza, S., Garbolino, P., and Aitken, C. 2014. Bayesian

Networks for Probabilistic Inference and Decision Analysis in Forensic Science,

Hoboken: Wiley.

Telesford, Q. K., Joyce, K. E., Hayasaka, S., Burdette, J. H., and Laurienti, P. J. 2011.

“The Ubiquity of Small-World Networks,” Brain Connectivity (1:5), pp. 367–375.

The Economist. 2003. “Open-source Software: Microsoft at the Power Point,”

The Economist.

The Economist. 2015a. “Blockchains: The Great Chain of Being Sure About Things,”

The Economist.

The Economist. 2015b. “Microsoft at Middle Age: Opening Windows,”

The Economist.

The Economist. 2017. “Buttonwood: Tech Stocks Have Regained their Dotcom-Era

Highs,” The Economist.

Thorseng, A., and Jensen, T. B. 2015. “Building National Infrastructures for Patient-

Centred Digital Services,” ECIS 2015 Proceedings.

References 121

Tilson, D., Lyytinen, K., and Sørensen, C. 2010. “Digital Infrastructures: The Missing

IS Research Agenda,” Information Systems Research (21:4), pp. 748–759.

Tiwana, A., Konsynski, B., and Bush, A. A. 2010. “Platform Evolution: Coevolution

of Platform Architecture, Governance, and Environmental Dynamics,” Information


Tiwana, A., Konsynski, B., and Venkatraman, N. 2013. “Special Issue: Information

Technology and Organizational Governance: The IT Governance Cube,” Journal of

Management Information Systems (30:3), pp. 7–12.

Trist, E. L., and Bamforth, K. W. 1951. “Some Social and Psychological

Consequences of the Longwall Method of Coal-Getting: An Examination of the

Psychological Situation and Defences of a Work Group in Relation to the Social

Structure and Technological Content of the Work System,” Human Relations (4:1),

pp. 3–38.

Truex, D. P., and Baskerville, R. 1998. “Deep Structure or Emergence Theory:

Contrasting Theoretical Foundations for Information Systems Development,”

Information Systems Journal (8:2), pp. 99–118.

Truex, D. P., Holmström, J., and Keil, M. 2006. “Theorizing in Information Systems

Research: A Reflexive Analysis of the Adoption of Theory in Information Systems

Research,” Journal of the Association for Information Systems (7:12), pp. 797–821.

Tukey, J. W. 1977. Exploratory Data Analysis, Reading, Mass.: Addison-Wesley.

Tukey, J. W. 1980. “We Need Both Exploratory and Confirmatory,” The American

Statistician (34:1), pp. 23–25.

Um, S., and Yoo, Y. 2016. “The Co-Evolution of Digital Ecosystems,” ICIS 2016

Proceedings.

Um, S., Yoo, Y., and Wattal, S. 2015. “The Evolution of Digital Ecosystems: A Case

of WordPress from 2004 to 2014,” ICIS 2015 Proceedings.

Um, S., Yoo, Y., Wattal, S., Kulathinal, R., and Zhang, B. 2013. “The Architecture of

Generativity in a Digital Ecosystem: A Network Biology Perspective,” ICIS 2013

Proceedings.

Valverde, S., Cancho, R. F., and Solé, R. V. 2002. “Scale-Free Networks from

Optimal Design,” Europhysics Letters (60:4), pp. 512–517.

Vasilescu, B., Blincoe, Kelly, Xuan, Qi, Casalnuovo, C., Damian, D., Devanbu, P.,

and Filkov, V. 2016. “The Sky is Not the Limit: Multitasking on GitHub Projects,”

ICSE 2016 Proceedings, pp. 994–1005.

Vassilakopoulou, P., and Grisot, M. 2012. “Technology Innovation in the Face of

Uncertainty: The Case of 'My Health Record',” ECIS 2012 Proceedings.

122 References

Venters, W., Oborn, E., and Barrett, M. 2014. “A Trichordal Temporal Approach to

Digital Coordination: The Sociomaterial Mangling of the CERN Grid,” MIS

Quarterly (38:3), pp. 927–949.

Volkoff, O., and Strong, D. M. 2013. “Critical Realism and Affordances: Theorizing

IT-Associated Organizational Change Processes,” MIS Quarterly (37:3), pp. 819–

834.

von Hippel, E. 1988. The Sources of Innovation, Oxford: Oxford Univ. Press.

von Hippel, E. 2005. Democratizing Innovation, Cambridge, Mass.: MIT Press.

von Hippel, E., and von Krogh, G. 2003. “Open Source Software and the “Private-

Collective” Innovation Model: Issues for Organization Science,” Organization

Science (14:2), pp. 209–223.

von Krogh, G. 2012. “How Does Social Software Change Knowledge Management?

Toward a Strategic Research Agenda,” The Journal of Strategic Information

Systems (21:2), pp. 154–164.

von Krogh, G., Haefliger, S., Spaeth, S., and Wallin, M. W. 2012. “Carrots and

Rainbows: Motivation and Social Practice in Open Source Software Development,”


von Krogh, G., and von Hippel, E. 2006. “The Promise of Research on Open Source

Software,” Management Science (52:7), pp. 975–983.

Walsh, I. 2014. “A Strategic Path to Study IT Use through Users’ IT Culture and IT

Needs: A Mixed-method Grounded Theory,” The Journal of Strategic Information

Systems (23:2), pp. 146–173.

Wareham, J., Fox, P. B., and Cano Giner, J. L. 2014. “Technology Ecosystem

Governance,” Organization Science (25:4), pp. 1195–1215.

Wasserman, S., Scott, J., and Carrington, P. J. 2005. “Introduction,” in Models and

Methods in Social Network Analysis, P. J. Carrington, J. Scott and S. Wasserman

(eds.), Cambridge: Cambridge University Press, pp. 1–7.

Watts, D. J. 2007. “A Twenty-First Century Science,” Nature (445:7127), p. 489.

Watts, D. J., and Strogatz, S. H. 1998. “Collective Dynamics of 'Small-World'

Networks,” Nature (393:6684), pp. 440–442.

Weiss, M., Moroiu, G., and Zhao, P. 2006. “Evolution of Open Source Communities,”

OSS 2006 Proceedings, pp. 21–32.

Weyns, D., and Ahmad, T. 2013. “Claims and Evidence for Architecture-Based Self-

adaptation: A Systematic Literature Review,” in Software Architecture, K. Drira

(ed.): Springer Berlin Heidelberg, pp. 249–265.

References 123

Whitley, E. A., and Darking, M. 2006. “Object Lessons and Invisible Technologies,”

Journal of Information Technology (21:3), pp. 176–184.

Williams, C., and Karahanna, E. 2013. “Causal Explanation in the Coordinating

Process: A Critical Realist Case Study of Federated IT Governance Structures,”


Wittrock, M. C. 1974. “Learning as a Generative Process,” Educational Psychologist

(11:2), pp. 87–95.

Woodard, C. J., and Clemons, E. K. 2014. “Modeling the Evolution of Generativity

and the Emergence of Digital Ecosystems,” ICIS 2014 Proceedings.

Woodard, C. J., Ramasubbu, N., Tschang, F. T., and Sambamurthy, V. 2013. “Design

Capital and Design Moves: The Logic of Digital Business Strategy,” MIS

Quarterly (37:2), pp. 537–564.

Wynn, D., and Williams, C. K. 2012. “Principles for Conducting Critical Realist Case

Study Research in Information Systems,” MIS Quarterly (36:3), pp. 787–810.

Yoo, Y. 2010. “Computing in Everyday Life: A Call for Research on Experiential

Computing,” MIS Quarterly (34:2), pp. 213–231.

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), pp. 227–236.

Yoo, Y., Boland, R. J., Lyytinen, K., and Majchrzak, A. 2012. “Organizing for

Innovation in the Digitized World,” Organization Science (23:5), pp. 1398–1408.

Yoo, Y., Henfridsson, O., and Lyytinen, K. 2010. “The New Organizing Logic of

Digital Innovation: An Agenda for Information Systems Research,” Information


Yoo, Y., Lyytinen, K., and Jr., R. J. B. 2008. “Distributed Innovation in Classes of

Networks,” HICSS 2008 Proceedings.

Zachariadis, M., Scott, S., and Barrett, M. 2013. “Methodological Implications of

Critical Realism for Mixed-Methods Research,” MIS Quarterly (37:3), pp. 855–

879.

Zhang, Z., Kulathinal, R., Wattal, S., and Yoo, Y. 2012. “Generative Diffusion of

Innovations: An Organizational Genetics Approach,” ICIS 2012 Proceedings.

Zhang, Z., Yoo, Y., Wattal, S., Zhang, B., and Kulathinal, R. 2014. “Generative

Diffusion of Innovations and Knowledge Networks in Open Source Projects,”

ICIS 2014 Proceedings.

Zimmer, M. 2010. “'But the Data is Already Public': On the Ethics of Research in

Facebook,” Ethics and Information Technology (12:4), pp. 313–325.

124 References

Zittrain, J. L. 2006. “The Generative Internet,” Harvard Law Review (119:7), pp.

1974–2040.

Zittrain, J. L. 2008. The Future of the Internet and How to Stop It, New Haven, CT:

Yale University Press.

Zittrain, J. L. 2009. “The End of the Generative Internet,” Communications of the

ACM (52:1), pp. 18–20.

Zollo, M., and Winter, S. G. 2002. “Deliberate Learning and the Evolution of Dynamic

Capabilities,” Organization Science (13:3), pp. 339–351.

Zucker, L. G., and Darby, M. R. 2005. “An Evolutionary Approach to Institutions and

Social Construction: Process and Structure,” in Great Minds in Management: The

Process of Theory Development, K. G. Smith and M. A. Hitt (eds.), Oxford: Oxford

University Press, pp. 572–588.

Publication list of the author XVII

Publication list of the author

A dissertation endeavor usually produces more research results than can become part of

the final thesis: adjacent topics are explored, digressions are abandoned, and cross-

pollinating research collaborations are entered. To give a comprehensive picture of

elaborated research, Table 27 lists all publications to which the author contributed

throughout the dissertation process. Besides the four articles included in this thesis

(highlighted like this), the overview includes additional journal articles, conference

articles and work-in-progress articles presented at international scholarly workshops.

Table 27. Comprehensive publication list with participation of the author

# Title Authors Outlet Year Type Status

1 Agile Delivery in Finance IT: Navigating in the Digital Age

Alexander Eck, Falk Uebernickel

IWI-HSG, University of St. Gallen

2014 S P

2 IT-unterstütztes Wissensmanagement im globalen Engineering

Olga Willner, Stefan Weber, Alexander Eck, Paul Schönsleben

Industrie Management (30:4), pp. 49–52

2014 J P

3 Fit for Continuous Integration: How Organizations Assimilate an Agile Practice

Alexander Eck, Falk Uebernickel, Walter Brenner

AMCIS 2014 Proceedings

2014 C P

4 Not all Information Systems are Created Equal: Exploring IT Resources for Agile Systems Delivery

Alexander Eck, Stefan Keidel, Falk Uebernickel, Thomas Schneider, Walter Brenner

Banking and Information Technology (15:3), pp. 21–34

2014 J P

5 Designing a Generative Digital Platform in Financial Services

Alexander Eck DESRIST 2015 Doctoral Consortium

2015 W L

6 Praktiken für das globale Engineering kundenindividueller Produkte

Olga Willner, Alexander Eck, Paul Schönsleben

IM + io (30:2), pp. 44–49

2015 J P

7 The Generative Capacity of Digital Artifacts: A Mapping of the Field

Alexander Eck, Falk Uebernickel, Walter Brenner

PACIS 2015 Proceedings

2015 C P

8 Agility Areas of Action in Finance IT – A Memorandum


IWI-HSG, University of St. Gallen

2015 S P

9 Organizational Control of Software Development Teams through Digital Platform Architecture

Alexander Eck 4th Innovation in Information Infrastructures Workshop

2015 W L

10 Exploring Generative Evolution in Open Source Software Ecosystems

Alexander Eck Social Study of IT Open Research Forum 2016

2016 W L

XVIII Publication list of the author

# Title Authors Outlet Year Type Status

11 Untangling Generativity: Two Perspectives on How Diverse Actors Create Unanticipated Change


ECIS 2016 Proceedings

2016 C P

12 Exploring How Digitized Products Enable Industrial Service Innovation – An Affordance Perspective

Matthias M. Herterich, Alexander Eck, Falk Uebernickel

ECIS 2016 Proceedings

2016 C P

13 Reconstructing Open Source Software Ecosystems: Finding Structure in Digital Traces


ICIS 2016 Proceedings

2016 C P

14 A Socio-Technical Approach to Study Consumer-Centric Information Systems

Benjamin Spottke, Alexander Eck, Jochen Wulf

ICIS 2016 Proceedings

2016 C P

15 Adapting New Capabilities or Enhancing Functionality? Two Sequence Patterns of Capability Redeem in Platform Ecosystems

Philipp Hukal, Alexander Eck

8th International Workshop on Software Ecosystems at ICIS 2016

2016 W P

16 Explaining Value Cocreation in Social Networking Services: Towards a Process Model of Resource Integration

Benjamin Spottke, Alexander Eck

Social Study of IT Open Research Forum 2017

2017 W L

17 Coordination Across Open Source Software Communities: Findings from the Rails Ecosystem

Alexander Eck MKWI 2018 Proceedings

2018 C P

18 Service Innovation in Social Networking Services: A Resource Integration Perspective on Facebook

Benjamin Spottke, Alexander Eck, Jochen Wulf

Under review at information systems journal (“A” tier, VHB JQ3)

2018 J R

Type Conference paper; Journal paper; Self-published working paper; Workshop paper Status Limited publication for presentation at workshop; Published; Review ongoing

Curriculum vitae XIX

Curriculum vitae

Personal information

Name Alexander Eck

Date of birth 28 April 1985

Place of birth Lippa

Nationality German

Education

2016 University of Warwick, United Kingdom

Visiting Research Student at Warwick Business School (WBS)

2013 – 2018 University of St. Gallen, Switzerland

Ph.D. Program in Management (Business Innovation)

2005 – 2011 Technische Universität Berlin, Germany

Dipl.-Ing. Program in Computer Sciences and Management

1995 – 2004 Ernst-Ludwig-Schule

Abitur (University Entrance Diploma)

Employment

Since 2016 Deloitte Digital, Germany

Senior Venture Architect / Manager

2013 – 2016 University of St. Gallen, Switzerland

Research Associate

2011 – 2013 European Center for ICT, Germany

Project Manager

generativity in digital ecosystems: how distributed ... · the contemporary digital age sees a...

Documents