Exploration of metaphors as a way to understand socio-technical phenomena: An emergent framework

Abstract

The intangible and (often) young nature of socio-technical phenomena makes them difficult to understand and communicate. Researchers and practitioners have responded by applying metaphors, prescribing an epistemological structure to these phenomena. While metaphors are frequently applied, researchers have paid limited attention to their applications and limitations. To address this gap, we applied a mixed-method approach, exploring the communities’ applications of metaphors for the conceptual development of socio-technical phenomena, using two communities: Open Government Data and IT Development and Maintenance. We synthesised 21 articles and two books into an emergent analytical framework, Communities’ Applications Of Metaphors (CAOM). We collected empirical material for each community’s academic and practical sides between 2015–2020, resulting in 100 articles and 263 documents. We conducted a word frequency analysis and an in-depth analysis of the empirical material, drawing on CAOM. The contributions are the emergent CAOM framework with the key concepts of metaphors’ usage, expression, and assemblages of metaphors. We conclude that the application of metaphors in communities is influenced by community type and metaphors’ role, while the topic influences metaphor selection. Metaphors are combined to create new ways of reasoning. We recommend that governments draw on action-oriented metaphors when writing policies for socio-technical phenomena, while digital government researchers should include action-oriented elements when they develop metaphors, which can help practitioners put new knowledge into practice.

Keywords

Metaphor community development socio-technical analytical framework open government data technical debtKey points for practitioners:•Communities of practitioners can approach metaphors as tools to discover new ways of working and acting departing from habits and traditional thinking;•Combining metaphors is essential to create new ways of reasoning and understanding socio-technical phenomena;•The emergent CAOM framework provides a base for combining and exploring metaphors.

1. Introduction

People’s continual work with and research of digital government bring about new intangible socio-technical phenomena (e.g., open government data and technical debt), requiring new vocabulary and thinking to collectively comprehend these young phenomena. However, people have a limited vocabulary and need to be economic in their usage of words (Ogden & Richards, 1923). An approach is to apply metaphors in the communication about socio-technical phenomena where an existing vocabulary and thinking are extended to encompass the young phenomena (e.g., archive, trash can, and desktop are office metaphors used for computers (Izwaini, 2003)). The application of metaphors has been common within research disciplines, such as information systems, management, and IT Development and Maintenance (e.g., Kendall & Kendall, 1993; Magnusson, Juiz, Gomez, & Bermejo, 2018; Morgan, 1998). For example, LNyhlen and Gidlund (2021) apply ‘Narratives’ as a syntax to examine policy documents, more precisely, government’s visions for digitization; Schedler, Guenduez, and Frischknecht (2019) apply ‘Smart’ as a frame to describe new digital governmental initiatives based on AI, IoT, big data analytic, and sensor systems; and Gerrits (2021) apply ‘Soul’ in the title as a rhetorical device to capture the receivers’ attention for the black-box of self-learning algorithms. While the above examples cover researchers, metaphors are also applied by practitioners (e.g., Accenture Federal Services, 2018; Carrara, Chan, Fischer, & Steenbergen, 2015; ODI, 2019).

A metaphor is an implicit or explicit assertion that something is like something else, bringing a specific structure to thinking (Lakoff & Johnson, 2008a; Morgan, 1998). Metaphors can help enrich communication by extending the vocabulary (Black, 1955; Lakoff & Johnson, 2008b) and can be viewed “as essential to human understanding and as a mechanism for creating new meaning and new realities in our lives” Lakoff and Johnson (2008b, p. 196). They can enable researchers and practitioners to understand and address phenomena with their preceding thinking (Mészáros, 1966). As a result, the application of metaphors can be beneficial for the conceptual development of new phenomena. On the other hand, the application of metaphors has some limitations. It is important to share sufficient knowledge to avoid misunderstanding the metaphors (Glucksberg, 1989; Ortony, 1975). The application of metaphors in ambiguous and knowledge-intensive situations with unclear purposes can hinder learning and lead to confusion (Hekkala, Stein, & Rossi, 2018). For example, ’Technical Debt’ explains the impact of suboptimal solutions made during the development of IT systems due to cuts in costs or time (Cunningham, 1992). It is a metaphor that describes a socio-technical phenomenon and the name of the said phenomenon, which has created confusion in the related conceptual development (Fairbanks, 2020). Metaphors can also strongly influence how people conceptualize and approach problems. People tend to search for information that confirms their metaphorical framing while seldom recognising the metaphors’ influence on their behaviours (Thibodeau & Boroditsky, 2011). Unsuccessful metaphors can generate confusion and despair (Ortony, 1975), while inappropriate metaphors can constrain thinking along undesirable lines. These metaphors embedded in formal communication may mislead or inhibit the learning of receivers (Hamilton, 2000). While previous digital government research has applied metaphors for conceptual development (e.g., Custers & Bachlechner, 2017; Smith & Sandberg, 2018; Zuiderwijk, Janssen, Van De Kaa, & Poulis, 2016), we lack knowledge about the said application of metaphors in communities of researchers and practitioners in the field of digital government. Hekkala et al. (2018) made a longitudinal study, illustrating how inappropriate applications of metaphors in an IS project led to problems, such as creating considerable uncertainty instead of order and hindering the learning of practitioners. As a result, we lack insights into if these digital government communities are reaping the benefits of metaphors in their conceptual developments, or if it limits them. A first step towards filling this knowledge gap is to explore the application of metaphors within these communities. This knowledge acts as a basis for future studies and debates concerning metaphors. It also provides researchers and practitioners with alternative ways they can apply metaphors in their conceptual developments.

The purpose of this article is to explore conceptual development in communities by studying researchers’ and practitioners’ applications of metaphors concerning socio-technical phenomena in their communication. We based our research on meme theory. This theory describes how concepts or ideas spread amongst people through imitation (Dawkins, 2016). Therefore, we synthesise literature on metaphors into an emergent analytical framework, which we call the ’Communities Application of Metaphors’ framework (CAOM framework). We select the Open Government Data (OGD) and Technical Debt (TD) communities as they are contemporary and active communities within research and practice, which are also available for us as researchers to investigate (see Subsubsection 3.2.1). OGD is data shared by public organisations for anyone to reuse without restrictions (Handbook, 2015). We study the communities over a period of five years and develop an emergent analytical framework based on meme theory (see Section 4). Our research was guided by the following research questions:

•
How have the OGD and TD communities applied metaphors in their conceptual development between 2015 and 2020, and why in this manner?
•
What are the lesson learned from analysing the application of metaphors in the OGD and TD communities?

The article starts with an overview of previous research and our research approach, then we present the CAOM framework. It continues by giving the quantitative and qualitative findings (in a section for each). We discuss the findings as the application of metaphors, studying metaphors, and the implications for researchers and practitioners. The article ends by presenting the conclusions with limitations and future research.
2. Background

This section presents the theoretical grounding for the emergent analytical framework synthesised in this research. It starts by explaining metaphors and then giving an approach to understanding how they spread within a community.

2.1 Metaphors

Information polity, digital government, and information system communities, in general, apply metaphors frequently to comprehend and communicate socio-technical phenomena. The anatomy of a metaphor is displayed in Fig. 1, where a sender and a receiver possess initial thoughts and initial words that they can apply to communicate the initial phenomenon to the receiver. A sender who encounters a new target phenomenon and experiences target thoughts about it but lacks target words can find it challenging to communicate. The sender can solve this problem by abstracting his thoughts about similar phenomena to produce a metaphor as a generalisation (Ogden & Richards, 1923).

Figure 1.

An anatomy of a metaphor.

The sender borrows initial words to apply as target words (metaphors). A metaphor also borrows partial thoughts of one phenomenon to apply to a target phenomenon, a form of generalisation. Morgan (1998) describes that “[ $\ldots$ a] metaphor proceeds through implicit or explicit assertions that A is (or is like) B.” (Morgan, 1998, p. 4). As such, metaphors are a conceptual approach to recycling and expanding previous knowledge to the unknown. The receiver infers similarities between an initial phenomenon and a target phenomenon from a metaphor that excludes others (Lakoff & Johnson, 2008a). The receiver must, therefore, sort out the differences and focuses on the similarities (Glucksberg, 1989).

However, some similarities may only be superficial, which can make a metaphor misleading (Mars, Bronstein, & Lusch, 2012). A metaphor aids by providing a specific perspective on a target phenomenon. However, the receiver’s understanding of a metaphor depends on his experiences with the initial phenomenon (Lakoff & Johnson, 2008a). A metaphor implies a way of thinking as it brings a specific structure to target thoughts based on initial thoughts and a way of understanding the world (Lakoff & Johnson, 2008a; Morgan, 1998). For example, technical debt refers to a debt that should be repaid; however, this is not always the case for technical debt. As a result, a sender and a receiver can evaluate a metaphor as beneficial or disadvantageous based on how well it conveys target thoughts about the target phenomenon.

2.2 Metaphors from a meme perspective

The application of metaphors in a community can be understood from the meme perspective (Dawkins, 2016). A meme is a convenient unit of cultural transmission (e.g., an idea or a tune) that can jump from one sender to another receiver through a process of imitation. A meme can be a metaphor, which we presume in this article. A metaphor spreads from sender to receiver through imitations where the metaphor can mutate or blend with other ideas (or metaphors). A metaphor spreads based on whether the receiver perceives the metaphor as having utility and appeal (Dawkins, 2016). A metaphor’s appeal relates to how well it can communicate a target phenomenon. The metaphor borrows the structure of the initial phenomenon to apply to the target phenomenon initially, and the focus is on the similarities between the initial phenomenon and the target phenomenon (Glucksberg, 1989; Lakoff & Johnson, 2008a; Ogden & Richards, 1923). A high-appeal metaphor entails being a well-known initial phenomenon and where the similarities between the initial and target phenomena are easily identified. The metaphor loses its appeal if the dissimilarities between the initial and target phenomenon create more problems than the similarities provide understanding. For example, the ‘e’ (electronic) is slowly dying as digital, online, and virtual is taking over (e.g., the eGovernment is becoming Digital Government). A beneficial and high-appeal metaphor can be characterised as self-explanatory (Ortony, 1975), while a low-appeal metaphor can generate confusion and despair (Ortony, 1975). However, a metaphor can have high appeal but also be disadvantageous, as such spreading quickly with consequences. On the other hand, a metaphor can lie dormant for years in various mediums (Dawkins, 2016), such as the metaphor “Technical debt”, which was first introduced in 1992 by Cunningham. It took a decade before it was picked up and discussed in the academic community (Tom, Aybuke, & Richard, 2012). A metaphor can co-adapt to ideas and other metaphors in its environment where associations are entrenched. It grows interdependent on other ideas and metaphors and can no longer exist without them. Possibly, they self-reinforce each other. This situation can lead to stable structures of assemblages of metaphors,1

¹
Originally, Dawkins (2016) called it complexes, but for clarity, we changed it to assemblages.

making it difficult for other metaphors to be adopted or accepted by receivers (Dawkins, 2016). In this article, we analyse the application of metaphors by examining how metaphors are used (usage), when they are used (expression), and the assemblages of metaphors used in connection with the metaphor in focus.

3. Research approach

Our study used a mixed-method approach to explore the application of metaphors within communities (see Creswell & Creswell, 2017). The mixed-method approach enables us to capture the applications of metaphors in different communities and to display the nuances of these applications and any related conceptual developments (Creswell & Creswell, 2017). It followed three phases: (1) we synthesised an analytical framework from previous research, (2) we chose and collected empirical material about two communities, and (3) we analysed the empirical material with word frequency analysis and then in-depth analysis based on the emergent analytical framework. Figure 2 presents an overview of our study.

Figure 2.

The research process.

3.1 Phase 1: Synthesis of an analytical framework

We used a hermeneutic literature review with citation tracking and citation analysis (Boell & Cecez-Kecmanovic, 2014) to review previous metaphor research. We combined the keywords “metaphor” or “metaphor development” with “communities” or “organisations” in Google Scholar. We skimmed titles, abstracts, and if they were relevant also the content. Then, we determined whether it would enable us to understand metaphors and their development. We classified relevant literature following title, purpose, research problem, keywords, limitations, implications, explanatory power, and interesting statements. In total, we classified 21 articles and 2 books.

We synthesised an emergent analytical framework based on the relevant literature. Its purpose is to analyse the application of metaphors by senders. An analytical framework, like a theoretical framework, draws concepts from a theory to shed light on a phenomenon Imenda (2014). We understand that an analytical framework leaves certain concepts open to be filled with empirical material and, as such, allows for the generation of new explanations and insights. The analytical framework guided and enabled us in explaining and interpreting the empirical material Imenda (2014). Moreover, the method of synthesis was concept mapping where we defined concepts and their relationships (Maxwell, 2012). This part of the phase started with an online discussion where we compared the relevant literature to identify concepts and relationships. This discussion was supported by a cloud-based diagram software where we could represent the framework. We continued to develop the framework drawing on the literature and when needed, we arranged for new discussions. In this process, several tentative frameworks were developed and discarded or merged based on perceived utility. We iterated between synthesis, data collection, and data analysis, allowing the final analytical framework to emerge over time. The outcome was the emergent analytical framework labelled the CAOM framework (see Section 4).

3.2 Phase 2: Data collection

We selected OGD and TD communities because of our existing knowledge of the two communities, and their accessibility. Additionally, these communities are interesting because of their differences in community characteristics, their active metaphor usage for a present socio-technical phenomenon, and the communities containing a mix of researchers and practitioners.

3.2.1 Introduction to the OGD and TD communities

We decided to study the OGD and TD communities because the communities are active both within research and practice and both want to communicate about a young, intangible socio-technical phenomenon. Thus, these communities enable us to study how metaphors have been applied (and help answer our research question) and explore their conceptual development (the purpose of this article). The OGD community is public, with key actors, and expresses itself abundantly on the Internet, while TD is less prominent on the Internet, contextually bound to the IT system, and local practitioners. They differ in the level of analysis: OGD contains a system perspective and focuses on actors and their interactions, whereas TD focuses on a local phenomenon related to a specific IT system. OGD has been understood as ecosystems (e.g., Dawes, Vidiasova, & Parkhimovich, 2016; Pollock, 2011), lifecycles (e.g., Attard, Orlandi, Scerri, & Auer, 2015; Folmer, Reuvers, Quak, van den Broek, & van Veenstra, 2011), and chains (e.g., Albano, 2013; Carrara, Chan, et al., 2015), while TD has been understood as loan (Kruchten, Nord, & Ozkaya, 2012), investments (Guo & Seaman, 2011), and shortcuts (Cunningham, 1992). The communities also apply a mix of metaphors. For example, in OGD, Davies (2011) applies ecosystem and infrastructure, Lindman, Kinnari, and Rossi (2015) apply ecosystem, network, and chain, and Harrison, Pardo, and Cook (2012) apply ecosystem, platform, network, and market metaphors. On the other hand, in TD, Kruchten et al. (2012) apply loan, evolution, landscape, and development, Cunningham (1992) apply shortcut, and Guo and Seaman (2011) apply investment metaphors.

3.2.2 Data collection in the communities

Our data collection was divided into researchers’ and practitioners’ communities. Figures 7 and 8 present a summary of the number of empirical documents over time collected in our study.

Research Communities. We created a list of outlets (journals and conferences) covering OGD or TD that contain English peer-reviewed, international research between 2015 and 2020. We perceived a high usage of metaphors in these communities in this time period. This is based on our own study and interactions with the communities. We excluded outlets that lacked a search engine and the ability to limit the searches by year. From this list, we chose three OGD outlets and three TD outlets based on the number of articles on their respective topics with a mixed sample in mind. We searched through the outlets to filter articles in two rounds of selection. First, we selected articles published between 2015 to 2020, written in English, and about the topics of OGD and TD. Second, we included accessible, full research articles, and excluded literature reviews, technical research, methodological research, and non-machine readable articles. In total, we identified 50 OGD articles and 50 TD articles for TD. Table 1 presents the articles divided over the outlets of their respective community with the two selection rounds. We analysed these articles in phase 3 of our study.

Table 1
The number of articles in the selected outlets from the OGD and TD communities

Outlet (Community)	Round 1	Round 2
Government Information Quarterly (OGD)	79	27
Information Polity (OGD)	36	10
International Conference on Electronic Government (OGD)	34	13
Journal of systems and software (TD)	20	15
International Conference on Technical Debt (TD)	40	31
International Conference on Software Engineering (TD)	6	4

Practical Communities. We explored the practical communities through desk research and identified data sources. We downloaded empirical material from the data sources, if it had sufficient quality and met our selection criteria (Table 2). We adapted our data collection to the differences in community characteristics of the OGD and TD communities. In the OGD community, we chose to study the European Data Portal (EUDP) and the Open Data Institute (ODI), as they provide rich streams of English public documents, act as international nexuses of OGD knowledge, and show some application of metaphors. We explored the actors’ website with a focus on knowledge-intensive documents, as we believed they could capture the application of metaphors in the OGD community. In total, we identified 34 documents for EUDP and 126 documents for ODI. In the TD community, we extended our exploration to include English and Danish due to limited search results. Danish had the highest number of hits amongst the Scandinavian languages. We chose 11 legitimate actors and identified a total of 103 documents. The TD community’s characteristics are local, semi-private and fragmented in nature, therefore to enter the community, we conducted 17 semi-structured interviews with respondents at a Danish Agency. The interviewees worked with the development and maintenance, either as business managers, law-graduate, IT architects, developers, or application managers. The interviews enabled us to capture how practitioners understand TD as a phenomenon and how metaphors are applied.

Table 2

Selection criteria for practical literature

$+/-$	Selection criterion	Reason
$+$	Published within 2015–2020	Capture the recent longitudinal development of the metaphors
$+$	Entailing ‘Open Government Data’ or ‘Technical Debt’	Only include articles focusing on relevant for the study
$+$	Public and knowledge-oriented documents	Increase the chance for metaphorical use and developments
$+$	Only legitimate and credible organisations as publishers	Authenticity and credibility that helps to ensure the quality of our research (Myers, 2013)
$+$	Possible to download the content	Accessibility to the empirical material to practitioners

3.3 Phase 3: Data analysis

Table 3
Our metaphor group dictionary

Open government data metaphor groups
Ecosystem	Cultivate, Element, Ecosystem, Evolve, Climate, Evolution, Cycle, Cultivation, Environment
Silo	Silo, Access, Close, Unlock, Lock, Release, Capture, Share
Creature	Lifecycle, Body, Heart, Interact, Consume, Adapt, Sit, Smart, Maturity, Capability, Mature, Life cycle
Market	Provider, Competitive, Consumer, Asset, Competition, Provision, Trade, Supply, Market, Demand, Supplier
Industry	Survey, Resource, Oil, Extract, Landscape, Plot
Adventure	Journey, Myth, Obstacle, Struggle, Launch, Hero, Champion
Technical debt metaphor groups
Economics	Portfolio, Forecast, Capital, forudsi, forudse, progn, kost, penge
Emotion	Dark, Shadow, Visible, Density, Bad, Negative, synliggøre, synlighed, tæt, dårlig, negativ
Financial	Financial, Spend, Save, Income, Finans, spare, forbrug, gemt, gemme, indtægt, bankruptcy
Investment	Invest, Return, afkast
Loan	Repay, Principal, Interest, rente
Person	Behavior, Immature, Lifecycle, Sprint, Smell, Brain, Grow, Retirement, Maneuver, Survival, Mature, Legacy, Inheritance, anatomy, heritance, infancy, candidate, Handsome, adfærd, cyklus, vokse, levetid, modern, hjern, arvet, bevæge
Purchase	Discount, Pay, Cost, omkost, betal
Repair	Shortcut, Hack, Self-fix, Maintain, Break, Fix, Refactor, genvej, vedlig, refak, brud, stykker, fiks

We first analysed the empirical material in MaxQDA using a word frequency analysis and then in-depth based on the CAOM framework (see Section 4). Both methods used metaphor density to inform us about possible shifts in the communities’ application of metaphors. Density was calculated per metaphor for each year by summing metaphor applications in the year and dividing this amount by the number of documents for that year.

Metaphor Frequency Analysis. The word frequency analysis (WFA) (see Izwaini, 2003) resulted in statistics about the frequency of certain words in the empirical material. It provided us with an estimation for metaphor usage, as the method does not consider the word’s context. This analysis followed three steps. First, we identified search terms for a dictionary. We understand categories as metaphor groups where the search terms are possible metaphorical words (Table 3). To remove unnecessary words, we applied WFA with filters of word frequency above 19, length above two, and the standard stop list of MaxQDA. We sorted out everyday, non-metaphorical, fragmented words into a stop list or our dictionary and included metaphorical words into a dictionary. We discussed uncertain words and decided their status. For OGD, we started with 373.297 possible words and ended up with 190 metaphorical words. For TD, we began with 393.103 possible words and ended up with 76 metaphorical words. Second, we discussed and rearranged the metaphorical words into metaphor groups, and we decided to focus on six OGD groups and six TD groups. Third, we applied the metaphorical groups to the empirical material for their respective community using MaxQDA’s category matrix browser. We examined the matrix for patterns. If we identified anything unexpected or unusual, we browsed the metaphor group and its search terms in the respective documents. If a word was used more in a non-metaphorical manner than in a metaphorical manner, we divided it into more specific metaphorical words. For example, the search term climate counts climate change, as such, we divided it into the metaphorical words political climate, business climate, and organizational climate. On the other hand, some words could not be specified because of complexities in context, these words were deactivated in our dictionary. For example, in TD, we deactivated negative as it is primarily used to describe results rather than as a metaphor for an ‘Emotion’. We visualised the final statistics as charts in Excel to reveal patterns. The charts were based on the metaphor’s usage density. We discussed the charts to draw tentative conclusions about the communities’ application of metaphors. The charts presented the statistics by year divided over the practical and research parts of the communities. Then, we applied the CAOM framework and examined the charts for metaphor usage and expression. We analysed usage by studying changes in metaphor density for and between years (short-time changes, revealing the speed of replication), while we analysed expression by examining the usage over time (illustrated by six patterns Section 5, revealing stability over time). The assemblages of metaphors could only be explored in the qualitative analysis.

Qualitative Analysis. We discussed the patterns to decide on one metaphor per community to explore deeper. We decided to explore the application of ‘Ecosystem’ (for OGD) and ‘Technical debt’ (for TD) based on their differences in application and metaphor density. We created an analytical template based on our emergent analytical framework (see Section 4) that we filled out with the empirical material for the selected metaphors. We applied the framework for each year and examined the density and application of the metaphor. We used the emergent analytical framework to examine how it has changed over time within a community. We paid attention to deviant cases or negative cases to ensure the quality of our findings and framework. We examined the individual items or cases which appear to deviate from the developed framework (Mays & Pope, 2000).

4. The emergent CAOM-framework

The Communities’ Application Of Metaphors (CAOM) Framework is a final product and contribution of this article (see Walsham, 1995). It can be iteratively used to analyse the application of metaphors within a community’s communication. It assumes the application of metaphors happens in the community’s conceptual development to understand socio-technical phenomena. The framework follows a quantitative and a qualitative approach.

The quantitative approach provides high-level insights into how the application of metaphors in a community develops over time. The framework conceptualises this communication as a stream of messages containing metaphors. The continued application of metaphors forms flows within this stream, such as an ecosystem or technical debt flow. This conceptualisation enables us to analyse a flow for increases and decreases in the application of metaphors, which provides us with a quantitative estimation of a metaphor’s longevity (stability over time) and fecundity (speed of replication) (Dawkins, 2016).

A metaphor’s copying-fidelity, referring to the senders’ ability to accurately replicate the metaphor (Dawkins, 2016), requires a study of changes and blends in the application of this metaphor. The qualitative approach employs us to analyse the community’s application of a metaphor in detail. Specifically, we analyse the application of a metaphor through its usage, expression, and assemblage.

Usage. We examine how a sender uses a metaphor in its communication. Whether it is merely mentioned, discussed and defined, or used as a tool. A sender or a receiver can use a metaphor for several purposes, such as (1) to enhance its understanding and learning of a target phenomenon, (2) to guide to possible directions for refinement and specification of thoughts, (3) to describe shapes, (4) to explain the unfamiliar or unknown, (5) to communicate abstract concepts, (6) to invite to conversations, and (7) to provide constraints to its thinking (Glucksberg, 1989; Haack, 1988; Hamilton, 2000; Ortony, 1975).

Expression. A metaphor’s expression entails examining where the metaphor appears in a sender’s communication and to what extent. An appearance in the introduction with no further explanation indicates the sender perceives the metaphor to have a high appeal and be immediately understood by receivers (Glucksberg, 1989; Lakoff & Johnson, 2008b). On the other hand, the metaphor can also appear several times in the text if the senders use the metaphor for one of the various purposes, such as (1) an analytical lens in qualitative research (Kaarst-Brown & Robey, 1999), (2) a help in methodology (Kendall & Kendall, 1993), and (3) a framework to conceptualise a targeted phenomenon in a particular way (Oates & Fitzgerald, 2007).

Assemblage. A sender’s communication contains an assemblage of metaphors and concepts (see Section 2). A metaphor can highlight some of the similarities between an initial phenomenon and a target phenomenon (Glucksberg, 1989; Lakoff & Johnson, 2008a; Ogden & Richards, 1923). A sender can, in their communication, apply and combine several metaphors (e.g., Izwaini, 2003). As such, we understand that it is possible for metaphors to be combined or supplement each other, highlighting different aspects of the target phenomenon where metaphors can have different roles. An assemblage could, therefore, reveal what similarities a sender is attempting to highlight and dissimilarities it is trying to abate.

Table 4
Overview of patterns in the metaphor frequency analysis for the OGD and TD communities

	OGD community		TD community
Pattern	Researcher	Practitioner	Researcher	Practitioner
Bumper	Creature	Adventure, Silo		Person, Purchase, Repair
Climber	Silo, Adventure, Industry		Economics, Emotions	Emotion
Duner		Creature
Quaker		Market, Industry	Purchase
Rocket	Ecosystem, Market		Loan, Person, Repair
Waver		Ecosystem		Economics, Loan

5. Patterns in the application of metaphors

In this section, we explore the application of metaphors through the density of estimated metaphor usage and identify patterns through the period 2015–2020 for the OGD and TD communities.

Six metaphor groups were analysed for the OGD community: ‘Ecosystem’, ‘Silo’, ‘Creature’, ‘Market’, ‘Industry’, and ‘Adventure’ (Table 4). No clear common pattern is visible. The metaphor density of the OGD researchers (see Fig. 4) is first stable with few fluctuations, ending with a sharp increase in density between 2019 and 2020. On the other hand, for OGD practitioners (see Fig. 3), the metaphor density fluctuates following a decreasing trend.

Figure 3.

Practitioners use of metaphors within the OGD community.

Figure 4.

Researchers use of metaphors within the OGD community.

The TD community’s six metaphor groups were: ‘Economics’, ‘Emotion’, ‘Loan’, ‘Person’, ‘Purchase’, and ‘Repair’ (Table 4). Similar to the OGD community, no clear common pattern is visible. For TD researchers (see Fig. 6), the metaphor density first fluctuates, followed by a sharp increase of all metaphor groups between 2019 and 2020. The metaphor density of the TD practitioners (see Fig. 5) is first stable, then expresses a sharp increase followed by a lower metaphor density for all metaphor groups.

Figure 5.

Practitioners use of metaphors within the TD community.

Figure 6.

Researchers use of metaphors within the TD community.

In the metaphor frequency analysis, six common patterns were identified for the OGD and TD communities. First, the waver pattern starts with a slow increase in metaphor density, reaching a plateau or peak, followed by a slow decrease in density. Second, the bumper pattern has first a higher stable period of metaphor density, which is abrupted by a bump (an increase, peak, and a decrease in density), ending with a lower stable density period. Third, the quaker pattern starts with a valley (decrease in density, bottom, and increase density). A solid increase follows the valley in density, peak, and a substantial decrease in density. Another valley follows this peak and then a smaller increase in density, peak, and decrease in density. Fourth, the climber pattern is a slow-moving plateau that has either an increasing or decreasing density over several years. Fifth, the duner pattern starts with a quick increase in density, reaching a peak, followed by a slow stable decrease in density, ending with a rapid decrease in density. Sixth, the rocket pattern starts with an increase in density, reaching a peak followed by a valley. It then expresses a sharp, high increase in density.

Figure 7.

Documents per actor over year for the OGD community.

Figure 8.

Documents per actor over year for the TD community.

Table 4 gives an overview of the identified patterns for the OGD and TD communities. The researchers’ communities have a higher metaphor density than the practitioners’ communities. This difference is possibly a consequence of researchers’ focus on concept development, while practitioners’ focus could be on using the metaphor. In addition, only the researchers’ communities contain the rocket pattern. Likewise, the waver pattern is only apparent in the practitioners’ communities. Whereas the Bumper pattern is not visible in the TD researchers’ community, but it appears in the remaining communities. Thus, there are similarities in patterns within the researchers’ communities and practitioners’ communities but not within the OGD community or the TD community. This result indicates the community’s application of metaphors is more aligned between practitioners and between researchers than between the community topic. Furthermore, in a cross-comparison, the TD researchers and OGD practitioners fluctuate in density. They are likely experimenting with metaphors or at least discussing them. A possible explanation is that a useful assemblage of metaphors has not been generated. On the other hand, the OGD researchers and TD practitioners seem stable. They have likely found a working assemblage of metaphors or do not need metaphors to describe the target phenomenon. This similarity could mean that different communities can express similar behaviours using metaphors for conceptual development. The two communities lack clear common patterns, which requires an in-depth analysis.

6. The application of metaphors from a COAM framework lens

In this section, we apply the CAOM-Framework (see Section 4) on the empirical material.

Usage. Senders’ usage of metaphors ranges from everyday, syntax, tool, and development. Everyday usage assumes the metaphor to be accepted amongst or known by the receivers. For example, Susha, Grönlund, and Janssen (2015) use the ‘Ecosystem’ metaphor to describe a research stream with a particular position. Syntax usage frames thinking, helping senders and receivers to perceive or approach a socio-technical phenomenon. For example, Charalampidou, Ampatzoglou, Apostolos, and Avgeriou (2017) use ‘Technical debt’ to create context and relevance for studying code smell; “[ $\ldots$ ] we explore code smells, which according to practitioners are the most commonly occurring type of TD in industry, by assessing the associated interest probability.” (Charalampidou et al., 2017, p. 1). Code smells are the symptoms of “code decay and potential maintenance problems that can be avoided by applying the right refactoring.” (Fontana, Ferme, Zanoni, & Roveda, 2015, p. 16) Tool usage help senders and receivers to see new ways to act or work with a socio-technical phenomenon. For example, ODI (2019) presents a tool to map an ‘Ecosystem’. The tool follows four steps: (1) map actors, (2) map the formal value exchanges, (3) map the ‘soft’ value exchanges, and (4) find opportunities. Develop usage can combine tool and syntax usage but focuses mainly on refining the metaphor for some purpose. For example, Siebra, Oliveira, Seaman, Silva, and Santos (2016) use grounded theory to develop ‘Technical debt’ into a theory;

“Thus, this work contributes to the effort in building a formal theory about TD and provides directions to assist the work of developers/managers who intend to identify and monitor TD items in their projects, given the practical nature of this study.” (Siebra et al., 2016, p. 219).

Expression. Senders can express metaphor differently in their communication: at the start, in the end, as crumbles, or as a mix. At the start, a metaphor helps to position the communication within a larger discourse or supply initial reasoning about a target phenomenon. For example, Carrara, Chan, et al. (2015) express ‘Ecosystem’ to frame the context of value creation. In the end, a metaphor helps to wrap up the communication or encourage the application of the metaphor. For example, while Altayar (2018) studies motivations for open data adoption from an institutional theory perspective, the author ends with the following recommendation: “Regarding the specific context of the study, it is important for policy makers in Saudi to take into consideration the development of OGD using an ecosystem approach [ $\ldots$ ]” (Altayar, 2018, p. 642). As crumbles, the communication entails sporadically mentions of the metaphor. For example, Carrara, Fischer, Oudkerk, van Steenbergen, and Tinholt (2015) mention ‘Ecosystem’ in the middle of the report. A mix is a combination of the the above expressions (e.g., Bonina & Eaton, 2020; Dawes et al., 2016; McBride, Olesk, Kütt, & Shysh, 2020).

Assemblage. A sender can apply and combine several metaphors and concepts in their communication, forming an assemblage. The focus is on the role of a studied metaphor within its assemblage. A metaphor can have a core role, providing a specific perspective on the targeted phenomenon. In TD, several sub-categories of the metaphor emerge, such as architectural debt and code debt (e.g., Fontana et al., 2015; Verdecchia, Malavolta, & Lago, 2018). The role of a metaphor can be supplementary, helping to create a new metaphor or to enhance the meaning of concepts, phenomena, or other metaphors. For example, Khayyat and Bannister (2017) focus on co-creation, but draws on ‘Ecosystem’ to put the activity within a larger whole; “Co-creation takes many forms. The type of co-creation that emerged in this research can best be described as a type of ecosystem in which various members or stakeholders interact and interdepend in various ways.” (Khayyat & Bannister, 2017, p. 14). A metaphor can have a peripheral role where it is briefly related to other metaphors, phenomenons, or concepts. On the other hand, Fontana et al. (2015) apply TD to set the scene for code smell’s relevance as a research topic; “Code smells can be used to capture symptoms of code decay and potential maintenance problems that can be avoided by applying the right refactoring. They can be seen as a source of technical debt.” (Fontana et al., 2015, p. 16). Lastly, the metaphor can have an insignificant role where it is merely expressed once or twice in a reference list, footnote, related posts, or without metaphorical meaning in the text.

Table 5
Overview of the in-depth qualitative analysis of the ecosystem and technical debt metaphors

Table 5 gives an overview of the in-depth analysis of the application of the ‘Ecosystem’ and ‘Technical Debt’ metaphors, drawing on the above dimensions. The in-depth analysis highlights that ‘Ecosystem’ is often used in an everyday manner where researchers develop the metaphor while practitioners used it as a tool. Its role is often insignificant or peripheral, with some instances of taking a core role. In comparison, ‘Technical Debt’ tends to be developed by researchers, and it has everyday usage by practitioners. Its role is often core within research, while it varies for practitioners.

7. Discussion

This research has explored the conceptual development of the OGD and TD communities by studying researchers’ and practitioners’ application of metaphors about socio-technical phenomena in their communities. While previous research tends to focus on a metaphor and its related target phenomenon (e.g., Glucksberg, 1989; Hamilton, 2000; Ogden & Richards, 1923; Ortony, 1975; Thibodeau & Boroditsky, 2011), our study broaden this research by using meme theory to perceive metaphors as part of a community’s conceptual development for a socio-technical phenomenon (see Section 4). The CAOM framework combines previous research about metaphors with a meme perspective, resulting in an original approach to understanding and studying metaphors. This section discusses the application of metaphors within the OGD and TD communities, our used research approach for studying metaphors, and implications for practice and research.

7.1 The application of metaphors

In this section, we focus on the identified patterns and assemblage of metaphors in the communities (see Sections 5 and 6). Our findings show that researchers discuss, share, and apply metaphors for conceptual developments, whereas practitioners primarily apply metaphors. Researchers display a higher metaphor density than practitioners and cite previous research. It is, as such, easier to trace the development and changes of metaphors in the research part of the communities. This difference is most likely a reflection of the researchers developing knowledge about less known socio-technical phenomena while practitioners are encountering or creating these phenomena (Matthiasen & Nielsen, 2008; Van De Ven, 2007).

Communities’ Application of Metaphors. The first research question asked how and why metaphors had been applied in the OGD and TD communities’ conceptual developments. The Sections 5 and 6 provides an answer for the how, but the findings need further highlighting to give an idea about why the communities apply metaphors in a certain manner.

The communities apply several metaphors each playing a specific role within an assemblage. OGD is a name for the core socio-technical phenomenon (the data that are published and reused following certain principles (Attard et al., 2015) but needs metaphors to be further developed and give context. For example, ‘Ecosystem’, ‘Market’, and ‘Industry’ give terms to describe the larger systems of actors and interactions surrounding OGD. On the other hand, ‘Technical Debt’ is both a developing metaphor, but also the everyday name for a specific socio-technical phenomenon. TD community develops more the metaphor compared to OGD because TD is the core of the TD field. While the socio-technical phenomenon influences what metaphors senders decide to apply in their conceptual development (see Section 2), we see no clear relationship between a socio-technical phenomenon and how senders apply metaphors in their conceptual development. An outliner is that OGD researchers never applied the Ecosystem metaphor as a tool within the explored time period, but we know from previous OGD research that it has been used as a tool (e.g., Harrison et al., 2012). Rather, we propose that the significance of a metaphor’s role in its assemblage influences how senders apply metaphors. The findings show that a peripheral or insignificant role tends to come with high everyday usage. While it is consistent for 19 of 24 years, there are some negative cases: 2016 for OGD practitioners, 2015 and 2020 for TD practitioners, and 2017 and 2020 for TD researchers. These negative cases cannot be explained by the used research approach, as such opening new research avenues.

Additionally, the OGD and TD communities vary further in community type. OGD is an open public community with more complex interactions, making it challenging to separate concepts and metaphors. We expected that this openness would have contributed to a higher degree of development for the Ecosystem metaphor, as ideas and knowledge can easily flow in and out of the community. However, this was not the case. Instead, they had high usage of various metaphors. On the other hand, TD is a semi-private community, with a strong core of researchers and software developers who are developing the metaphor. They appear to take ownership of the socio-technical phenomenon and try to solve the challenges it poses. The phenomenon is expanding to other communities (e.g., engineering) and other job functions (e.g., management). A surprising insight is that an open community applies more peripheral metaphors through an everyday usage, while closed communities developed more core metaphors.

As an answer to the ‘why’-question, we propose that while the socio-technical phenomenon influences what metaphors senders apply in their conceptual development, the metaphors’ role within an assemblage and community type influence how senders apply them in their conceptual development. Thus, we expand on Lakoff and Johnson (2008a) arguments that metaphors structure our thinking and are influenced by how we perceive their origin.

Patterns in Metaphor Frequency. In the period between 2015 and 2020, we identified six patterns in the application of metaphors for the OGD and TD communities (see Section 5). While we identified no clear common patterns between the communities, we found that there are similarities in the patterns within the researchers’ communities and within the practitioners’ communities but not within the OGD communities nor the TD communities. This difference indicates that the community’s application of metaphors is more aligned between practitioners and between researchers than based on a common socio-technical phenomenon. At the same time, only the researcher communities contain the Rocket pattern; likewise, the Waver pattern is only apparent in the practice communities. Researchers could be more sensitive to the appeal of a metaphor, seeking to experiment and test it, while practitioners take longer to accept a metaphor because they need to have practical utility. Whereas the Bumper pattern is not visible in the TD researcher community, it appears in the remaining communities. The other communities might, thus, be more open to experimentation with metaphors. In a cross-comparison, the TD research community and OGD practice community display fluctuation in density. They are likely experimenting with metaphors or at least discussing them. It is possible that a useful assemblage of metaphors has not been generated. On the other hand, the findings show that the OGD research community and TD practice community are more stable in their metaphor density. They have likely found a useful assemblage of metaphors or do not need metaphors to describe the target phenomenon. This similarity could mean that different communities can express similar behaviours towards applying metaphors for conceptual development. This conclusion is in contradiction to Stubblefield (1998)’s statement that metaphors can have lifecycles, as our findings show that metaphors are part of ever-evolving processes following similar patterns.

Assemblage of Metaphors. Our findings show a tendency that assemblages of metaphors can be constructed to create a new logical syntax. For example, ‘Technical Debt’ is combined with ‘Repair’ rather than ‘Repay’. The metaphors fit with the target phenomena but could be understood as incompatible if compared to each other. However, Mészáros (1966) warns that the metaphorical fallacy is to draw a conclusion from something that is only metaphorically established. It is, as such, possible that senders and receivers create similarities and reasoning, which are not (fully) applicable to the target phenomenon. For example, in the OGD community, the ecosystem is combined with business and health. Where businesses and ecosystems behave differently (Mars et al., 2012) and health is a metaphor that originates in medicine for an organism and is now stretched to groups (Rapport, Gaudet, Constanza, Epstein, & Levins, 1998). As a result, the senders’ and receivers’ construction of assemblages of metaphors is similar to using their minds as a lab to discover new ways to express understanding and thinking about a target phenomenon by drawing on previous knowledge. This insight means that metaphors are more than a tool to economize a limited vocabulary, structure thinking, create new meanings, and generalize (Black, 1955; Lakoff & Johnson, 2008b; Ogden & Richards, 1923). Metaphors can be combined to go beyond traditional thinking, embracing new ways of reasoning and perceiving about socio-technical phenomena. In addition, our research indicates that senders can employ assemblages to help receivers sort similarities and differences. These assemblages can oscillate over time with some recurrent configurations (e.g., in the OGD community, ‘Ecosystem’ tends to be paired with ‘Business’ or ‘Production’). At the same time, Mészáros (1966) explains that a metaphor is more than a word, as it can help receivers to better understand the message. A metaphor can provide a logical syntax for thinking. The metaphor-assemblage approach opens questions about inter-metaphorical emergence, meaning senders and receivers could create new meanings through combinations of metaphors that could not be expressed by a single metaphor (as shown in the ‘Technical Debt’ example above). It is possible that the oscillations of the assemblages are attempts by the senders to capture target phenomena difficult to communicate. Also, senders and receivers can apply a metaphor as a tool to reason and make sense of new or complex phenomena (e.g., Hekkala et al., 2018; Stubblefield, 1998; Thibodeau & Boroditsky, 2011). Therefore, an assemblage of metaphors could help senders create new ways of reasoning and understanding (a new logical syntax) about a target phenomenon, which a single metaphor could not express even if the metaphors when literally compared would be incompatible. Thus, we expand on Stubblefield (1998)’s findings that metaphors are valuable as designing tools, as metaphor assemblages can be valuable as a tool to understand a new logical syntax, as Izwaini (2003) display in the various metaphors applied in relation to computers and the internet.

7.2 Studying metaphors

To our knowledge, other researchers have not studied the application of metaphors in relation to conceptual development using an approach similar to ours. Thus, our findings provide guidance to future research of interesting topics to explore further using similar methods to this article.

The study of metaphors in a community is complicated as several senders can participate in the communication through various outlets. Some senders explain their metaphors, while other senders assume the receivers will spring to the right spontaneous conclusions (Ogden & Richards, 1923). In addition, the application of metaphors can differ in usage, expression, and assemblage where they are sometimes not applied as metaphors or at least border between being a metaphor and being everyday language or an unfettered concept. Therefore, it can be difficult to determine the exact meaning of a metaphor. We can draw some lessons learned about researching metaphors related to the OGD and TD community characteristics, the emergent CAOM framework, the patterns in the application of metaphors, and understanding the application of metaphors from a COAM framework lens.

Community Characteristics. The community characteristics influence how senders communicate to receivers but also act as an entry barrier for newcomers to the community. The OGD community is open, public, and international, allowing easy access for receivers (and researchers). The challenge was to find relevant key outlets where the majority of essential communication happens (relevant empirical material). The format and quality of this communication varied, which made the research time-consuming, giving less gain per resource invested (as seen as an increase in the ‘Ecosystem’s role as insignificant in assemblages. See Table 5). Moreover, the TD community is semi-private and local, which can create an entry barrier for newcomers but also increase the quality of the material. It requires more effort from the researcher to access or become part of the community but also provides greater gains. The entrance barriers include cultural and language barriers (e.g., the researchers are very strict on what can be defined as technical debt).

The Emergent CAOM Framework. We synthesised the CAOM framework from previous research about metaphors and memes. The metaphor research gave a foundation for the framework, while the meme research provided an approach to studying and understanding the spread and application of metaphors within a community. This combination is an original contribution to previous research. The framework focused our attention on important parts of the empirical material while giving structure to the analysis. It enabled us to bridge the use of a quantitative and a qualitative method.

Patterns in the Application of Metaphors. The quantitative analysis used word frequency analysis to estimate the application of metaphors within the empirical material over time. While it required several iterations to become satisfactory, it was useful as a tool to gauge the senders’ application of metaphors, as it helped to identify patterns. However, this method does not consider the context of words, giving an estimation where the researchers need to decide to exclude or include words in the grey zone (words used in both a metaphorical and non-metaphorical manner). The analysis method is fast to apply but time-consuming to set up and difficult to get right. At the same time, the lack of consideration for context meant that assemblages of metaphors were only analysable in the in-depth analysis. This exclusion allowed the frequency analysis to focus on the usage and expression of metaphors over time. It revealed when a metaphor had a high potential to influence the conceptual development of a community but also when it could have limited potential.

The Application of Metaphors from a COAM Framework Lens. The in-depth analysis focused on a metaphor per community using the CAOM framework. It allowed a richer understanding of a metaphor’s usages, expressions, and assemblages (application). We saw a correlation between the density, usage, expression, and assemblage of metaphors. For example, the TD researcher community has a high density, tending to primarily apply the metaphor for development purposes, the expression is mixed, and the metaphor has a core role. Although the in-depth analysis was time-consuming, it helped to explain the usage and trends of metaphors within the two communities. It was an essential part to understand assemblages of metaphors.

7.3 Implications

Our research has some implications for practitioners’ and researchers’ application of metaphors in conceptual developments. They can actively experiment with assemblages of metaphors to create new logical syntax and meanings, considering the potential impacts metaphors could have on each other. Inter-metaphorical emergence can be beneficial but could lead to metaphorical fallacies or nonsensical communication. They should be aware that assemblages of metaphors can cause unwanted associations for receivers, as they are the catalyst who interpret the metaphor and its assemblage. As such, it is important for senders to be clear when communicating original assemblages of metaphors, highlighting important similarities and dissimilarities for the receivers. This approach could help to increase the benefits gained from conceptual developments using metaphors.

Research. Researchers can use the CAOM framework as an analytical lens. It captures the underlying mechanisms of metaphor usage, expression, and assemblage (application). The framework is developed to be used in mixed methods and qualitative research; as such, it is ample for use in case study research. Researchers could use the CAOM framework as part of their analysis, as an introductory guide to metaphors, and as a starting point for future research that seeks to theorize metaphor usage in communities (see Eisenhardt, 1989; Walsham, 1995). On the other hand, our findings imply that researchers should be aware of metaphors and their assemblages. We believe assemblages can give insights into what is perceived as new by practitioners and researchers, which could direct future research to original contributions. Similarly to practitioners, researchers could discuss the emergence of new meanings when combining various metaphors to increase the speed of conceptual developments. We recommend that digital government researchers include an action-oriented element when they develop or apply metaphors, as it can improve practitioners’ usage of their research.

Practice. Practitioners have a repertoire of actions and interpretations for the application of metaphors. They should reflect on what a metaphor entails for their thinking and actions, but also its possibilities. Senders could use the CAOM framework to analyze metaphors, which could help them avoid pitfalls in their communication and create assemblages that foreground what they seek to communicate. Receivers can use the CAOM framework to gain a deeper insight into a metaphor. Especially, if the sender is attempting to communicate something only expressed through the metaphorical emergence of an assemblage of metaphors. We recommend that practitioners approach metaphors as a serious tool that they can apply to quickly expand previous knowledge to encompass new and unknown socio-technical phenomena. This process can involve evaluations and reports on the success of a metaphor and its assemblage, but also a focus on the development of metaphors. We recommend that governments seeking to communicate with practitioners through policies or documents should draw on action-oriented metaphors, which can guide the practitioners towards an intended goal.

8. Conclusion

Our study aimed to explore conceptual development in communities by studying researchers’ and practitioners’ application of metaphors about socio-technical phenomena in their communications. We employed a mixed-method approach, analysing the application of metaphors between 2015 and 2020, focusing on the TD and OGD communities. We use meme theory and synthesise metaphor literature into the emergent CAOM analytical framework with the key concepts of metaphors’ usage, expression, and assemblages of metaphors (applications), which is a contribution of our study (see Section 4). Our study was guided by two research questions, which it answered:

First, how have the OGD and TD communities applied metaphors in their conceptual development between 2015 and 2020, and why in this manner?

•
The OGD community applies metaphors to supplement the main concept, while the TD community applies TD as a syntax to develop the main metaphor further. Practitioners combine metaphors and concepts, focusing on action, while researchers expand upon metaphors and concepts with other metaphors and concepts, focusing on knowledge development. Generally, researchers and practitioners blend metaphors to create new ways of reasoning and understanding. Some combined metaphors could be understood as incompatible if literally compared to each other. Our findings also show that metaphors are part of ever-evolving processes following similar patterns even between communities and socio-technical phenomena.
•
We propose that while a socio-technical phenomenon influences what metaphors senders apply in their conceptual development, how the metaphors are applied is influenced by community type (e.g., researcher or practitioner; open or semi-private) and metaphors’ role (e.g., core or peripheral).

Second, what are the lessons learned for analysing the application of metaphors for the OGD and TD communities?

•
The boundary between metaphor and concept is blurry where some terms could be both, making them challenging to capture in both quantitative and qualitative analysis.
•
The community characteristics influence how senders communicate to receivers but also act as an entry barrier for newcomers to the community.
•
The emergent CAOM framework provides analytical dimensions, enabling the exploration of metaphors’ application in communal communication. It helped us bridge the use of quantitative and qualitative analysis.
•
The quantitative analysis revealed when a metaphor had a high potential to influence the conceptual development of a community but also when it could have limited potential.
•
The qualitative analysis explained the application and trends of metaphors within the two communities. It was an essential part of understanding assemblages of metaphors.

8.1 Limitations and future research

Our research has some limitations and opens new avenues for future research. In our research, we followed the suggestions of Mays and Pope (2000) to improve our research’s validity. We have used methodological triangulation, fair dealings, attention to negative cases, and lastly, a clear exposition of methods of data collection and analysis. We also followed Creswell and Creswell (2017); Myers (2013); Saunders and Lewis (2012) to further ensure research quality. Our synthesis of the emergent CAOM framework focused on understanding, as such it is possible that further previous research could be used to enhance the framework. We believe the framework gives sufficient understanding to analyse the application of metaphors. However, our research has mainly focused on public empirical sources (e.g., blogs and reports) with some interviews from the TD community. This approach excludes forums and events where practitioners and researchers could meet to discuss and develop metaphors. It is possible that these social gatherings are important locations for the application of metaphors. On the other hand, if so, their output should sooner or later be expressed in the texts we have analysed; allowing us to identify them indirectly. Moreover, we used word frequency analysis (WFA) and then a qualitative analysis (QA). WFA is limited in its ability to capture the context of used words. We revisited the instances with high frequency to ensure it was used metaphorically. QA should have offset some of these problems by allowing us to study the metaphors in their context, but this correction is not included in the statistics. We believe the combination of WFA and QA has mitigated some limitations of each other and contributed to a better validity of our research. In the end, we believe our research to have good quality, but is one step towards understanding the development of metaphors. It has opened for future research about (1) the application and combination of metaphors, (2) comparative, longitudinal studies for the application of metaphors in communities, and (3) metaphors as a method for knowledge development about socio-technical phenomena between practitioners and researchers.

Footnotes

Author biographies

Mille Edith Kjærsgaard Hansen has recently been awarded a Ph.D. for her research on technical debt management. She has published in the Government Information Quarterly journal and in the conferences: EGOV-CeDEM-EPART and ICEGOV. She has previously worked at the IT department at the Danish Business Authority and is currently working as an IT consultant.

Jonathan Crusoe is a Ph.D. in Information Systems at the University of Borås, Sweden, and Postdoctor at the University of Gothenburg. His research focuses on open government data as reform and ecosystem, which his recently defended doctoral thesis covers. His research is often multi-disciplinary with international collaborations. Recently, he published and presented three articles about open data at the EGOV-CeDEM-EPART conferance and has published in the journal of Transforming Government – People, Process and Policy.

Ulf Melin, Professor in Information Systems, Linköping University, Sweden. His research is focused on public sector digitalization. Ulf has published in e.g. Government Information Quarterly, Transforming Government – People, Process and Policy, International Journal of Electronic Governance, International Journal of Public Information Systems; and is active in several IS conferences.

References

Accenture Federal Services. (2018). Decouple to innovate how federal agencies can unlock it value & agility by remediating technical debt.

Albano

C. S.

(2013). Open government data: a value chain model proposal. In Proceedings of the 14th annual in ternational conference on digital government research, pp. 285-286.

Altayar

M. S.

(2018). Motivations for open data adoption: An institutional theory perspective. Government Information Quarterly, 35(4), 633-643.

Attard

Orlandi

Scerri

, & Auer

(2015). Asystematic review of open government data initiatives. Government Information Quarterly, 32(4), 399-418.

Black

(1955). Xii-metaphor. In Proceedings of the aristotelian society (Vol. 55, pp. 273-294).

Boell

S. K.

, & Cecez-Kecmanovic

(2014). A hermeneutic approach for conducting literature reviews and literaturesearches. CAIS, 34, 12-12.

Bonina

, & Eaton

(2020). Cultivating open government data platform ecosystems through governance:Lessons from buenos aires, mexico city and montevideo. Government Information Quarterly, 37(3), 101479-101479.

Carrara

Chan

Fischer

, & Steenbergen

E. V.

(2015). Creating value through open data: Study onthe impact of re-use of public data resources. EuropeanCommission.

Carrara

Fischer

Oudkerk

van Steenbergen

, & Tinholt

(2015). Analytical report 1: Digitaltransformation and open data.

10.

Charalampidou

Ampatzoglou

Chatzigeorgiou

, & Avgeriou

(2017). Assessing code smellinterest probability: a case study. In Proceedings ofthe XP2017 Scientific Workshops, 1-8.

11.

Creswell

J. W.

, & Creswell

J. D.

(2017). Research design: Qualitative, quantitative, and mixed methods approaches. Sage publications.

12.

Cunningham

. (1992). The wycash portfolio management system. ACM SIGPLAN OOPS Messenger, 4(2), 29-30.

13.

Custers

, & Bachlechner

(2017). Advancing the eudata economy: Conditions for realizing the full potentialof data reuse. Information Polity, 22(4), 4291-309.

14.

Davies

(2011). Open data: infrastructures and ecosystems. Open Data Research, 1-6.

15.

Dawes

S. S.

Vidiasova

, & Parkhimovich

(2016). Planning and designing open government data programs:An ecosystem approach. Government Information Quarterly, 33(1), 15-27.

16.

Dawkins

(2016). The selfish gene. Oxford university press.

17.

Eisenhardt

K. M.

(1989). Building theories from case study research. Academy of Management Review, 14(4), 532-550.

18.

Fairbanks

(2020). Urtechnical debt. IEEE Software, 37(4), 95-98.

19.

Folmer

Reuvers

Quak

van den Broek

, & van Veenstra

A. F.

(2011). Walking the extra byte:A lifecycle model for linked open data. In Folmer

Reuvers

, & Quak

(Eds.), Linked open data-pilotl inked open data nederland (pp. 95-111). Remwerk.

20.

Fontana

F. A.

Ferme

Zanoni

, & Roveda

(2015). Towards a prioritization of code debt: A code smell intensityindex. In 2015 IEEE 7th International Workshop on Managing Technical Debt (MTD), 16-24.

21.

Gerrits

(2021). Soul of a new machine: Self-learning algorithms in public administration. Information Polity, 26(3), 237-250.

22.

Glucksberg

(1989). Metaphors in conversation: How are they understood? why are they used? Metaphor andSymbol, 4(3), 125-143.

23.

Guo

, & Seaman

(2011). A portfolio approach to technical debt management. Proceedings of the 2ndWorkshop on Managing Technical Debt, 31-34.

24.

Haack

(1988). Surprising noises: Rorty and hesse on metaphor. In Proceedings of the aristotelian society(Vol. 88, pp. 293-302).

25.

Hamilton

(2000). Metaphor in theory and practice: the in uence of metaphors on expectations. ACM Journal of Computer Documentation (JCD), 24(4), 237-253.

26.

Handbook

O. D.

(2015). What is open data? Retrieved 2020-08-31, from http://opendatahandbook.org/guide/en/what-is-open-data/.

27.

Harrison

T. M.

Pardo

T. A.

, & Cook

(2012). Creating open government ecosystems: A research and developmentagenda. Future Internet, 4(4), 900-928. Retrievedfrom http://www.mdpi.com/1999-5903/4/4/900/doi:10.3390/fi4040900.

28.

Hekkala

Stein

M.-K.

, & Rossi

(2018). Metaphors in managerial and employee sensemaking in an informationsystems project. Information Systems Journal, 28(1), 142-174.

29.

Imenda. (2014). Is there a conceptual difference between theoretical and conceptual frameworks? Journal ofSocial Sciences, 38(2), 185-195.

30.

Izwaini

(2003). A corpus-based study of metaphor in information technology. In Proceedings of the workshopon corpus-based approaches to figurative language, corpus linguistics (pp. 1-8).

31.

Kaarst-Brown

M. L.

, & Robey

(1999). More on myth, magic and metaphor. Information Technology & People.

32.

Kendall

J. E.

, & Kendall

K. E.

(1993). Metaphors and methodologies: Living beyond the systems machine. MIS quarterly, 149-171.

33.

Khayyat

, & Bannister

(2017). Towards a model forfacilitating and enabling co-creation using open governmentdata. Information Polity, 22(4), 211-231.

34.

Kruchten

Nord

R. L.

, & Ozkaya

(2012). Technical debt: From metaphor to theory and practice. LeeeSoftware, 29(6), 18-21.

35.

Lakoff

, & Johnson

(2008a). Metaphors we live by. University of Chicago press.

36.

Lakoff

, & Johnson

(2008b). Metaphors we live by. University of Chicago press.

37.

Lindman

Kinnari

, & Rossi

(2015). Business roles in the emerging open-data ecosystem. IEEE Software,33(5), 54-59.

38.

LNyhlen

, & Gidlund

K. L.

(2021). In conversation withdigitalization: Myths, fiction or professional imagining? Information Polity, Pre-press(Pre-press), 1-11.

39.

Magnusson

Juiz

Gomez

, & Bermejo

(2018). Governing technology debt: beyond technical debt.Proceedings of the 2018 International Conference onTechnical Debt, 76-84.

40.

Mars

M. M.

Bronstein

J. L.

, & Lusch

R. F.

(2012). The value of a metaphor: Organizations and ecosystems. Organizational Dynamics, 41(4), 271-280.

41.

Matthiasen

, & Nielsen

(2008). Engaged scholarship in is research. Scandinavian Journal of InformationSystems, 20(2), 1-18.

42.

Maxwell

J. A.

(2012). Qualitative research design. SAGE. Publications Inc.

43.

Mays

, & Pope

(2000). Assessing quality in qualitative research. Bmj, 320(7226), 50-52.

44.

McBride

Olesk

, K\̈refauthor{missing}, , & Shysh

(2020). Systemic change, open data ecosystem performanceimprovements, and empirical insights from estonia:A country-level action research study. InformationPolity, 25(3), 377-402.

45.

Morgan

(1998). Images of organization: The executive edition. Thousand Oaks, CA.

46.

Myers

M. D.

(2013). Qualitative research in business and management. Sage Publications Limited.

47.

Mészáros

(1966). Metaphor and simile: Preliminary notes to a discussion of norms and values. In Proceedings ofthe aristotelian society (Vol. 67, pp. 127-144).

48.

Oates

B. J.

, & Fitzgerald

(2007). Multi-metaphormethod: organizational metaphors in information systemsdevelopment. Information Systems Journal, 17(4), 421-449.

49.

ODI. (2019). Data ecosystem mapping tool the odi. Retrieved 2020-11-09, from https://theodi.org/article/data-ecosystem-mapping-tool/#1561999285194-8defa99a-39a5.

50.

Ogden

C. K.

, & Richards

I. A.

(1923). The meaning of meaning: A study of the in uence of thought and ofthe science of symbolism. Harcourt, Brace & World, Inc.

51.

Ortony

(1975). Why metaphors are necessary and not just nice. Educational Theory, 25(1), 45-53.

52.

Pollock

(2011). Building the (open) data ecosystem. Open knowledge foundation Blog, 31.

53.

Rapport

Gaudet

Constanza

Epstein

, & Levins

(1998). Defining ecosystem health.RAPPORT, D.; COSTANZA, R.; EPSTEIN, PR;GAUDET, C, 18-33.

54.

Saunders

M. N.

, & Lewis

(2012). Doing research in business & management: An essential guide to planningyour project. Pearson.

55.

Schedler

Guenduez

A. A.

, & Frischknecht

(2019). How smart can government be? exploring barriers tothe adoption of smart government. Information Polity,24(1), 3-20.

56.

Siebra

Oliveira

Seaman

Silva

, & Santos

(2016). Theoretical conceptualization of td: A practicalperspective. Journal of Systems and Software, 120, 219-237.

57.

Smith

, & Sandberg

(2018). Barriers to innovating with open government data: Exploring experiences acrossservice phases and user types. Information Polity,32(3), 249-265.

58.

Stubblefield

W. A.

(1998). Patterns of change in design metaphor: a case study. In Proceedings of the sigchiconference on human factors in computing systems (pp. 73-80).

59.

Susha

, Gr\̈refauthor{missing}, k., & Janssen

(2015). Driving factors of service innovation using open government data:An exploratory study of entrepreneurs in two countries. Information Polity, 20(1), 19-34.

60.

Thibodeau

P. H.

, & Boroditsky

(2011). Metaphors we think with: The role of metaphor in reasoning. PloS One, 6(2), e16782.

61.

Tom

Aybuke

, & Richard

(2012). A consolidated understanding of technical debt. In European conference on information systems (ecis) (pp. 1-13).

62.

Van De Ven

(2007). Engaged scholarship: A guide for organizational and social research. Oxford UniversityPress on Demand.

63.

Verdecchia

Malavolta

, & Lago

(2018). Architectural technical debt identification: The research landscape.2018 IEEE/ACM International Conference onTechnical Debt (TechDebt), 11-20.

64.

Walsham

(1995). Interpretive case studies in is research: nature and method. European Journal of InformationSystems, 4(2), 74-81.

65.

Zuiderwijk

Janssen

Van De Kaa

, & Poulis

(2016). The wicked problem of commercial value creationin open data ecosystems: Policy guidelines forgovernments. Information Polity, 21(3), 223-236.