Keywords

1 Introduction

Open source software (hereinafter OSS) is a user-driven, collaborative innovation produced by self-organizing teams of contributors dynamically formed through online interactions [1]. With the emergent development in the last decades, the OSS has become an important phenomenon in both economics and culture. Increasing number of IT giants has realized the value of OSS and created their own platforms for hosting the OSS projects. For instances, Google established Google Code on 2005 for providing fundamental tools for OSS developers to share their projects to the public; Microsoft subsequently set up the CodePlex for allowing the engineers and computer scientists to share their ideas and OSS projects although Microsoft used to strongly disagree the OSS campaigns [2].

Various benefits can be found with the emergence of OSS. We think three key benefits are predominantly important. First, OSS software can dramatically reduce the cost of development for IT company. Such cost does not only include the development cost but the innovation cost as well. The outstanding software received the effort and contribution from the talented developments throughout the world. Second, the OSS can facilitate the organizations to implement the information technology systems into their business process, especially for those small medium enterprise (hereinafter SMEs) or public institutes. There is no doubt that the IT can make the business better. However, due to the cost of IT implementations, the SMEs or public institutes may not implement the IT system due to their limited budget. Introducing the OSS systems can mitigate such dilemma. Last but not least, the information and knowledge can be diffused in terms of participating in OSS projects, which is beneficial for those people who want to have a good command of programming language since they can learn via practice.

To this end, great studies have been made to understand the OSS predominantly, with a focus on viewing OSS project as an organization for technological innovation. These investigations include the investigation of motivation to participate in the OSS development [3–6], the role of social network in promoting the OSS performance [7–10, 27], the governance and management of an OSS project [11, 12], and the influence of policy and legalization on OSS [13–15]. In addition, various measurements have been employed to assess to the extent the OSS success, such as number of downloads, frequency of CVS commits, extent of code reuse etc. Thus, it is imperative to provide a comprehensive extent of literature reviews on OSS studies. In this article, we surveyed the most representative OSS studies and summarized them into three streams based on their research topics. In addition, the assessments of OSS success are also summarized with their pros and cons. The future agenda and potential research gap are given in the end of the article.

2 Prior OSS Studies

In this section, a comprehensive extent of literature reviews is conducted. Based on the reviews, three subsections are listed by the research topics of prior OSS studies. In particular, the first stream includes the literatures studying the individual motivations to contribute or participate in the OSS projects. The second stream summarized the literatures discussing how the network characteristics influenced the OSS performances. In the third stream, we reviewed how the legalization, especially the OSS licenses, influences the OSS performances.

2.1 Individual Motivations

Thousands of individuals participate in OSS projects for diverse purposes. Their contributed product will be released to the public for free usage. Why are they willing to contribute to such OSS projects since their no economic returns from their contributed projects? What are their motivations? After surveying prior literatures, we found two key motivations, which can be concluded as intrinsic motivations and extrinsic motivations.

Intrinsic Motivation. The intrinsic motivations have been studied since 1970s. It is driven by an interest or enjoyment in the task itself and exists within the individual instead of depending on external pressure or desire for incentive reward [16]. The intrinsic motivation was widely regarded as a key motives for individuals to participate in the OSS projects. Previous literatures found more than half developers indicated that the enjoyment in programming and the sense of satisfaction originated form the participation in OSS software constantly motivate them to sustainably contribute to OSS projects. Besides the sense of enjoyment and satisfaction, Lakhani and Wolf [17] found the several individuals could show their creativity and new ideas in terms of implementing them into the OSS projects, which conferred great sense of accomplishment for them. Such sense was believed as another key motivation for constantly contribution. Although the intrinsic motivations were found as the original motives for breeding the OSS campaign, however, Roberts et al. [5] thought the intrinsic motivations had several defects, such as short effectiveness and strong self-direction, which might challenge the sustainability of OSS projects in future. Thus, it is imperative to unveil the extrinsic motivations for OSS participants.

Extrinsic Motivation. Extrinsic motivation denotes performing an activity is built upon the desired outcome like momentary incentives, reputations, and profits etc., which is the opposite of intrinsic motivation [16]. In other words, with the disappearance of external incentives, then the extrinsic motivations will decrease or even disappear. Von Krogh and his collaborators [18] investigated a large OSS project and interviewed several participants and summarized that the participation was built upon the pursuit of communal resources, which includes reputations, control of technology, and learning opportunities. The first dimension of extrinsic motivation is the acquisition of reputation. The reputation can be obtained in terms of (1) actively participating in the OSS projects, (2) providing solutions to the existing bugs or problems, (3) providing innovative revision or modification to the current OSS. With the increase of reputation, the participants will be conferred with higher authority, which will eventually enable him or her to dominate the entire project, such as the recruitment of developers or the decision power in the project management. To this end, the reputation in the OSS community is widely regarded as an extreme important extrinsic motivation for encouraging OSS developers to constantly contribute to OSS project. The second acquired communal resource is the “control over technology”. Such extrinsic motivation is formalized from the demand of self-usage. It is found that the key reason to participating in OSS project is to customize such project for own usage [3]. Although the novel and creative ideas can be realized through OSS project, however, the overall quality of OSS project cannot reach the industrial standard due to the relaxed management system and software testing procedures. Thus, those experienced users may compile their revised OSS and incidentally update it. In terms of long-term investment, those users may have a good command of this technology, which would benefit them in the future. The third communal resource is the learning opportunities. The participants of OSS project do not only include those experienced developers but those users or fresh programmers. By joining in OSS project, these people were provided a very good opportunity to learn (1) programming by collaborating with talented programmers in the world, and (2) knowledge about project management in software engineering as well. Besides these three key extrinsic motivations, prior literatures also identified several other extrinsic factors motivating individuals to contribute to the OSS project such as sense of reciprocity and job opportunities. Sense of reciprocity is prevailing in the OSS user support community, in which those end-users proactively respond or answer other’s questions in order to obtain the helps from others when they need in the future. In addition, prior literatures [19, 20] found some IT companies might recruit some talented programmers from the OSS community, which encourage some participants to diligently work on their contributed OSS projects.

Besides the explanation from motivation theory, several studies found institutional management or leadership also affected individual motivations. For instances, by interviewing the developers from Debian project, O’Mahony and Ferraro [12] found the democratic management style outperformed the bureaucratic one thought the later was found to be efficient. In addition, Li and her collaborators [11] found individuals preferred to join those OSS project which were employed the transformational leadership style. In other words, such leadership style signals that everyone could be conferred as a project leader in the future.

2.2 Network Characteristics

In second stream, prior studies investigated how network structures affected the OSS performances [7–10, 27]. The development of OSS project cannot be done without the collective actions. In this regard, prior literatures employed the social network analysis to articulate such collective actions, i.e. collaborative behaviors, which include inter-project individual-individual collaborative networks [7], project-affiliated networks [8, 9, 27], and intra-project communication or collaborative networks [10] etc. Thus, the social capital theory was mostly adopted as a theoretical underpinning to explicate such social relationships.

2.2.1 Inter-project Network

Due to the open nature of the OSS projects, the participants can freely contribute to multiple projects. In terms of such shared participants or concurrently contributed projects, two types of affiliated networks can be initialized. The first one is call project-affiliated network, in which the projects with shared participants are interconnected. The second one is called participant-affiliated network, in which the developers/project administrators who contributed to same projects are interconnected. Prior studies found the characteristics of participant-affiliated network, such as network distances or network density, had significant impact on the OSS performance [21]. After five-year observation on more than 2000 OSS projects hosted in Sourceforge.net, Singh and his collaborators [27] found internal collaborative cohesion had significantly positive impact on the OSS performances, manifested by the extent of CVS commits, but the external cohesion presented an inverted-U impact on the OSS performances. Hahn et al. [8] found the OSS developers preferred to join in those projects initialized by those who had collaborated before in terms of investigating the collaborative ties. Grewal et al. [7] argued the OSS performances was significantly influenced by the extent of network embeddedness in terms of studying the developer-affiliated network.

2.2.2 Intra-project Network

Besides the inter-project network, prior studies also found that the intra-project network also played a role on OSS performances [22]. Differing from the works studying OSS performances, the intra-project studies mainly discussed the individual collective behaviors. For instances, Conaldi et al. [10] established an analytical model for collaborative network and empirically verified it by using the debugging network from a large OSS project. Sowe et al. [23] categorized the participants from Debian (a leading Linux OS distributor) into three key roles, which included knowledge seekers, knowledge contributors, and knowledge brokers in terms of analyzing the emailing communication network, and argued that the knowledge brokers could facilitate the information flow and distribution. By studying the internal communication networks of two leading OSS projects, Singh and Tan [24] found the stability and efficiency cannot be concurrently reached, and advocated the OSS project administrators had to adjust the balance based on their key goals.

In sum, we can find the literatures on inter-project network mainly discussed how OSS projects could be outperformed in terms of network ties or vertex positions. The intra-project network studies mainly concentrated on particular behaviors or actions in OSS project development, such as debugging, communications, or collaborations. A brief summarization is given in Table 1.

Table 1. Summarization of prior OSS studies using network analysis
Full size table

2.3 Policy and Legalization

Comparing with the works studying individual motivations or network characteristics influencing the success of OSS project, the studies in third stream mainly discussed how external policy or legalization affect the OSS performances. In particular, two sub-streams in this domain can be found, i.e. the characteristics of OSS licenses and the intellectual property lawsuits enforcement [13–15].

The OSS supports encouraged the source codes could be shared, revised, and redistributed, however, the rights of the intellectual property of the creators cannot be overlooked. In this regard, several OSS licenses have been given to restrict the right of usage or revisions. There are more than 60 different OSS licenses prevailing throughout the world. The extent of restrictiveness of those OSS licenses is different. For instances, comparing with BSD whose restrictiveness is very low, the GNU license is widely known for its restrictive copyleft. Prior studies found OSS project administrators and users preferred the highly restrictive OSS licenses, but the OSS developers did not welcome such license [13]. In addition, Sen et al. [14] found the OSS project with moderately restrictive OSS license cannot attract the developers and users. Besides the effectiveness of OSS licences, Wen et al. [15] found the intellectual property lawsuits enforcement would increase the OSS usage cost, which would reduce the OSS participants’ interests.

3 Assessment of OSS Success

Several metrics have be employed to assess the OSS success, and such assessments were developed in consideration of the research propose and the audiences, such as number of subscribers, number of active developers, number of downloads, frequency of CVS commits, and the extent of code reuse etc. [25]. After comprehensively reviewing prior literatures, the number of downloads and the frequency of CVS commits were most widely used to assess the OSS project success. However, we though either number of downloads or frequency of CVS commits cannot best represent the performance of OSS, i.e. success of OSS success. The number of downloads can only serve as a proxy indicator of software exposure, but cannot represent the success of an OSS due to the unknown conversion rate [25]. For instances, end users can simply remove their downloaded OSS after a quick trial, which cannot be reflected in the number of OSS downloads. To reconcile such defect, some scholars advocated to use the frequency of CVS commit to assess the success of OSS projects. Comparing with the number of downloads which represent the extent to which OSS prevails in end-users, the frequency of CVS commit serves as a proxy indicator of development vitality. The higher frequency of CVS commit indicates the developers constantly contributed to the OSS project. However, the limitation of such indicator is also obvious. The higher frequency of CVS commit may also indicate the fundamental quality of this OSS project is not good, and the developers had to constantly fix the bugs found in the newly distributed versions.

Encouraging the literatures in innovation and strategic management, we thought the extent of code reuse could be a representative indicator to assess the success of OSS project. Haefliger et al. [26] found the qualified sources codes were extensively reused in other OSS projects. In innovation literatures, the innovation capability of an organization is measured by the extent to which their patents are cited. In similar vein, we argued the extent of code reuse could serve as an important indicator to assess the OSS project success, i.e. the innovation performance.

4 Conclusion and Future Agenda

In this article, we provide a comprehensive literature reviews on prior OSS studies. In particular, we categorized the previous literatures into three streams based on the research topics, which included individual motivations to contribute or participate in the OSS projects, the relationship between network characteristics and OSS performances, and the external factors like OSS licenses and policy influencing the OSS performances. In addition, we also surveyed the prior literatures and summarized several key indicators assessing the OSS project success, and found the pros and cons of the existing measurement. Encouraged by the innovation literatures, we expected the future research could use the extent of code reuse to indicate the innovation quality of OSS.