Reviewing the Health of Software Ecosystems { A Conceptual Framework Proposal

Abstract

The health of a software ecosystem is an indication of how well the ecosystem is functioning. The measurement of health can point to issues that need to be addressed in the ecosystem and areas for the ecosystem to improve. However, the software ecosystem eld lacks an applicable way to measure and evaluate health. In this work, we review the literature related to the concept of software ecosystem health and the literature that inspired the software ecosystem health literature (a total of 23 papers) and (i) identify that the main source of inspiration is the health of business ecosystems while also in uenced by theories from natural ecosystems and open source, (ii) identify two areas where software ecosystems di er from business and natural ecosystems, and (iii) propose a conceptual framework for de ning and measuring the health of software ecosystems.

Keywords

• software ecosystems
• ecosystem health
• software ecosystem health framework
• software ecosystem health measurement

1 Introduction

The notion of software ecosystems (SECOs) is gaining popularity as a means of expanding development, better positioning in the market, or increasing revenues. There is a number of de nitions of SECOs in the literature [ 1, 2, 3, 4 ]. In this work we de ne a software ecosystem as \the interaction of a set of actors on top of a common technological platform that results in a number of software solutions or services. Each actor is motivated by a set of interests or business models and connected to the rest of the actors and the ecosystem as a whole with symbiotic relationships, while, the technological platform is structured in a way that allows the involvement and contribution of the di erent actors" [ 5 ]. Today, software ecosystems come with a wide variability of characteristics: platform structure, actor participation, ecosystem orchestration, and revenue models, to name a few. This makes the establishment of methods for measuring and evaluating the activity of the ecosystem challenging. The SECO literature refers to the concept of \health" of an ecosystem as a way to monitor ecosystem activity, identify and predict areas for improvement, and evaluate changes in the ecosystem. However, the measurement of the SECO health is not yet fully achieved.

We tentatively de ne the health of a software ecosystem as the ability of the ecosystem to endure and remain variable and productive over time. In this work, we aim to get closer to SECO health measurement by reviewing the literature that is elaborating on SECO health. In doing so, we identify that the SECO health literature is borrowing de nitions and measurement of health from other elds and expand our literature review focus to include additional ecosystem health elds (explained in section 2). We review the wider ecosystem health literature body and report the health de nitions and measurements (section 3). We identify two main di erences between SECOs and business and natural ecosystems and, based on previous work, we propose a conceptual framework for de ning and measuring the health of SECOs (section 4). Finally, in section 5 we discuss threats to validity and future work and conclude in section 6.

2 De ning the Health Literature Body

The method used for de ning the literature body consisted of the following steps: (i) De ning the SECO health literature. To de ne the literature related to the SECO health, we used as input the papers identi ed in our recent systematic literature review [ 5 ]. In [ 5 ], we identi ed a number of papers referring to the concept of ecosystem health. The papers have a wide variability on the level of detail they provide on the ecosystem health ranging from mere reference to the concept (e.g., [ 6, 7, 8 ]) to papers in which health forms part of the main focus (e.g., [ 9, 10 ]). (ii) De ning wider ecosystem health literature. While examining the SECO health literature, we noticed that the de nition and analysis of health is borrowed from other types of ecosystems not covered by the SECO health literature . Using the \snowballing technique" [ 11 ], we followed the references of the SECO health literature and evaluated whether these are related to the health of an ecosystem 1. The criteria for accepting a paper in the literature was that it would (a) de ne the health or sustainability of an ecosystem or (b) elaborate on ways of measuring health. 1 We also followed references of the selected references that appeared relevant, resulting in a number of papers ([ 12, 13 ]) 2 Paper [ 27 ] is de ning the eld of \IT ecosystems", though, as a BECO with IT products.

Type SECO BECO OSS Total:

[ 28 ] Natural ecosystems [ 12 ] [ 13 ] Paper

Source [14, 10, 15, 9, 18,6,119,7,270,, 1281,, Health literature from [ 5 ] 6] [ 14, 9, 21, 10, 15, 23, 24 ] [ 14, 7, 10, 16, 6, 23 ] [ 14, 10, 9, 16, 18 ] [ 18, 8, 23 ] [ 20 ] [ 9 ] in uenced SECO health. In this work, we have not looked at the health de nition and measurement in the other elds outside the references of the SECO health literature and, thus, do not claim that these papers are representative of each eld.

3 Ecosystem Health

Following the separation of ecosystem elds in Table 1, we list and discuss the papers per ecosystem that have in uenced the SECO health literature.

3.1 Natural Ecosystems

The eld of (natural) ecosystems inspired the rest of the ecosystem elds examined here (BECO and SECO) and it is the eld where the concept of ecosystem health was initially formulated. Costanza [ 12 ], de nes a healthy ecosystem as \being `stable and sustainable'; maintaining its organization and autonomy over time and its resilience to stress". In addition, Rapport et al. [ 28 ], referring to a collection of papers in the literature, de ne three indicators for health of an ecosystem: Vigor that indicates how active or productive an ecosystem is, Organization that indicates the variability of species, and Resilience that indicates the ability of the ecosystem to \maintain structure and function in the presence of stress". The characterization of an ecosystem in terms of structure and function is also discussed by Schae er et al. in [ 13 ]. They parallelize ecosystem health with human health and de ne it as the \absence of disease". They identify structure as \numbers of kinds of organisms, biomass etc." and function as \activity, production, decomposition etc.". These are seen as measures used to de ne the ecosystem health. Furthermore, Schae er et al., referring to the literature, list four ways that structure and function may be connected: (a) tightly connected, where neither can change without the change of the other, (b) structure changes does not a ect function, (c) function changes do not a ect structure, and, (d) structure and function appear unconnected.

3.2 Business Ecosystems

BECO health is the area that has inspired most of the SECO health literature. In the BECO literature, the concept of health is mainly de ned as the ability of a BECO to provide \durably growing opportunities for its members and for those who depend on it" [ 26 ]3. Iansiti and Levien [ 25, 26, 22 ] and Iansiti and Richards [ 27 ] de ne the health of a business ecosystem using three measures: Productivity. Inspired by natural ecosystems' ability to create energy from input sources (e.g., sunlight or mineral nutrients), BECO productivity is the ability of an ecosystem to \convert raw materials of innovation into lowered costs and new products and functions" [ 26 ]. Productivity in BECOs can be measured by means of (a) total factor productivity, (b) productivity improvement over time, and (c) delivery of innovations, the ability of the ecosystem to adapt and deliver to its members new technologies, ideas, or process.

Robustness. The ability of the ecosystem to sustain shocks, perturbations, and disruptions. Robustness is measured in terms of (a) survival rates, the survival of actors over time, (b) persistence of ecosystem structure, the extent to which actor relationships are kept unchanged, (c) predictability, the extent to which even if shocks alter the relationships of actors, a main core of the ecosystem remains solid, (d) limited obsolescence, whether the ecosystem has a limited invested technology or components that becomes obsolete after a shock, and (e) continuity of use experience and use cases, the extent to which products gradually evolve in response to new technologies rather than changing abruptly.

Niche Creation or Innovation. The ability of the ecosystem to increase meaningful actor diversity over time. Niche creation is measured in terms of (i) growth in company variety and (ii) growth in product and technical variety (value creation) that measures the increase in value the growth brings.

Iansiti and Levien and Iansiti and Richards also propose three ecosystem actor roles, inspired by natural ecosystems, that a ect the health of a BECO: Keystone. Is an actor that normally occupies or creates highly connected hubs of actors and promotes the health of the ecosystem by providing value to 3 Similar de nitions appear in [ 22, 27 ] the surrounding actors. Keystones promote the health of the ecosystem by increasing the variability, provide value to the connected actors and thus increase productivity, and increase robustness by protecting connected actors from external shocks.

Dominators. Are the actors that control the \value capture and value creation" [ 22 ] of the ecosystem. They tend to expand by taking over the functions of other actors thus eventually eliminating the actors. Dominators are harmful for the health of an ecosystem as they reduce diversity Niche (players or rms). Usually form the main volume of the ecosystem actors drawing value from the keystones. A niche player aims to separate from the other niche players by developing special functions.

Typically, a keystone provides value to a number of actors that can be either niche players trying to develop or dominators trying to dominate the functions of the surrounding actors. The roles of the BECO actors are also examined by Iyer and Lee [ 24 ]. They classify the actors in an ecosystem in (a) hubs, (b) brokers that connect two sets of actors, and, (c) bridges that are essential for the connectedness of the ecosystem. A hub can demonstrate keystone, dominator, or niche player characteristics.

Hartigh et al. [ 23 ] use the work of Iansiti and Levien and Iansiti and Richards (referred to as \Iansiti" hereafter) to measure the health of the Dutch IT industry. They de ne BECO health using two long-term parameters: the nancial well-being and strength of the network and break down the health in two components: partner health and network health. Partner health evaluates the health of each individual actor of the ecosystem. A healthy ecosystem is composed of productive actors contributing to the productivity of the ecosystem while unproductive actors will have di culty surviving. The survival of the actors is analogous to the Iansiti robustness measure. Network health is measured in terms of actor connectivity. Highly connected actors contribute to the robustness of the ecosystem as the actors are not easily a ected by external shocks. In addition, a healthy ecosystem contains clusters of di erent nature, thus increasing the possibility of niche creation. 3.3 OSS Wahyudin et al. [ 29 ] study the concept of health in OSS projects. They de ne the health of an OSS project as \survivability", the ability of the project to survive throughout time. An OSS project is healthy and survives if the software produced by the project is used by a number of users and maintained by a number of developers. They identify three measures that a ect the health of an open source project: The developer community liveliness. The project should attract new developers and keep the existing by boosting their motivation. Wahyudin et al. break down an OSS developer's motivation in intellectual stimulation, skill enhancement, and access to source code and user needs. The user community liveliness. The users of OSS software play an active role in the evolution of the project by reporting bugs and requesting new features. A large, active user community indicates that the software produced is usable and of good quality.

The product quality. A product that is competitive with commercial products in use and quality will attract users and developers, increase the activity in the project, and therefore enhance survivability.

3.4 Software Ecosystems

In the eld of SECO, Berk et al. [ 9 ] propose SECO-SAM, a model for the assessment of a SECO strategy based on SECO health. In their model, they make an analogy between the health of an ecosystem and human health and propose that SECO health is in uenced by the biology of the ecosystem, the lifestyle, the environment, and the intervention of healthcare organizations while they measure the SECO health adopting the Iansiti productivity, robustness, and niche creation (PRN) measures. Jansen et al. [ 10 ] elaborate on a three-level model of SECOs, published in [ 7 ], consisting of the organization scope level, SECO level, and software supply network level. They de ne SECO health as a characteristic of the software supply network level using the Iansiti PRN measures. Additionally, they propose the application of the Hartigh et al. [ 23 ] measures for de ning the health at the SECO level. Angeren et al. [ 14 ] show that SECO robustness of the Iansiti PRN measures is an important factor for vendors that choose to depend on a SECO.

In OSS, McGregor [ 20 ] translates the Iansiti PRN measures to measures that can be applied to open source projects, while Kilamo et al. [ 18 ] propose a framework for going from a proprietary to a Free/Libre/Open Source Software (FLOSS) SECO. One of the framework activities is setting up a \community watchdog" that assesses the community, the software, and \how well the objectives of the company are met". The watchdog indirectly assesses the health while they provide a number of measures to be applied in FLOSS SECOs.

Looking at the SECO health literature, we note that the main source of inspiration is BECO health when trying to de ne and measure SECO health or health-related parameters (e.g., keystone-dominator strategies) with 11 out of the 13 papers referring to at least one of the Iansiti authored papers [ 22, 25, 26, 27 ]. Although the health of a BECO is very similar to the SECO health, we identify a number of di erences between the two. In the next section, we explain the di erences and build on top of the existing literature to de ne a framework for SECO health.

4 A SECO Health Proposal

When analyzing the health of SECOs, we identify that similarly to BECOs and natural ecosystems, the set of actors, their activity and the network they form is an indication of the level of prosperity and sustainability of the ecosystem. However, one main di erence of SECOs from BECOs and natural ecosystems is in the nature of the products of the actors and, eventually, of the whole ecosystem. The BECO approach explained in the previous section is aligned with the natural ecosystem approach of actors and products, where the products of the ecosystem (i.e. energy) are represented by the actors (i.e., species) by enclosing energy in the energy ow between the species. In other words, a herbivorous species eats a plant and is eaten by a carnivorous species. This herbivorous species is both an actor and a product in the ecosystem and changes in the health of this species (e.g., number decrease) a ects the energy enclosed (product) by this species and, thus, the carnivorous species.

In SECOs, the actors are di erentiated from the products. The main product of the actors is software, either as a common software/technological platform, as software components, or services based on software components. The symbiosis of this software can in uence the health of a SECO. The in uence that the software components have on the SECO health is independent of the actor health. An example would be an actor that creates a software component that enhances the component interoperability and increases the use of the platform and thus contributing to the SECO health. At the same time, this actor might not have a successful revenue model for this software component and end up loosing a big part of the invested e ort. The actor will have a negative in uence on the SECO health because of low productivity and possibly robustness, while the software component will have a positive in uence.

One additional di erence of SECOs to BECO/natural ecosystems is that in SECOs there is an entity organizing and managing the ecosystem, the orchestrator. The orchestrator, whether a for-pro t organization or an OSS community, is typically managing the ecosystem by running the platform and creating rules and processes for actors and software. The orchestration of the SECO thus has a signi cant e ect on the health of the ecosystem.

The proposed SECO health framework can be seen in Figure 1. We depict three main components that a ect the SECO health: (i) the actors, (ii) the software and (iii) the orchestration. In (i), we separate between the individual health of an actor and the health of the network of actors and similarly in (ii) between the individual component health, the ecosystem platform health and the software network health.

4.1 Individual Actor Health

The health of the individual actors in uences the overall health of the ecosystem. The actor health can be measured in similar terms to a BECO actor. The actor's productivity and robustness in uence the ecosystem. The active participation and engagement of actors brings value to the ecosystem, while the actor's robustness increases the probability that the actor exists and remains involved in the ecosystem activity in the future. If the SECO is a proprietary ecosystem or consists of for-pro t organisations, the partner health measures of Hartigh et al. [ 23 ] can be directly applied. If in the OSS domain, the actor health can be

Actors

Individual Actor Health

Actor Network Health

SECO Health

Software

Platform

Health

Software Component Health

Software Network Health

Orchestration

assessed in a way similar to Wahyudin et al [ 29 ]: measuring the actor activity in the ecosystem (commits, mailing list activity etc.). In that case, an indication of actor robustness is the active participation in the ecosystem over a long period of time. An actor being an active participant in the ecosystem for a long period of time has lower probability of dropping out of the ecosystem than an actor that recently started contributing to the ecosystem.

4.2 Actor Network Health

The network of actors and their interaction plays an important role in the SECO health. The PRN measurements are applicable here, so is the network health perspective of Hartigh et al. [ 23 ]. Additionally, the individual actor health may be weighted according to the role of the actor in the network. A keystone with low productivity or robustness will have greater e ect in the ecosystem than a niche player with low productivity or robustness.

4.3 Software Component Health

The health of a software component can be measured in terms of, among others, (i) reliability, (ii) availability, (iii) modi ability and prevention of ripple e ects, and (iv) interoperability, the ability to interact with, to the extent applicable, the platform and other components. In SECOs, the software components are, in most cases, also the products of the ecosystem. The health of such a software component is also in uenced by the relative demand and product quality, e.g., how popular is the product and how it is performing in comparison to possible alternatives. This demand is also a ected by whether the product is internal, i.e., products intended for use mainly by the ecosystem actors, e.g., the technological platform or external, i.e., products consumed externally to the ecosystem.

4.4 Platform Health

The health characterization of the software components above can be applied to the technological platform of a SECO, since it is a software component itself. However, the technological platform, might have an additional role: depending on how the SECO is organized and managed, the platform re ects possible orchestration actions (rules, processes, or management decisions). The measurement of the platform health should not re ect how the orchestration a ects the SECO health (as this is re ected in the orchestration in uence on SECO health seen below), but the e ectiveness of applying the orchestration actions.

4.5 Software Network Health

The software components are connected and interacting with other components in the ecosystem forming the software network. Graph measures such as connectivity and clustering coe cient show to what extent the components interact [ 30 ]. Additionally, the categorization of the activity of hubs into keystone and dominator indicate the level of healthy interaction. Analogous to the Iansiti descriptions in the previous section, an example of keystone activity can be a component that provides interfaces to parts of its functionality for the neighboring components to consume, while in a dominator activity the component would intent to take over functionality of the neighboring components.

4.6 Orchestration In uence to Health

The orchestrator can monitor the health of the ecosystem and take measures to promote ecosystem health if necessary. This requires that the orchestrator has a good overview of the ecosystem and is consulting e ective measurements (e.g., ecosystem health). Additionally, the orchestrator can act by creating/re ning rules and processes for the actors, communicating plans to the actors (e.g., by road-mapping), organizing the ecosystem development through, e.g., release management, making changes to the platform and other software components, changing the revenue model for internal products, and controlling the actor population and motivation by modifying the model by which the actors participate in the ecosystem. The orchestration of a SECO, i.e., the actions of the orchestrator, possibly based on monitoring and evaluation, in uences SECO health.

4.7 Other In uences on SECO Health

Additionally, there might also be in uences on the SECO health that are external to the ecosystem. This kind of in uences are referred to as \(external) perturbation" in the literature [ 28, 22, 26, 27 ] and are disturbances that are outside the control of the ecosystem actors. In uences of this kind might be the establishment or rise of a competitive ecosystem or a radical technological or legal change.

5 Threats to Validity and Future Work

The wider ecosystem health literature used in this study was identi ed through the references in the SECO health literature as our focus was literature that in uenced the SECO health literature. As already mentioned, the literature on each eld (apart from SECO) is not necessarily the representative or most in uential work in the eld. Identi cation of the most in uential work and possible literature mapping of the health in each of the elds (BECO, natural ecosystem, OSS/FLOSS) might bring perspectives into the SECO health that have been overlooked. Additionally, we speculate that the in uence of the di erence elds to the SECO health is not necessary re ected in the number of papers appearing in this work. Natural ecosystem have had a greater impact on SECO health concepts, but most of it is indirect through the health of BECOs.

Moreover, the proposed conceptual framework, at this point, does not go into detail on the di erent kinds of actors. An expansion of the model would further analyze on the nature of the actors, e.g., developing companies, resellers, valueadding-resellers, and possibly include their in uence on health. Additionally, although the model included products, it did not include customers or end-users. The in uence of entities of this kind could be discussed in future work.

6 Conclusion

In this paper, we analyzed the concept of software ecosystem (SECO) health. In order to de ne SECO health and its measurement, we examined the SECO health literature, a literature body of 13 papers touching upon the concept of SECO health. We identi ed that the health research is mainly inspired by three elds: business ecosystems (BECO), natural ecosystems, and open source software, with BECO being the main source of inspiration in 11 out of the 13 SECO health papers. We reviewed the wider ecosystem health literature, consisting of 23 papers, explained how they de ne and measure the health of an ecosystem and concluded with two contributions: (i) We identify two di erences between the SECO and business and natural ecosystems: (a) they perceive products in the ecosystem di erently. BECOs and natural ecosystems perceive actors as a product per se, while in SECOs an actor produces software components or services. (b) SECOs have an orchestrator entity managing the ecosystem, something that does not appear in the BECO/natural ecosystem literature. (ii) We propose a logical framework for de ning and measuring the SECO health consisting of the health of (a) each individual actor, (b) network of actors, (c) each individual software component, (d) platform, (e) software network, and (f) orchestrator. The purpose of this study is to create a discussion on the particularities of SECO health and bring the community closer to a measurable way of de ning the health of software ecosystems.

Acknowledgements

This work has been partially funded by the Connect2Care project4. 4 http://www.partnerskabetunik.dk/projekter/connect2care.aspx

References

1. Jansen , S. , Finkelstein , A. , Brinkkemper , S.: A sense of community: A research agenda for software ecosystems . In: Software Engineering - Companion Volume, 2009 . ICSE-Companion 2009 . 31st International Conference on. (may 2009 ) 187 { 190

2. Bosch , J. , Bosch-Sijtsema , P. : From integration to composition: On the impact of software product lines, global development and ecosystems . Journal of Systems and Software 83 ( 1 ) ( 2010 ) 67 { 76

3. Messerschmitt , D. , Szyperski , C. : Software ecosystem: understanding an indispensable technology and industry . MIT Press Books 1 ( 2003 )

4. Lungu , M. , Lanza , M. , G ^rba, T., Robbes , R. : The small project observatory: Visualizing software ecosystems . Science of Computer Programming 75 ( 4 ) ( 2010 ) 264 { 275 Experimental Software and Toolkits (EST 3): A special issue of the Workshop on Academic Software Development Tools and Techniques (WASDeTT 2008 ).

5. Manikas , K. , Hansen , K.M.: Software ecosystems { A systematic literature review . Journal of Systems and Software 86 ( 5 ) ( 2013 ) 1294 { 1306

6. Mizushima , K. , Ikawa , Y. : A structure of co-creation in an open source software ecosystem: A case study of the eclipse community . In: Technology Management in the Energy Smart World (PICMET) , 2011 Proceedings of PICMET '11: . (august 2011 ) 1 { 8

7. Boucharas , V. , Jansen , S. , Brinkkemper , S. : Formalizing software ecosystem modeling . In: Proceedings of the 1st international workshop on Open component ecosystems. IWOCE '09 , New York, NY, USA, ACM ( 2009 ) 41 { 50

8. Viljainen , M. , Kauppinen , M. : Software ecosystems: A set of management practices for platform integrators in the telecom industry . In Regnell, B. , Weerd , I. , Troyer , O. , Aalst , W. , Mylopoulos , J. , Rosemann , M. , Shaw , M.J. , Szyperski , C., eds.: Software Business. Volume 80 of Lecture Notes in Business Information Processing . Springer Berlin Heidelberg ( 2011 ) 32 { 43 10.1007/978-3- 642 -21544-5 4 .

9. van den Berk, I., Jansen , S. , Luinenburg , L. : Software ecosystems: a software ecosystem strategy assessment model . In: Proceedings of the Fourth European Conference on Software Architecture: Companion Volume. ECSA '10 , New York, NY, USA, ACM ( 2010 ) 127 { 134

10. Jansen , S. , Brinkkemper , S. , Finkelstein , A. : Business network management as a survival strategy: A tale of two software ecosystems . In: First International Workshop on Software Ecosystems (IWSECO-2009) , Citeseer ( 2009 ) 34 { 48

11. Denscombe , M.: The good research guide . Open University Press ( 2010 )

12. Costanza , R.: Toward an operational de nition of ecosystem health. Ecosystem health: New goals for environmental management ( 1992 ) 239 { 256

13. Schae er, D.J., Herricks , E.E. , Kerster , H.W. : Ecosystem health: I. measuring ecosystem health . Environmental Management 12 ( 4 ) ( 1988 ) 445 { 455

14. van Angeren, J. , Blijleven , V. , Jansen , S. : Relationship intimacy in software ecosystems: a survey of the dutch software industry . In: Proceedings of the International Conference on Management of Emergent Digital EcoSystems. MEDES '11 , New York, NY, USA, ACM ( 2011 ) 68 { 75

15. dos Santos , R.P. , Werner , C.M.L.: A proposal for software ecosystem engineering . In: Third International Workshop on Software Ecosystems (IWSECO-2011) , CEUR-WS ( 2011 ) 40 { 51

16. Jansen , S. , Brinkkemper , S. , Souer , J. , Luinenburg , L. : Shades of gray: Opening up a software producing organization with the open software enterprise model . Journal of Systems and Software 85 ( 7 ) ( 2012 ) 1495 { 1510

17. dos Santos , R.P. , Werner , C. : Treating business dimension in software ecosystems . In: Proceedings of the International Conference on Management of Emergent Digital EcoSystems. MEDES '11 , New York, NY, USA, ACM ( 2011 ) 197 { 201

18. Kilamo , T. , Hammouda , I. , Mikkonen , T. , Aaltonen , T. : From proprietary to open source-"growing an open source ecosystem" . Journal of Systems and Software 85 ( 7 ) ( 2012 ) 1467 { 1478

19. Dhungana , D. , Groher , I. , Schludermann , E. , Bi , S. : Software ecosystems vs. natural ecosystems: learning from the ingenious mind of nature . In: Proceedings of the Fourth European Conference on Software Architecture: Companion Volume. ECSA '10 , New York, NY, USA, ACM ( 2010 ) 96 { 102

20. McGregor , J.D.: A method for analyzing software product line ecosystems . In: Proceedings of the Fourth European Conference on Software Architecture: Companion Volume. ECSA '10 , New York, NY, USA, ACM ( 2010 ) 73 { 80

21. van Ingen, K. , van Ommen , J. , Jansen , S. : Improving activity in communities of practice through software release management . In: Proceedings of the International Conference on Management of Emergent Digital EcoSystems. MEDES '11 , New York, NY, USA, ACM ( 2011 ) 94 { 98

22. Iansiti , M. , Levien , R.: The keystone advantage: what the new dynamics of business ecosystems mean for strategy, innovation, and sustainability . Harvard Business Press ( 2004 )

23. den Hartigh, E., Tol , M. , Visscher , W.: The health measurement of a business ecosystem . In: Proceedings of the European Network on Chaos and Complexity Research and Management Practice Meeting . ( 2006 )

24. Iyer , B. , Lee , C.H. , Venkatraman , N.: Managing in a small world ecosystem: Some lessons from the software sector . California Management Review 48 ( 3 ) ( 2006 ) 28 { 47

25. Iansiti , M. , Levien , R.: Strategy as ecology . Harvard Business Review 82 ( 3 ) ( 2004 ) 68 { 81

26. Iansiti , M. , Levien , R.: Keystones and dominators: Framing operating and technology strategy in a business ecosystem . Harvard Business School , Boston ( 2004 )

27. Iansiti , M. , Richards , G.L. : The information technology ecosystem: Structure, health, and performance . Antitrust Bull . 51 ( 2006 ) 77

28. Rapport , D. , Costanza , R. , McMichael , A. : Assessing ecosystem health . Trends in Ecology & Evolution 13 ( 10 ) ( 1998 ) 397 { 402

29. Wahyudin , D. , Mustofa , K. , Schatten , A. , Bi , S. , Tjoa , A.M.: Monitoring the health status of open source web-engineering projects . International Journal of Web Information Systems 3 ( 1 /2) ( 2007 ) 116 { 139

30. Hansen , K.M. , Manikas , K. : Towards a Network Ecology of Software Ecosystems: an Analysis of two OSGi Ecosystems . In: Proceedings of the 25th International Conference on Software Engineering & Knowledge Engineering (SEKE' 2013 ). ( 2013 )