Definition Synthesis of Agility in Software Development: Comprehensive Review of Theory to Practice

Information system/technology agility has been highlighted as a critical pillar of modern businesses [1]. In particular, software development as a complex domain adopts increasingly more agile approaches and has nowadays reached wide adoption [2]. The basic idea of agile software development is to manage complexity with the need to be able to respond to change [3]. Nevertheless, it is interesting that while the notion of agility in the software development domain is not new, the term has no consistent, complete, precise, agreed, uniformed and standard definition yet [4, 5, 6, 7, 8, 9, 10, 11, 12].

As Turner [13] pointed out, there are nearly as many definitions of agility as there are agile practitioners. In that case, the understanding and definition of agility remarkably vary [14]. As a result, many practitioners are confused about agility [15]. This confusion involves practitioners [15, 16], theorists [16], and even the recognized gurus [17]. As a result, the definition of agility in software development has not been well understood; people do mean different things when they are talking about the concept [17]. Moreover, organizations implement agile approaches without a clear understanding of how to define the word of agility [18].

People's different agility perceptions make application and evaluation of it very challenging [17, 19]. The definitions can also be taken to promise what Agile Software Development has to offer. Its misleading definitions will create unrealistic results, which will ultimately damage Agile itself. Inaccurate definitions and the corresponding base that such definitions establish will negatively affect all other Agile phenomena (frameworks, organizations, teams, etc.) built on that base. As long as agility and its attributes cannot be defined, it is hard to understand how a software development method might contribute to a development team's ability to be agile [20]. Hence, it is not feasible to discuss the effectiveness and value of Agile methods [21]. For instance, whatever is constructed on an agility definition directly proportional to speed, synonymous with flexibility or constrained firmly by adaptation, naturally deviates organizations from accurate agility and its benefits. As a result, a more in-depth understanding of the agility concept is crucial.

Apart from the studies providing their agility definitions, as for now in the academic literature, study [17] (not systematically) reviewed the definitions of agility in the software development domain up until 2013 and listed a range of facets of agility. To clarify the meaning of agility, Conboy, and Fitzgerald [22] and Conboy [23], the extended version of the former one, carried out a (not systematic) literature review on the concept of agility across several disciplines, including management, manufacturing and business. They also distinguished intertwined concepts of flexibility and leanness along with their relationship with agility. Then, they form the iterative definition of agility by merging underlying aspects from flexibility and leanness. Apart from these two studies, we have not encountered any other study that reviews the definitions of agility in the field of software development. This situation implies that there is still no systematic and up-to-date (covering also the last decade) study in this area.

Motivated with this gap, this paper first aims to review the existing definitions of agility from the literature. Then, it discovers and synthesizes the underlying facets of the identified definitions. The paper uncovers the inadequacy of existing definitions and provides conceptual differentiation among the agility concept with other similar paradigms (such as flexibility, adaptability, leanness, robustness) to describe agility and proposes an overarching definition of it for software development. Additionally, the process of agility is proposed to realize the inner mechanism experienced during agility actions, then to reach and serve for a clearer and healthier definition of agility. Consequently, this study focuses on three main issues:

• •    Identifying current definitions of agility in software development domain from the literature

• •    Investigating at what extent they provide a proper definition for each individual source and in general

• •    If there is a lack of a proper definition at the final, proposing an overarching one based on the appropriate parts of the previous definitions and after clarifications of the relevant concepts with the support of the literature

2.    Background: Process of Agility 3.    Research Method

Our study stands out with its up-to-date coverage, benefiting from the non-IT definitions, using a method for analysis, providing inputs to this analysis and synthesis with a comprehensive literature review, and the comprehensive evaluation of the facets with the support of the literature. The paper’s strength is that it aims to solve a large problem in finding the overarching definition of agility. Since there are a lot of definitions already in existence, taking on the multitude of definitions is a key strength in attempting to fill the gap. Additionally, one other strength is its access to a large amount of updated research with definitions of agility. In terms of proposing a definition, most of the studies provide a “take it or leave it” definition [23]. Unlike them, the definition of agility in our study is developed step by step, through a deep investigation and evaluation of all relevant facets.

The paper is organized as follows. Section 2 outlines the background for the process of agility that will be used in the further stages of the paper, especially during the development of the new definition of agility. Section 3 depicts the proposed method applied in the study. Section 4 delivers the details of the literature review along with its results. Section 5 elaborates the concepts encountered during the literature review stage as the relevant ones to agility concepts to clarify their difference, similarity and correspondence. Section 6 provides discussions on the current definitions and proposes a new one based on the background and baseline in Section 2 and 5. Finally, Section 7 delivers conclusions and limitations.

As any process, the agility process starts with a trigger. This trigger may be a change or a need for creating an action desired to be produced for a possible change that will be happening. The trigger can come from outside or from inside the entity’s system. An example from inside is the emergence of the system's self-improvement need as a trigger coming from within the system. This change can be any kind of predicted or unpredictable, expected or unexpected change. Making a change is another step. This step happens first in reactive behaviors and after in proactive behaviors by anticipating. The issue is that the later the entity detects it, the more time, money, and defects will be incurred [22]. After or before a change is happening, sensing and anticipating it (what is happening, or what will be happening) is the next step. Sensing is about detecting and collecting data from the source that suggests the need for change [1]. After sensing and anticipating, diagnosing, filtering, and interpreting input data occurs [1]. The entity evaluates and decides on what response to prepare. This response can also be null (including not delivering any response). The last step is the execution of the prepared response. This stage is where flexibility also plays a role; an inflexible entity may find executing a given response harder than a flexible one does. However, a flexible organization may not be agile if it is unable to detect and decide quickly and appropriately [1] and unable to make quick sensing (detecting and anticipating)

[24, 25] or able to do so with a wrong sensing and responding that could be fatal for the significance of agility [25].

Agility is not a concept unique to software development [22]. Indeed, it first appeared in the mainstream business literature in 1991, when a group of researchers at the Iacocca Institute in Lehigh University introduced the term “agile manufacturing” [26]. However, as argued by Conboy and Fitzgerald [22], “the search for a definitive, all-encompassing concept of agility is not to be found simply through an examination of agility in other fields” that are having the same problems as those studying agile methods in software development. Instead, it is to be found through an examination of the underlying concepts of agility. Thus, this paper aims to develop a comprehensive conceptual definition of agility applicable to the software development domain by considering useful definitions in other fields, enabling a more satisfactory agility degree. To do so, we applied an exploratory, inductive research design to synthesize primary studies for the purpose of making contributions beyond those achieved in the original studies, as proposed by Hoon [27]. It involves the accumulation of previous studies’ evidence, and more specifically its extraction, analysis, and synthesis in empirically consolidating primary studies [27]. To come up with a customized method for that purpose, we were inspired by Hoon [27] which fits well into our study’s context. Accordingly, we applied these research process steps:

• •    Framing the Research Question/Objective: The research objective of this paper is to describe and discuss the definition of agility in detail; and then to propose a formal definition of it for the software development domain. With this purpose, the study aims to reach other related studies and related contents in the literature.

• •    Locating Relevant Research: A literature review on the concept of agility was conducted. This review includes research on agility in software development primarily and Information Technology/Information System (IT/IS) domain secondarily, and in other domains such as manufacturing, business, organizational agility lastly. When it provides a domain-independent agility definition to serve our purpose, we appreciate the multidisciplinary nature and evolution of the concept.

• •    Inclusion/Exclusion Criteria: The next step is to determine the appropriate inclusion of relevant studies by defining and applying the inclusion and exclusion criteria to ensure the synthesis's validity [28].

• •    Extracting Data: The definitions of each agility and agility-related terms were extracted regarding the