The dynamics of information system development in developing countries: From mutual exclusion to hybrid vigor

The integration of isolated software components into large‐scale and complex information systems is a topic that attracts the interest of many information systems practitioners and researchers. However, less attention is given to the intricate processes by which these infrastructures are built using multiple independent software components. Different actors introduce these components, which can lead to a “battleground of functional roles” where components are replacing and blocking each other. To better understand this phenomenon, we conducted multiple case studies involving the development of a large‐scale information system for communicable disease detection, prevention, and control in Vietnam. We collected empirical data through various methods, including interviews, focus groups, and participation observation. We identified different strategies employed by actors to block, remove, or replace other software components, protect their position, and find a way to join the existing and established systems. This paper contributes to this topic by theorizing a strategy called “hybrid vigor,” which allows for the combination of the strengths of individual software components and enables sustainable evolution.

toward a certain goal is an important technique espoused by both industry and academia. The tenet of this approach advocates designing and building a system from what already exists rather than from scratch. In other cases where there is no (or little) installed base to utilize, one can incrementally build systems through the strategy of bootstrapping, referring to a process of producing and providing incentives to mobilize the first users of a technology and using this set of users as an installed for further development.
Recently, approaches to the information infrastructure design have been augmented by several researchers in a variety of ways. Sanner et al. (2014) proposed grafting as a strategy to embed a component in the network of components by making sure this component can adapt well to the existing installed base through various social-technical adjustments. To that end, the choices of the right place and the right time are crucial for the success of the grafting. Other researchers have recognized the role of architecture in shaping development trajectories, arguing that a resilient and open architecture is critical for information infrastructure successful evolution. Others have called for partial or complete devolution of control over information infrastructures to maximize the innovation and creativity to enable success and sustainability of information infrastructures.
Until now, most research on information infrastructure focuses on finding innovative approaches that help cultivate information infrastructures through various strategies such as bootstrapping and recently grafting. In such studies, constitutive elements or components are introduced to either be extended, complement or be integrated with the existing network. Within the scope of this research, we employ the following definition by Szyperski et al. (2003) on software components: "A software component is a unit of composition with contractually specified interfaces and explicit context dependencies only. A software component can be deployed independently and is subject to composition by third parties" (p. 41) We believe this definition is generic and flexible enough to use in a variety of contexts. As the unit of evaluation of our research is information infrastructures which are usually complex information systems, a software component could be understood as a sub-system or a module that have their own stakeholders (owner, development team, sponsor, advocator) and development agenda.
Till now, not many studies have been conducted on situations where multiple software components are deliberately introduced to replace existing ones and potentially block other alternatives in future replacement attempts. One exception is the work by Nielsen and Saebø (2016), who conceptualized the interplay between software components and the strategies behind different, potentially competing software components as "functional architecting." They distinguish three strategies used by proponents of different software components: charting, encroaching, and connecting. This article extends the work of Nielsen and Saebø by focusing on situations where different software components meet and established functional architecture is challenged and changed. We will illustrate that the consequences of these strategies can lead to severe clashes between old and new components. Conflicting strategies may result in the replacement of existing components and potentially the entire network by a single new component.
Our specific focus is on situations where multiple actors directly compete to serve the same functional needs and struggle for their own existence. Therefore, the purpose of this study is to understand how relatively weak components can fight for survival. Our research question is: "What strategies can be implemented to prevent mutual exclusion and promote the evolution of information infrastructures?" To answer this question, we conducted a multiple-case study of four individual cases involving the negotiation, development, and implementation of information systems for communicable disease monitoring in Vietnam between 2010 and 2015. We were directly involved in one of these initiatives and participated in general discussions and collaboration with stakeholders in the other three. Through our analysis of these cases, we discovered a different strategy for functional architecting called "hybrid vigor." We define hybrid vigor as a strategy that aims to improve the overall functional attributes of an infrastructure. Our definition also includes three dimensions of hybrid vigor: the politics of functional negotiation, the power of governmental agencies, and reconfiguration of the resulting infrastructure.
We chose to conduct our study in developing countries in general and Vietnam in particular for a number of reasons. Firstly, developing countries receive significant funding from international donors to develop information systems to monitor their development goals (Braa et al., 2007).
These funds create multiple information systems with overlapping functionalities and often lead to them being replaced by one another. Secondly, Vietnam is a rapidly growing country with a large population. Large donors such as the Asian Development Bank, United States Agency for International Development, and the US Centers for Disease Control are all present in Vietnam.
The remainder of the article is organized as follows. In Section 2, we provide a review of related research. In Section 3, we present the methods and approaches used in this study. In Section 4, we introduce the case of epidemic reporting systems in Vietnam, and in Section 5, we provide our analysis and discussion. The article is concluded in Section 6.

| Information infrastructure, and health information infrastructure
Conceptualizations of large and complex information systems focus on networks or networks of systems that are comprised of heterogeneous components. Hanseth et al. (2001) characterize them as "shared, open, evolving, heterogeneous, and socio-technical systems" (Hanseth & Monteiro, 1998). Information infrastructures, a genre of information systems, cannot be built or constructed using conventional software engineering methods, which are developed for standalone and homogeneous systems (Somerville et al., 2013).
There is a common argument in the literature on information infrastructures that the complexity, uncontrollability, and unintended consequences associated with these systems are the root causes of the failure of many large IT projects (Aanestad & Jensen, 2011). The development and implementation of large-scale and complex information systems require new approaches that view technology as more than individual tools (Tilson et al., 2010). To address this demand, design theory has been developed by Hanseth and Lyytinen (2010) to tackle the dynamic complexity in the design for information infrastructures by addressing two key challenges: bootstrap and adaptability. The theory outlines a list of design principles that guide designers on how to "generate attractors to propel users to adopt the IT capability so that its growth will reach momentum" (Hanseth and Lyytinen, 2010, p. 8) and "guarantee that the information infrastructure will grow adaptively and re-organize constantly with new connections between information infrastructure components" (Hanseth and Lyytinen, 2010, p. 13).
Another important aspect of large-scale and complex information systems is the lack of centralized control and, as a consequence, the nature of how they change. The process of actualizing an information infrastructure in a piecemeal and incremental manner, based on what already exists, is referred to as cultivation. The installed base of the information infrastructure slowly evolves, emphasizing that due to its complexities, an information infrastructure is hardly built or designed quickly. It is not practically possible to ignore the legacy of the "historical accumulation of socio-technical arrangements around it" (Sanner et al., 2014, p. 221).
Accepting the complex nature of information infrastructures has led researchers to study the different strategies pursued by different actors within these large-scale and complex networks. For example, Sanner et al. (2014) recently introduced the concept of grafting to describe a strategy where "local organizational goal-oriented information system innovations become viable extensions of shared and evolving information infrastructure" (Sanner et al., 2014, p. 221). Identifying the right moment and position to graft a novel software component to the rootstock (the installed base of software systems) is critical for success. The grafting strategy also entails effectively managing relationships with stakeholders who are in control of parts of the existing installed base and the deliberate choice of rootstock to ensure compatibility. Another example is the concept of co-evolution, which focuses on how different technologies developed by different actors may intersect from time to time. Jansen and Nielsen (2005) introduced and applied the concept of co-evolution to analyze the parallel evolution of information infrastructures using the case of two wireless communication platforms, universal mobile telephone system (UMTS) and wireless local area network (WLAN) in Norway. They concluded that the intra-and interlinkages of technologies, politics, interests, and user preferences within each information infrastructure strongly influence the trajectory of the co-evolution. In a similar vein, based on the work of Benbya and McKelvey (2006), Shaw (2009) developed a co-evolutionary framework for managing the complexity of hospital management information systems development in resource-constrained settings. Another example is the literature on platforms, discussing the distribution and redistribution of control and innovative capacity between different actors and the technologies they control. For example, Tiwana et al. (2010) argued that the evolutionary dynamics of information infrastructures are significantly influenced by the "co-evolution of endogenous choices by platform owners and the dynamics of an ecosystem's exogenous environment" (Tiwana et al. 2010, p. 687). Additionally, several scholars, including Nielsen and Aanestad (2006) and Hanseth et al. (2001), cleverly use the concept of devolution to describe a situation in which information infrastructures might better evolve if centralized control is balanced with autonomy. The reduction of strict control on the information infrastructure growth would permit "the distribution of resources, risks and abilities and willingness to innovate" (Nielsen & Aanestad, 2006, p. 185).
The existing literature on information infrastructure primarily discusses the introduction and combination of innovative systems, subsystems, and components into existing social-technical arrangements, and how these constituents shape and are shaped by social, technical, and institutional contexts. However, there is a lack of research that clearly addresses the overlap and duplication of new and incumbent software components, their competition, and potential substitution. An important exception is the work of Nielsen and Saebø, who conceptualize the interplay between and the strategy behind different and potentially competing software components as functional architecting. They identify three different strategies when a software component from one domain is moved into another, such as when a corporate accounting system also offers functionality for human resources management. Charting is one strategy where a software component is moved into a new domain to meet an unmet functional need. Connecting is another strategy where software components from different domains are connected to leverage the benefits of complementary functionality. Finally, encroaching is described as a strategy where a software component is moved into a new domain by offering functionality in direct competition with functionality already provided by other components.
Health information systems in developing countries, such as in our case study, are often incomplete, unreliable, obsolete, and of poor quality (Braa et al., 2004(Braa et al., , 2007Braa & Hedberg, 2002;Garde et al., 2007;Haux, 2006;Heeks, 2006;Heeks et al., 1999). Despite substantial investment from local governments and international donors, the situation has not improved, and in some cases has worsened, due to the lack of effective coordination and technical competence, as well as poor and possibly corrupt governance. In many cases, poor coordination leads to wasteful overlap and duplication of investment in software systems because donors and government agencies support the development of disparate systems that provide the same functionality for the same setting, making them direct competitors. Braa et al. (2004) raised another problem in public health systems, which they refer to as the "all or nothing dilemma." There is a common need to scale ICT innovations to full coverage, that is, all districts in a province or all provinces in a country, to make them useful to health managers. However, different systems are typically implemented in an uncoordinated and fragmented fashion, and there is no single system that meets the full coverage criterion. These systems are at risk of being substituted by new systems that cover a larger area. At the same time, developing countries are commonly too poor to afford the substitution of all directly competing systems.

| From system builders to hybrid vigor
The information infrastructure literature depicts the development and evolution of infrastructure as emergent and out of limited central control, but there is still room for intervention. Based on his work on the history of the electrical system in the US, Hughes coined the term "system builder" (Hughes, 1979(Hughes, , 1987. Hughes showed how this electrical system was not only technical, but also related to and based on various institutions, manufacturers, and investors. Such systems cannot be satisfactorily treated in isolation from organizational, political, and economic matters, but must be integrated with their context to work and grow. To facilitate this integration is the key role of the "system builders"-the creators of large-scale and socio-technical systems. "System builders" preside over technological projects from the concept and preliminary design stages through research, development, and deployment. In order to do so, they need to cross disciplinary and functional boundaries and become involved in funding and political stage-setting. The core competency of the "system builders" lies in their ability to integrate heterogeneous physical, human, and organizational components into a working and goal-oriented system: "to force unity from diversity, centralization in the face of pluralism, and coherence from chaos" (Hughes, 1987). According to Hughes, "system builders" should have a holistic focus and see the entire system, rather than only its components. Through control and management, and with attention to the interconnection between the system's different components, "system builders" believe that the system will not evolve and grow without someone viewing it as a coherent system. Inspired by the concept of "system builders," we conceptualize the hybrid vigor strategy to capture and understand situations in which new systems are introduced and live side-by-side with existing ones, and what is needed to make them do so. As we have discussed earlier, there is a gap in the information infrastructure literature that deals with directly competing software components. So far, information infrastructure researchers have rarely provided discourses on the issues of co-existence of direct competing software components and solutions to address that challenge. One exception includes the case of asymmetric integration reported by Sahay et al. (2009). In their study, Sahay et al. (2009) criticized that integration is not merely technical issues but involves political and institutional interests of different stakeholders. Even though the integration of different software components could contribute to the evolution of information infrastructures, it is hard for more powerful actors to easily accept new components. That raises the issue of asymmetric integration, i.e. unbalanced power distribution. The solution is that the smaller actor must configure their systems to match with the more dominant ones. However, Sahay et al. (2009) did not highlight the issue of directly competing software components that offer overlapping functionalities. In such cases, a system can be easily substituted by other systems backed by more powerful stakeholders such as governmental agencies, international donors, etc.
We borrow the term "hybrid vigor" from biology, where it is used to describe the improved or increased function of any biological quality in a hybrid offspring (Shull, 1948). This occurs through cross-breeding of plants or animals from the same species or genus.
We have decided to use the term "hybrid vigor" in the context of information infrastructure development for several reasons. Firstly, the development of information infrastructure has many characteristics similar to the growth and evolution of biological ecosystems (Manikas & Hansen, 2013).
The development of information infrastructure is incremental, as it occurs in small steps, complex, as it involves many components and depends on internal and external factors, and evolutionary, as it is gradually upgraded in terms of functionalities and complexities (Henningsson & Hanseth, 2011).
Secondly, information infrastructure researchers have been extensively using terms borrowed from biology to describe the development of information infrastructures. For example, Hanseth and Monteiro (1998) used the term "evolution of information infrastructures" to describe the development of information infrastructures as a natural and biological process, similar to the growth of biological entities. They also compare this process to the growth of potatoes. Hanseth (2010) adopted the term "cultivation," an agricultural process in which seeds or trees are planted, to propose a strategy for building information infrastructures. Sanner et al. (2014) used the term "grafting," a biological technique that helps transplant parts of one tree to another, to refer to the process of adopting, transferring, and implementing technology, often from North to South.
All in all, there are some limitations on the use of these biological concepts given the dynamic nature of information infrastructure development.
The concept of evolution is relatively generic and descriptive, that is, it is mainly about how the information infrastructure grows but not about how to design information infrastructure to enable its healthy growth. The concept of cultivation has been criticized for its lack of precision in capturing specific goal-oriented organizational interventions (Sanner et al., 2014). Therefore, we still need additional concepts in other to unpack the complexities of information infrastructure development. As such, we propose the use of hybrid vigor to better understand the information infrastructure development process. Hybrid vigor is a strategy used under circumstances in which direct competition between software components exists, and where certain actors enable the combination of components that offer almost identical functionalities. Hybrid vigor is the strategy of actors understanding the potential of and pursuing the opportunities which the cross-breed of rival software components can yield the evolutionary information infrastructure.
In situations of direct competition, existing components and actors run the risk of being relegated and replaced. In some instances, only one "winning" component that is best adapted to the environment will remain, while all others will disappear. The hybrid vigor strategy is based on the belief that such an outcome would be suboptimal. If competing components can find a way to coexist and collaborate instead of mutually excluding each other, then information infrastructures will evolve more dynamically, support innovation and offer more flexibility. In this case, the strengths and weaknesses of each component will be combined and complemented, promoting flexibility and potential for growth. The three strategies developed by Nielsen and Saebø (2016) for functional architecting did not fully capture the dynamics of how key system builders, including government and non-government agencies, regulations, and politics, are shaping the context. Our contribution is the conceptualization of hybrid vigor as a fourth strategy for functional architecting in such contexts. We further explore the hybrid vigor strategy in our case study of the development of several communicable disease reporting systems in Vietnam by identifying and outlining its various dimensions. Our theoretical framework is presented in Figure 1.

| Research method
This research is based on multiple case studies, as chosen for its ability to provide in-depth exploration of complex phenomena within specific contexts, as noted by Rashid et al. (2019). The aim is to understand the dynamics of software component interaction and integration in developing country settings. The multiple case study approach allows for a deeper understanding of how these components evolve and keep pace with changes in their peers, given the subject of the study is the interplay of components backed by different stakeholders. By analyzing similarities and differences between cases, both within and across cases, a more comprehensive picture can be obtained. This study is considered an embedded multiple case study rather than a holistic multiple case study, as it examines each software component as a subpart of a larger information infrastructure, making data collection and analysis easier. However, this method demands a larger effort for data collection and analysis and can be time-consuming and prone to errors.
Our research approach comprised several steps. Firstly, we tracked the development and implementation of the Vietnamese Epidemic Notification System (ENS) by the Administration of Medical Services (VAMS) following a devastating measles outbreak in early 2014 that claimed the lives of hundreds of children. The first author of this paper was an active participant in the VAMS effort, both in a managerial and technical capacity.
Secondly, we investigated three additional initiatives that were running parallel to VAMS and aimed to support data collection, reporting, and monitoring of communicable diseases in Vietnam. These three cases include the Electronic Communicable Disease System (eCDS) backed by the General Department of Preventive Medicine (GDPM), a new version of eCDS developed by a state-owned telecom company (VCom), and the Communicable Disease Dashboard System supported by international NGOs (iNGOs).
In addition to separately studying each of these four cases, we also focused on their intersections and the collaboration between different actors coordinated by iNGOs to identify opportunities for a joint effort.
Using the case study approach, case selection is crucial (Merriam, 1998). While being partially opportunistic and based on our access to the field, our selection of cases was primarily driven by our goal of comprehending the developing situation related to disease surveillance systems in Vietnam. The combination of the strengths of these cases lies in both access and richness. On one hand, our long-term involvement and access to the health-care sector in Vietnam and communicable disease monitoring, in particular, was essential for a rich and longitudinal study which provides useful insights into the evolution of information infrastructures. On the other hand, our use of extreme case selection method-selecting a F I G U R E 1 The conceptual framework of information infrastructure design strategies. case based on its extreme value of the independent or dependent variable-was also influential (Seawright & Gerring, 2008, p. 301). This supports our aim of theorizing hybrid vigor as extremes or ideal types often define theoretical concepts (Henfridsson & Bygstad, 2013, p. 914). We believe that cases involving four separate attempts to build systems with similar functionality are uncommon.

| Data collection
The collection of data for the first case was carried out simultaneously with the development and implementation of the ENS, in which the first author was directly involved. The other three cases were conducted between late 2015 and early 2016 and were initiated by the first author, who was invited to participate in a joint effort led by iNGOs to integrate existing communicable disease reporting systems. In this case, our research can be seen as action research, blending software prototyping and problem solving for organizational change (Davison et al., 2021). We also took into consideration the recommendations for maintaining ethics in action research (Davison et al., 2022).
In gathering data for the study, we followed the approach outlined by Benbasat et al. (1987) who emphasized the importance of using multiple sources of data to increase the reliability of the research. To ensure that no precious data were lost and to maximize the time spent on sites, we carefully planned our data collection and meticulously kept records during and after the collection. Our methods of data collection included participative observation, interviews, focus groups, and archival records.
Participative observation was employed in our collaborative efforts with iNGOs. Our exposure to meetings and events gave us the opportunity to absorb and take note of the details, actions, and subtleties of the field environment (Benbasat et al., 1987, p. 374). In total, we attended four meetings organized by the iNGOs, one of which lasted for a full day. These meetings provided us with insights into the architectural and functional design of different systems, which was crucial in understanding the issue of direct functional competition between software systems.
The participative observation method was also extensively used in the case where the first author was involved in developing and implementing a system. This involvement gave us a unique opportunity to observe the political and technical negotiation processes. However, being too closely involved in the case we were studying could lead to potential biases, which we were aware of and deliberately tried to avoid. To address this, we applied data triangulation techniques and purposely sought neutral and alternative explanations for the same phenomena. The second author played a key role in asking critical questions.
Secondly, we also collected data through interviews, including focus group interviews. These were conducted to gain broader information about systems we were not directly involved in and to validate and discuss data gathered from other sources. This helped to resolve any contradictions and inconsistencies. We conducted a total of eight interviews, each lasting approximately 30 minutes to an hour. The first group of informants included staff responsible for operating the eCDS at the General Department of Preventive Medicine (GDPM) (4 interviews). The second group consisted of managers and developers working for VCom on the project to redevelop the eCDS system at GDPM (4 interviews).
The summary of interviews is provided in Table 1.
To obtain a comprehensive understanding of the development and implementation of both the eCDS system supported by GDPM and the updated version developed by VCom, we carefully selected our interview participants from various backgrounds and positions. We believed that they could provide valuable and unique perspectives. To ensure that the data collected was comprehensive and detailed, we employed a combination of open-ended and semi-structured interview methods. Open-ended questions, which allow respondents to provide unstructured answers, are often used in qualitative research to gather subjective information and capture respondents' thoughts and feelings in their own words (Kvale, 1983). Semi-structured interviews, which incorporate elements of both structured and unstructured interviews, are commonly used in qualitative research to gather detailed information and explore complex issues, such as in our study (Kvale, 1994). All interviews were recorded and transcribed for analysis. Additionally, the author conducted various informal interviews with iNGOs, GDPM, and VCom, which enriched the data collected from other sources.
To gather collective views, experiences, and beliefs from participants from GDPM, iNGOs, and VAMS, we used focus group discussions during four meetings organized by iNGOs. The meetings took place between October 2015 and January 2016, and the purpose varied depending on T A B L E 1 Interviews and informants. the time and participants. The focus, however, was to ensure that all stakeholders understood the current situation of each system and to provide opportunities for expression of interests and discussion of plans.
Thirdly, the method of archival records was also employed to gather data that offered general information regarding the legal aspects of communicable disease administration in Vietnam. Email exchanges, project reports, and proposals were some of the types of archival records used. Table 2 describes the list of significant archival records.
The data collection process highly influenced by our prior hypotheses, assumptions, and knowledge about the research topic was conducted in an iterative manner. That means that data analysis was performed simultaneously with the data collection process. In general, our coding procedures started with predefined codes and was gradually augmented by codes emerged from the collected data. In this aspect, our thematic analysis can be seen as mix between framework analysis and grounded theory. Framework analysis involves organizing data into pre-determined categories or frameworks.
On the other hand, grounded theory allows the data to drive the development of themes, rather than starting with pre-determined categories. The reason for this was we first started with an attempt to understand II evolution but later we shifted into mutual exclusion and competing systems.
When analyzing data, we did not use a software but rather a manual process. Despite its advantages using a software for data analysis has also some limitations (Saldaña, 2015). There are concerns when using software for qualitative data analysis that it may lead to more rigid and predetermined processes, a focus on coding and retrieval techniques, and the loss of meaning and depth in the data. Additionally, researchers may feel pressure to prioritize quantity and breadth over quality and depth (John & Johnson, 2000). However, the decision not to use software for analysis in this paper was simply a personal preference.
As we followed qualitative tradition in this research, qualitative analysis method was employed. In qualitative research, categories and codes are two different ways of organizing and analyzing data (Braun & Clarke, 2006). A category is a broad grouping of similar data or concepts. It is a way of organizing data into general themes or topics. Coding, on the other hand, is a way of identifying specific pieces of data or segments of text within a larger dataset that are relevant to a particular category or theme (Braun & Clarke, 2006).
In our study of II for health care, we first identified several broad categories: "II design principles", "II development strategies", "II control", "II evolution and devolution". These theme guided our data collection and analysis. However, in the later phase, we found out more categories that we found relevant to our study such as "II functional overlapping", "II mutual exclusion", "II direct competition".
In regarding to data analysis process, within the category of "II development strategies", we have coded a number of factual data: • "iNGOs wanted a data warehouse for data sharing and decision making, a collaborative approach was proposed to re-use existing systems" This factual data were collected during a number of archival record analysis from email exchanges between stakeholders. For example, the following email extract sent by iNGOs on January 25, 2016: If it's okay, I'd love to join the meeting on Wednesday as well. Perhaps we can meet briefly afterwards to discuss how to link laboratory data into the DW, and how we can use existing tools within the DHIS2 platform to support laboratory messaging (e-referral, result reporting, etc.). (Email archive) • "VAMS proposed to use their system to serve as a intermediaries to collect data from health facilities" This coded data were analyzed based on an interview with a VAMS's specialist: VAMS systems connect well with health facilities. It could enable a faster data collection link than GDPM-health facility one.
(Interview transcripts, December 10th 2015) T A B L E 2 Data sources used in the research. • "GDPM needed more functionalities and greater vertical scale to their current system" This coded data were found when analyzing project document archives (email) and informal meeting notes with stakeholders: Longer term, VN MOH wants to expand the system past communicable diseases to encompass all diseases. That is Dr. P request stated several times -opening and closing remarks -with emphasis to plan with that scale in mind. (Email archive, dated 26th January 2016) Through a number of reflections and rereading the materials, larger themes were identified, which was an important step in finding answers to the research questions. Initially, we traced the competition among software systems and early conceptualized hybrid vigor as a strategy that helps reconcile and neutralize conflicts between systems' proponents. Examining the strategy, we gradually discovered multiple layers of its dynamics and delineated several aspects, which we referred to as dimensions of hybrid vigor.
To reduce bias and increase credibility and validity of our research, we attempted thorough research triangulation. Triangulation involves the use of multiple data sources, methods, or theories in order to triangulate, or cross-check, findings. The goal of triangulation is to provide a more complete and accurate understanding of the research question (Denzin & Lincoln, 2011). In this research, data triangulation and investigator triangulation was used. Data triangulation involves using multiple data sources, such as interviews, surveys, and observations, in order to provide a more comprehensive understanding of the research question (Denzin & Lincoln, 2011). Investigator triangulation uses multiple researchers to collect and analyze data in order to reduce bias and increase the credibility of the findings (Denzin & Lincoln, 2011).
The following extract shows an example of investigator triangulation between the first and second authors when discussing using a diagram to illustrate the relationship between the software components: still think the figure is a bit confusing. And figures should clarify, not confuse. My current concern is the arrowsthey are illustrat- Apart from investigator triangulation, we also employed data triangulation to increase accuracy and validity of our data. An example of data triangulation was when we interviewed GDPM, we assumed that infectious disease reporting data from health facilities were transferred electronically to the eCDS. However, when visiting GDPM offices and looked the software, we realized that data were transferred by papers and later entered into the system manually.
As part of the analysis process, the cases description is presented in the following session.

| FOUR INITIATIVES ADDRESSING THE COMMUNICABLE DISEASES REPORTING
This section presents the four implemented initiatives and systems to address data collection, reporting, and monitoring of communicable diseases in Vietnam, described in chronological order.
4.1 | The electronic communicable disease reporting system (eCDS) The preventive medicine reporting system in Vietnam can be characterized as "passive," with its primary focus being data collection and reporting rather than being used for epidemic control and prevention. This weakness is attributed to a lack of skilled staff, inadequate infrastructure, and the absence of standards and protocols for reporting and monitoring outbreaks. In 2008, the Vietnamese National Assembly passed the Law on The pilot phase aimed to evaluate the software and determine the necessary infrastructure for successful nationwide implementation. The implementation team consisted of staff from GDPM, PHSS, VAHIP, WHO, and regional epidemic control institutes, who provided training and continuous support to health workers at the provincial and district levels. The pilot phase was successful, with the software providing most of the required functionalities for communicable disease reporting as mandated by Circular 48. The system allowed users from various levels to access and enter data, which facilitated collaboration and ensured interoperability of data from multiple sources.
The positive outcome of the pilot phase led to increased interest from donors in expanding the implementation of the information system across Vietnam. ADB continued their support by sponsoring the implementation, training, support, and evaluation of the system in 45 provinces.
WHO provided support for servers, hosting, and hardware services and also sponsored the implementation, monitoring, and evaluation in three provinces. The Centers for Disease Control and Prevention in the United States also supported the implementation in seven provinces. GDPM used their budget to implement the system in the remaining eight provinces.
By June 2014, the information system had been implemented in all provinces, including 11 that had implemented it down to the ward level.
In provinces where it was not implemented at the ward level, district PMCs entered data from paper reports sent by ward clinics. Despite the potential to provide full coverage of timely and sufficient communicable disease data, it was not the case. Circular 48 regulated reporting templates that health facilities and PMCs must use to report communicable disease data in weekly, monthly, and yearly basis. Fatal and severe communicable diseases classified as Group A by Circular 48 required immediate reporting for timely intervention. The system included upward reporting flow (district health centers to provincial health centers), peer reporting flow (district hospital to district health center), and feedback flow (upper-level PMCs send confirmed case lists to lower-level PMCs for verification and intervention). However, due to a lack of adequate skills and equipment in PMCs at district and provincial levels, most communicable diseases confirmed cases were only diagnosed and discovered by hospitals. 1 consistently inadequate, incorrect, and delayed data collection. PMCs require infectious disease data for their interventions, but they cannot actively and directly collect it. They are forced to rely on hospitals, causing significant delays in data reporting. This has led PMCs to frequently send staff to larger hospitals for the sole purpose of data collection, resulting in missing or delayed data from other hospitals.

| The epidemic notification system (ENS)
In 2014  A key challenge in this situation was that no agencies or authorities could provide accurate information on fatalities, infections, and hospital admissions. The official source of the Ministry of Health reported 25 deaths, but data from one hospital indicated nearly three times that number.
It became clear that Vietnam lacked appropriate and effective information systems to support the monitoring of communicable diseases. At the time, the eCDS, backed by the GDPM, was still in its pilot phase. Its original design focused on providing detailed data needed to control a measles outbreak, such as the total number of infections and fatalities, but did not support tracking data like the number of inpatients with serious complications. This information was critical for the efficient allocation of scarce resources, such as skilled doctors and respirators. To ease the situation, an official at VAMS discussed with the first author the feasibility of using the annual accidents and injuries reporting system (AIRS) to create a daily reporting system to monitor the measles epidemic. The first author had previously supported the Ministry of Health in implementing systems such as a health professionals licensing system, hospital inventory, and patient complaint system, which were part of a larger infrastructure covering most medical examination and treatment activities. A team was quickly formed to work on the extension and, to speed up the process, they reused most of the functionality, such as data entry forms and data dictionary management, from an existing opensource platform in Vietnam, DHIS2. After 2 weeks of intensive work, the team completed the basic data entry and analysis system.
The success of prior systems convinced MoH leaders to approve the rollout plan. An official letter was sent to all provincial health departments and hospitals, requesting them to use the system daily to report measles and other infectious diseases. Recognizing the urgency of the epidemic, most health facilities agreed to the request and began reporting data immediately. A few days later, the system collected and consolidated data on the nationwide measles epidemic, playing a crucial role in further controlling the outbreak. Compared to the eCDS, the system developed by VAMS provided more detailed data, including not only basic information about infectious and fatal cases, but also monitoring the treatment progress of measles patients. The dataset included additional elements such as the number of discharged or transferred patients, lab tests, patients in critical condition, and more, totaling over 20 additional elements compared to the eCDS. This justified having an additional system focused on collecting clinical data for communicable diseases. Figure 2 shows an example of daily admissions for measles reported by the VAMS system from May to October 2014 from hospitals across the country.

| The new electronic communicable disease system (the new eCDS)
VCom is a state-owned corporation that initially operated only in the telecom sector but has since expanded into software and services. The company has a significant international presence, including countries in Asia, Africa, and Central America, and has gained considerable success. With the slowdown in the growth of the telecom market, VCom has gradually broadened its business scope. In the software sector, the company has ambitious plans to develop enterprise resource planning (ERP) solutions, e-Commerce, and e-Government-related products.
In 2015, after several rounds of discussion, VCom signed an agreement with the Ministry of Health (MoH) to collaborate in the development and implementation of ICT solutions for healthcare until 2020. Since then, VCom has worked closely with various departments within the MoH to conduct situation analysis and propose plans to improve or redevelop the systems being used. Most of the existing systems have therefore been substituted, developed or redeveloped using VCom's architecture and platform. eCDS to more modern frameworks like Model-View-Controller (MVC). Additionally, the team identified several design flaws in the eCDS that posed security risks and could negatively impact performance. The team's proposal to redevelop the system was supported by the leaders of GDPM.
Initially, the team estimated that the project would take only three months to complete, but various technical and communication challenges prolonged the timeline. After one year, the first release of the new eCDS was complete and ready for pilot testing. The new system includes all of the features offered in the previous eCDS, as well as new features such as an automated case-based data entry process. This process was added to increase the speed and accuracy of data reporting by reducing manual data entry, as most communicable disease cases are discovered and diagnosed in hospitals.
The data exchange plan involved exchanging information between the VAMS epidemic notification system and the new eCDS through either web services or file transfers. In the latter case, a spreadsheet-based template was used for data interchange. To support this collaboration, technical meetings were held with representatives from GDPM and VAMS. However, progress was slow due to a lack of resources from VAMS to upgrade their system.

| The communicable disease data warehouse (CDDW) for emergency operations center (EOC)
In 2014, Vietnam joined the Global Health Security Initiative (GHSI), a US-led program to enhance global collaboration in detecting, preventing, and rapidly responding to infectious disease threats. The initiative, joined by 25 other countries and key international agencies such as the WHO, FAO, and OIE, was supported by the US CDC. With this support, Vietnam launched a project to improve its laboratory systems and establish an Emergency Operations Center (EOC). For the successful running of the EOC, it is crucial to have an operational data warehouse for infectious diseases. CDC and its implementing partner PATH have worked with local stakeholders, including GDPM, VAMS, and regional health institutions, to secure resources and define the processes, tools, and technologies needed for the data warehouse. A CDC team member emphasized the significance of this initiative, saying: The data warehouse and visualization platform will offer opportunities for historical analysis, trend detection, and rapid disease outbreak detection and prediction, leading to improved resource allocation, program planning, and communication and collaboration between national and international public health entities. (Pre-meeting materials) At a meeting between CDC and VAMS to discuss the infectious disease data warehouse, VAMS welcomed GHSI's support and requested assistance from CDC and partners to upgrade their current epidemic notification system. A Vice Head of VAMS recommended the CDC team to meet with the first author of a related paper for further discussion. Afterward, a short meeting was held, where a CDC technical expert asked the first author about the existing electronic patient records system in Vietnam and suggested reusing existing systems to avoid duplicated data entry.
A few months after the initial meeting, the first author was invited to attend a series of technical meetings at the PATH office in Hanoi, where the health information system (HIS) leads from multiple MoH agencies, implementing partners, and donors gathered to foster data sharing and communication between the systems and plan for the development and implementation of the data warehouse. During these meetings, three important processes were identified: identifying a data source for the data warehouse, defining a list of essential data elements, and consenting to health information exchange mechanisms needed for the data warehouse operations. The meetings resulted in several decisions that stimulated collaboration among stakeholders. To quickly produce useful results, the focus was given to building tools to analyze and visualize the collected data on graphs and maps, to support informed decision making. Additionally, training and other capacity building activities would be organized to prepare adequately skilled personnel for operating the emergency operations center (EOC).

| The current relationships among the different infectious diseases systems
In this section, we'll examine the relationships between the systems. Circular 54 and new systems like the VCom-backed eCDS and iNGOadvocated infectious disease data warehouse have disrupted the dominance of eCDS and ENS. To maintain their presence in the ecosystem, both eCDS and ENS have been redesigned proactively.
The various systems have adapted according to the factors that influence them, such as their roles, resources, and control. The stakeholders involved in these systems have taken strategic actions to shape the systems to suit their needs: 1. VCom leveraged its partnership with the Ministry of Health to highlight the weaknesses of the existing eCDS system and offer a new solution that was more functional, efficient, and secure.
2. GDPM realized that it did not have the capability to upgrade its eCDS to comply with Circular 54, so it chose to relinquish its existing system and exercise control over the design of VCom's new eCDS, requiring that it have the same functionality as the original.
3. VAMS, which mainly works with hospitals, added a feature to its ENS that allowed the export of disease data to Excel files compatible with the new eCDS. VCom intends to connect the systems through standardized web services in the future. 4. iNGOs, which operate as part of a treaty between Vietnam and the Global Health Security Initiative, has the freedom to choose where to intervene and when to develop. Initially, it will utilize data from both the eCDS and the infectious disease data warehouse and collaborate with VAMS to upgrade ENS to streamline data collection and improve data quality. Table 3 summarizes the relationships among the different software systems before and after the configuration:

| FINDINGS AND DISCUSSION
In this section, we explore the concept of hybrid vigor to gain insight into the competition between software systems. By examining the elements of hybrid vigor, we aim to understand how this strategy has been effectively utilized by various parties to fundamentally alter the outcome of the competition.

| Overlap and duplication
The investment in health information systems in developing nations often results in overlap and duplication, typically blamed on a lack of coordination between international donors and local governments (Sahay et al., 2009). However, the data from the cases studied challenges this perception. The motivations and interactions of stakeholders have shaped the development of four seemingly overlapping systems for infectious disease reporting, creating an ecosystem that serves the need for disease surveillance in Vietnam. • Retrieve case-based and aggregate data from eCDS and ENS for dashboard and business intelligence The GDPM's eCDS was the first effort aimed at providing a comprehensive solution for infectious disease reporting and monitoring, with substantial investment and support from both donors and authorities. However, during the 2014 measles outbreak, the eCDS was unable to provide timely and accurate data for epidemic control due to its delay in implementation. Additionally, the eCDS's inflexible design prevented the adjustment of data collection frequency from weekly to daily, and lacked crucial data elements such as the number of inpatients requiring respirators and the ability to track inpatient progress with complications. This led to the need for a system that could collect measles-related data from inpatient hospitals. The VAMS ENS quickly filled the gap by providing the missing functionalities, as described by Nielsen and Saebø (2016) as "charting." The ENS was not created from scratch but instead was an extension of VAMS' larger infrastructure used to manage various activities such as medical licensing, hospital quality, and patient complaints. Although both eCDS and ENS had some overlapping functionality and collected data, they also each offered unique features and data that complemented one another.
The introduction of a new eCDS system developed by VCom with the goal of replacing the previous eCDS system is an example that demonstrates overlap and duplication are not always a result of poor coordination. The new eCDS was a well-planned intervention initiated by a strategic agreement between the Ministry of Health and VCom, followed by a series of situation analyses and recommendations related to the existing information infrastructure. Although the new eCDS offered identical functionality as the old system, VCom incorporated innovative features such as automatic data import from hospital electronic medical records to attract users and win over stakeholders.
The data warehouse for communicable diseases coordinated by international non-governmental organizations (iNGOs) is another example of the caution taken by international donors in approaching health information systems in developing countries like Vietnam. The iNGOs conducted a comprehensive situation analysis, aware of the overlapping systems used for disease reporting and monitoring. With their extensive experience in fragmented and uncoordinated efforts in many developing countries, they chose to intervene in the weakest part of the information infrastructure, data visualization and dashboard, in order to avoid further fragmentation. The intervention was planned in piecemeal phases with specific goals over a period of 5 years.

| Direct competition and mutual encroaching
The introduction of four separate health information systems in Vietnam was indicative of the varied interests and goals of multiple stakeholders.
Despite offering similar features and the potential to substitute one another, competition between the systems in the preventive health sector, where the Ministry acts as the sole client, resulted in the exclusion of one system. Unlike a free market where multiple similar products or services can coexist and be used by a variety of customers, the presence of multiple overlapping systems in this setting was not feasible.
The mutual exclusion between the old and new eCDS was a result of various social, political, and technical factors. Politically, the decision to replace the old eCDS with the new version was backed by the support of the top management at GDPM and MoH. Technically, the VCom team highlighted the need for a complete overhaul of the old eCDS and presented the new version as a solution. The development of the eCDS by GDPM and the ENS by VAMS also contributed to the overlap and competition between the systems. Although they were built to address different issues and served different user groups, they both offered similar functionalities and had the potential to be extended for use by both hospitals and preventive health centers. Initially, the iNGO focused on building a data warehouse and dashboard to visualize communicable disease data, but once fully functional, it could easily evolve to include data collection as well.

| Unpacking the dynamics of hybrid vigor
The case study reveals that the situation involves more than just competition and substitution. It is rather a tale of multiple stakeholders engaging in negotiations and making strategic adjustments in response to changes brought about by others.

| The politics of negotiation of functional roles between the different systems
Our empirical data showed that although the different stakeholders were competing for functional roles, they also pursued strategies of collaboration, negotiation, and mutual dependencies. Each stakeholder was dependent on at least one other stakeholder, whose interests in turn were dependent on another. For example, GDPM required support from VCom to meet the requirements of Circular 54. VCom, in turn, relied on VAMS for complete and timely data, while VAMS needed help to improve the ENS. Furthermore, iNGOs were dependent on GDPM for data to run the EOC's data warehouse. The interwoven dependencies among the stakeholders are depicted in Figure 3.
In the complex network of socially and technically heterogeneous actors with diverse interests as seen in this case, the interdependency among stakeholders plays an important role in stabilizing and sustaining the whole network. If one system could not provide any useful functionality to the network, it would probably be excluded quickly by other software components taking their functional role.
This functional interdependency is crucial for negotiations between stakeholders to take place. A change in strategy of one actor thus will affect other actors, and thus their strategies need to be adjusted accordingly. For example, when the VCom introduced functionality to import case-based data from Excel files in its new eCDS, VAMS provided corresponding functionality to export data to Excel files that can be used in the new eCDS. Second, when iNGOs came to know that eCDS and ENS are in place and function well as communicable disease data collection tools, it has changed the approach to focus on data visualization. The two examples clearly demonstrated that the reconfiguration strategy of an actor heavily relied on the dynamics of others' strategies and vice versa. For example, if VAMS rejected to participate in the functional boundary delineated by other actors, the eCDS might encroach upon the ENS and the ENS would be gradually replaced.
Certainly, the balance of mutual interests as analyzed in our case would not last forever. It is a result of a complex and ongoing process involving a myriad of negotiations and interests among stakeholders (Callon, 1984). Thus, what is presented in Figure 3 is merely a snapshot of an everevolving network. Future actions and new actors will trigger changes that lead to the formation of newer version of the same network or completely new networks.

| The power of government agencies in shaping functional roles
In our study, both GDPM and VAMS are governmental actors with the power to shape policies and impact the favoritism of certain systems.
However, the extent of their influence is contingent on their respective functional domains. For instance, GDPM holds significant control over the formulation of policies and the development and implementation of information infrastructure relating to infectious diseases in Vietnam.
Meanwhile, according to the division of responsibilities within the Ministry of Health, VAMS is accountable for all activities related to the administration of medical services across hospitals and clinics in the country. This level of authority empowers VAMS to deploy ENS in response to measles outbreaks without waiting for GDPM to act. Given that the majority of infectious disease diagnoses are performed at hospitals and clinics, VAMS has sole access to nationwide and near-real-time infectious disease data.
Despite being unable to control the functional architecting, the two non-governmental stakeholders VCom and iNGOs had their own strategy to actively respond to the functional plan that had been outlined by VAMS and GDPM. iNGOs gained their legitimacy through the Global Health Security Initiative and were assigned to work with local stakeholders to actualize an infectious data visualization and dashboard system as part of the US aid package aimed at strengthening Vietnam's ability to detect, prevent, and control infectious diseases. The focus of iNGOs shifted to data analysis as the data entry was being handled by other systems, a choice heavily shaped by the powerful government actors. Meanwhile, VCom secured necessary political support and took advantage of the issuance of Circular 54 to re-define its functional role. The circular introduced changes to the templates, structure, and flows of reports used for communicable diseases, which required an update of the eCDS. This gave legitimacy to VCom's plan to replace the eCDS with a newly developed system.
In summary, government agencies play a crucial role in shaping functional boundaries between systems through legal means such as circulars, decrees, and decisions. In the case being discussed, Circular 54 played a significant role in three ways: 1. It initiated a functional reconfiguration process by putting pressure on all involved actors, forcing them to change their strategies to maintain their roles. Prior to the circular, both the ENS and eCDS existed as separate systems without any interaction.
2. It redefined the functional boundaries between systems by requiring hospitals to use the eCDS for data reporting to health centers, posing a threat of exclusion to the ENS. Before the issuance of the circular, there was a clear division of functions between the eCDS and ENS, with the eCDS being used in health centers and the ENS in hospitals. However, Circular 54 redefined this boundary by requiring the eCDS to encroach on the ENS.
3. It favored one system over others by promoting the eCDS, which was backed by GDPM, thereby undermining the role of other systems. This led to the stakeholders cleverly acting within the predefined functional boundaries to protect and ensure their interests.

| The co-configuration of individual systems and the re-configuration of the resulting ecosystem
As previously discussed, direct competition poses a threat of exclusion. However, it is suggested that threat and opportunity often go hand in hand, and higher risk usually leads to higher potential rewards. The high number of direct competitors in the sphere of communicable diseases in Vietnam can be seen as an opportunity to not only utilize the strengths of each individual system but also leverage the power of collaboration to improve the information infrastructure.
There are several possible outcomes of the evolution of the information infrastructure in competitive environments. For instance, in our case study, the mutual exclusion scenario could occur if one of the four systems encroaches on the functional roles and replaces all others. On the other hand, the hybrid vigor scenario occurs when the resulting system inherits some or all of the strengths from each individual system, resulting in better functionality compared to any individual system. This superior quality leads to more adoption and support.
Functional co-configuration, which is an indirect result of the negotiation process between the systems' proponents, helps individual systems neutralize tensions and reduce the risk of exclusion. This leads to the transformation of the whole ecosystem into a better system that provides more useful and diverse functionality.

| What is exactly hybrid vigor strategy?
The interplay between different software components is also a topic in the software ecosystem literature. A shared definition is missing (Manikas & Hansen, 2013), and the common approach is to view software ecosystems as complex environments in which multiple software components exist and interact with a software platform (the host). The platform and the components on top of it are interwoven in a web of symbiotic relationships, with the components relying on the platform to function and become useful, and the platform needing components to offer value to users and grow. The platform plays a critical role in the existence of its components, while other components can be easily replaced and substituted. A systematic literature review by Manikas and Hansen (2013) identified three innate attributes of a software ecosystem, one of which is the existence of a central and common software, referring to the platform that all other modules or components are based on to function.
However, from our long-term engagement with Health Information Systems in developing countries, we have observed many cases in which the interplay of multiple software systems takes place in a considerably different way. In our case, there is not one central or common software element. Therefore, the Hybrid Vigor strategy is the answer to the research question we previously posed, which we will elaborate on in the coming paragraphs.
This paper delves into the realm of software ecosystems where similar systems compete to meet functional needs. These systems often have a high degree of identical and overlapping functionality, and in order to survive, each system must continuously align and realign itself with other systems. During this process, the system may reconfigure itself by either reducing or expanding its functionality. The legal control exercised by government agencies serves as a central driving force behind the evolution of the software ecosystem, much like platforms in other ecosystems.
The ecosystem can be impacted by various factors such as changes in the environment (new policies or requirements) or the introduction of new actors, which trigger complex negotiating processes. These changes can result in shifts that favor some actors and disadvantage others. Nielsen and Saebø (2016) had identified three strategies that actors can use to reconfigure their functional roles in the ecosystem: connecting, charting, and encroaching.
"Connecting" involves the integration of systems with complementary functionality. "Charting" means extending a system's functionality to meet unmet needs, while operating freely without competition. On the other hand, "encroaching" involves introducing competing functionality to challenge and overtake other systems' functional roles.
The three strategies discussed by Nielsen and Saebø (2016)  It is clear that the three concepts of connecting, charting, and encroaching by Nielsen and Saebø (2016) did not provide sufficient analytical power to fully understand our case. Even the concept of asymmetric integration by Sahay et al. (2009) did not offer sufficient insight into the dynamics of overlapping and substitution between directly competing systems, as both systems discussed in their study offer relatively separate functionalities, such as collection of aggregation data and detailed name-based data. We therefore argue that the use of the new concept of hybrid vigor is necessary in analyzing the emerging empirical data.
Drawing inspiration from biology and genetics, we propose a fourth strategy for the reconfiguration of software in a wider ecosystem: hybrid vigor. In biology, hybrid vigor refers to the superior qualities that emerge from the crossbreeding of different plants or animals. Analogously, in the context of software ecosystems, hybrid vigor refers to the superior functional attributes that emerge from the combination of multiple software components. Unlike the connecting strategy discussed by Nielsen and Saebø (2016), which focused on data sharing and complementary functionality, our concept of hybrid vigor emphasizes the strengthening of the functional attributes of the ecosystem as a whole. This concept is based on the assumption that the presence of multiple software components represents a strength and a source of innovation (Sahay et al., 2009).
We will now delve into a deeper analysis of the hybrid vigor concept to fully understand its potential in studying the development and evolution of information infrastructures and software ecosystems. We believe the concept provides analytical power in comprehending and potentially predicting the dynamics of complex systems such as these. We will focus on three main aspects of the concept: cross-breeding of software genres, the unique qualities of vigor in the context of information infrastructures, and the generative mechanisms of hybrid vigor in these systems.
Our goal is to make a solid case for the use of this concept.

| Cross-breeding of different software genres
As previously mentioned, hybrid vigor in biology refers to cross-breeding of plants or animals within the same species or genus. The same concept can be applied to information infrastructures and software ecosystems, where it emphasizes the significance of combining closely related software components. For example, the software components in our case studies all provide functionality for managing and monitoring infectious diseases. The level of relatedness required for hybrid vigor is still unclear. Additionally, the software components must exhibit diversity in design, functionality, architecture, technology, and stakeholders. As summarized in Table 4, the four software components in our cases are distinct.
The essence of hybrid vigor arises from cross-breeding different breeds or genera of a species. Mapping these concepts to information infrastructures can be challenging. We propose that breeds and genera in information infrastructures can be viewed as software components that do not provide identical functionality and/or use differing technologies. This diversity is believed to be the source of generativity in resulting information infrastructures. We will revisit this idea when examining the underlying mechanisms of hybrid vigor in information infrastructures.

| Specificalities of vigor in the context of information infrastructures
In biology, hybrid vigor is defined as improved or enhanced function in a hybrid offspring for traits such as stature, biomass, fertility, etc. In the context of information infrastructures, what constitutes a quality of a hybrid infrastructure? We can adopt a list of key software system qualities commonly used in software engineering and information infrastructure literature and discuss them based on our empirical data. The six qualities are functionality, user experience (UX) and user interface (UI), scalability, resilience, evolution (Sahay et al., 2009), and innovation (Grisot et al., 2014).

More complete set of functionalities
Although both the new eCDS and ENS offer basic reporting functionality, they lack the power to serve as an intelligent analytic tool that endusers can use to quickly identify and predict new outbreaks and the likelihood of a pandemic. CDDW, developed by a team of specialists with expertise in disease and epidemic modeling, provides end-users with many functionalities not available in the new eCDS and ENS, such as the ability to analyze data over different time periods and trends and to quickly detect and predict disease outbreaks. UI design is sometimes controversial, but we believe that combining different software components results in a better user experience. In the case of hospitals, users are already familiar with the flow and UI of ENS, so introducing another software would entail additional learning and negatively impact the user experience. Furthermore, using two separate software components to enter the same data would require double entry and significantly detract from the overall user experience. Therefore, combining software leads to a better user experience.
Better scalability Sahay and Walsham (2006) argued that scalability in information infrastructures should not be viewed as merely a technical issue, but should involve other aspects such as institutional, political, and human resources. From this perspective, the information infrastructure in the Vietnamese health-care system exhibits improved scalability traits. In the Vietnamese health-care system, there are two relatively separate streams: public health centers and hospitals. While the eCDS (both old and new) is strong in supporting public health centers, it struggles to work with hospitals, where nearly all use some form of electronic patient records. The introduction of eCDS into hospitals would result in extra burden for hospital staff and cause fragmentation problems. By combining eCDS and ENS, the resulting information infrastructure would provide full coverage of data collection and be already scaled to both public health centers and hospitals.

Higher resilience
Modern software architecture best practices emphasize the need for micro-service or modular design, even though it may result in extra communication and integration. The overall system will have increased resilience in terms of partial failure. Deploying the three software components by different stakeholders on different hardware infrastructure can maintain higher resilience of the information infrastructure against failures such as hardware failure, power outages, and network interruption.

Better innovation and evolution
For a long time, researchers in the field of information infrastructure have been searching for strategies and mechanisms to enhance innovation and allow for the evolution of information infrastructures (as cited in Grisot et al., 2014;Henfridsson & Bygstad, 2013). Our empirical data indicate that crossbreeding software components can create many possibilities for innovation. For instance, using ENS software to gather data directly from hospitals is an effective way to guarantee data quality and breadth. Hospitals also have the necessary resources to diagnose and verify positive cases. Incorporating a data collection software tool in hospitals is a novel approach to improving data quality.
Another example is using an intelligence tool for data analysis along with a software component that has the capability to identify and forecast epidemics (CDDW).

| Underlying generativity mechanisms of hybrid vigor in information infrastructures
The mechanism of hybrid vigor remains controversial in its original context. However, some biologists agree to some extent that genetic mixing is the driving force behind hybrid vigor. In the realm of information infrastructure, we can draw parallels based on our empirical data. 5.3.5 | The resulting information infrastructure for epidemic control in Vietnam and who were the followers of hybrid vigor strategy?
The current status of the infectious disease reporting system in Vietnam can be seen as the outcome of a hybrid vigor strategy with components successfully reconfigured to collectively (hybrid) offer superior (vigor) functionality: 1. The old eCDS has acted as a functional reference for the new eCDS, despite being replaced.
2. The ENS provides clinical data of infectious cases to the new eCDS electronically, playing its role in the ecosystem.
3. The CDDW relies on data from eCDS and ENS for data visualization and analysis functionality.
As the process of (re-)configuring is contingent, the evolution of software ecosystems may take many different trajectories. What we have seen is that software components which gains strong political favor may get the required backing to expand to encroach upon other components and the functionality they offer. Other components are thus marginalized and potentially excluded. While this new configuration can offer the needed functionality, it will at the same time undermine further innovation and new cycles of reconfiguration based on the hybrid vigor strategy.
Actors pursuing an encroaching strategy with the result of excluding other systems may also experience political repercussion as there are vested interests in any software component. Depending on the political climate, the power of the involved actors and endogenous configurability of components, hybrid vigor can be achieved on different configurations.
For the strategy of hybrid vigor to be effective, it must be consciously embraced by key stakeholders. In the context of our discussion, two actors, VAMS and iNGOs, have been noted as having intentionally pursued this strategy. As they have a comparatively minor and legitimate role in the arena of infectious diseases data compared to GDPM and VCom, the adoption of hybrid vigor is vital to their systems' survival. These actors play a crucial role as "system builders." This argument contributes to the ongoing debate on the role of centralized control by government agencies in the development of information infrastructure. While the role of government agencies is important, it does not mean that the roles of other actors should be ignored or dismissed. The outcome of a situation can change, even if it initially appears to be predetermined. Future research can examine scenarios not discussed in this case, such as what the resulting infrastructure might look like if (a) government agencies pursue hybrid vigor, or (b) none of the actors follow it.

| CONCLUSION
The findings and discussions in this paper challenge the widespread but oversimplified view of information infrastructure that largely disregards the complexity of the development of its software components, including their overlap, duplication, and potential substitution. The study provides a closer look at information infrastructures as software ecosystems and how they evolve. It highlights the interplay and dynamics of directly competing components as they struggle to survive throughout the evolution process. Theoretically, the study introduces the concept of hybrid vigor as a strategy for functional architecting, offering new insights into the evolutionary trajectories of information infrastructures as competitive environments, or battlegrounds. The study identifies three dimensions of hybrid vigor (political negotiation, power of government agencies, and functional reconfiguration) to characterize the strategy and offer a better tool for analyzing threats and opportunities posed by crossbreeding directly competing components. The study also delves into the core of the hybrid vigor concept by examining various aspects of the concept (cross-breeding, specificities of vigor, and mechanisms of hybrid vigor) and linking it back to its original context in biology. The study answers the research question that hybrid vigor is the strategy to follow when dealing with situations when rival software components pose threats of encroaching on each other.
The findings have several implications for both theory and practice. The tension of direct competition is inevitable in many cases, and in these situations, it is important to adopt flexible and elastic strategies that can leverage and combine the strengths of separate but competing components, instead of encouraging a fight until only one component remains. The resulting ecosystem, as a result of the hybrid vigor strategy, will support innovation and become more responsive to any contingencies or changes in the environment. These findings could extend beyond the health-care domain and be useful in analyzing similar phenomena in other areas such as business and social networking.

ACKNOWLEDGMENT
Open access publishing facilitated by RMIT University, as part of the Wiley -RMIT University agreement via the Council of Australian University Librarians.

DATA AVAILABILITY STATEMENT
Research data are not shared.

ORCID
Thanh Ngoc Nguyen https://orcid.org/0000-0003-3052-1571 Petter Nielsen https://orcid.org/0000-0003-3723-6976 7. still think the figure is a bit confusing. And figures should clarify, not confuse. My current concern is the arrowsthey are illustrating so many different things: How different systems are influencing each other, replacing each other and exchange of data. Perhaps the relationship between the new and the old eCDS should not be there?
8. I still find this a bit odd. Is there such a strategy, or is each of the systems just fighting for their own position? I am not convinced.
9. I think the interesting part in this case is the (1) central role of the government/policies and (2) the co-existence of systems with overlapping functionality.
10. I do not really see the relevance of thisdo you have much empirical data on the role of the non-functional?
11. I think it is better to position this paper related to the political nature of the process, the politics of how systems influence each other and continue to exist.
12. I have suggested a slight rewrite.
13. What the challenge is then is to make the hybrid vigor concept to capture the three dimensions of politics, role of governmental agencies and the functional configuration of the ecosystem. I think the hybrid vigor concept so far only have discussed the last dimension so far.
14. I am still struggling with the hybrid vigor concept. Is it most sensible to discuss this as an end-state or a as a forth strategy? I have attempted to describe it as a forth strategy. Dr Petter Nielsen is a Professor at Department of Informatics, University of Oslo, Norway. His research interest is related to large-scale and complex information systems (such as information infrastructures, digital platforms, and digital global public goods)-how they are evolving, how we influence them, how we can govern them, the role of their architecture and how they facilitate innovation. He has empirically studied this related to content services for mobile phones, e-commerce as well as health information systems.