1. Three Critical Matters in Big Data Projects for e- Science Kerk F. Kee, Ph.D. Assistant Professor, Chapman University Orange, California www.ekerk.com Kerk.kee@gmail.com A CKNOWLEDGMENT The research reported in this paper is supported by the National Science Foundation, Grants ACI #1453864 and ACI # 1322305. 1

2. Introduction Big data projects are springing up across various fields and industries. Saltz – little is known about the development methodologies, team processes, and short-term frameworks in order to facilitate, guide, and support big data projects towards effective, productive, and successful outcomes. Saltz –based on his review, more than 50% of big data projects never get completed. In fact, many more big data projects fail to achieve their project objectives Focus on the context of e-science. Critical matters are most salient during the design, development, adoption, use, implementation, integration, and the ultimate diffusion of computational tools for big data analytics, simulations, visualizations, and mathematical modeling in science and engineering research. 2

3. Unique Challenges for Big Data Projects in e-Science Inter-disciplinary, Multi-institutional Funded on a limited basis (i.e., three to five years by NSF and other federal agencies) With the goal of advancing science (not directly the technologies) With a mix of full-time and part-time participants all working on a portion of their employment Create experimental tools to harness big data for grand challenges in science. 3

4. Three Critical Matters 3 Different User Groups Better identify their needs and unique situations for designing project methodologies. Mutually Constitutive Perspective Identify strategies to develop innovative team methodologies and improve big data projects. Virtual Organizational Capacity The lack of organizational capacity in various areas presents serious threats to big data projects’ success. 4

5. Critical Matter #1 – Three Groups of Users Lee, Bietz, and Thayer – traditional conceptualizations of technology ‘developers’ and ‘users’ do not fully capture the participants’ roles in e-science projects and stakeholders of cyberinfrastructure systems. Scientist-developers Understand the scientific questions (i.e., requirements), but have limited computational skills Hard to keep up with both domains (i.e., science & programming) Co-producers Tools are often developed to serve the domain scientists involved. NSF funds science (i.e., requirements), and not the technology Pure users of existing tools Open source tools; Adaptation to the design of the tools. Technology integration – they do not always have prior experience and tacit knowledge to successfully adopt and integrate the tools Big data project methodologies should recognize their unique needs 5

6. Critical Matter #2 – Mutually Constitutive Perspective Leonardi (2009) argues that we have made two wrong assumptions ‘technological development and technology use’ Technological development (i.e., object) vs. Technology use (i.e., practice) separated by the act of implementation. Need to work closely in sprints, what Saltz describes as agile software development methodology. A bias for actions – to give users an early prototype to test and critique ‘technology and organizing’ Technology in an organization – the technology, and the people working with the technology, and the people working with each other around or via the technology. Organizing enabled by technologies – the people interacting with each other around the technology, interacting through the technology, and the technology itself. Fundamental units of analysis are people, technologies, and interactions. Big data project methodologies should be agile, and focus on identifying the main people and technologies, then designing strategic interactions to link the people and objects in productive ways. 6

7. Critical Matter #3 – Virtual Organizational Capacity & Capacity Building Strategies Cyberinfrastructure as a multidimensional innovation: Material objects (such as networks, hardware, software, big data, etc.) Organizational practices (such as co-production of project-driven computational tools, distributed collaborations of multi-disciplinary experts, etc.) Philosophical ideologies (such as the belief that mathematical modeling, computer simulation, and computational visualization of big data is as valid a research approach as the traditional theoretical and experimental approaches). 7

8. Critical Matter #3 – Virtual Organizational Capacity & Capacity Building Strategies Cyberinfratructure (CI) as a dynamic innovation Static – pre-designed, mass-produced, bought off-the-shelf, and used-as- instructed after full product development, so the technologies remain static while the adopting organizations undergo new changes during implementation. Dynamic – user-driven, custom-made, produced on demand, permanently beta, and more importantly, being put to use while simultaneously being developed in CI-enabled virtual organizations, which constantly co-evolve along with the technologies. 8

9. Critical Matter #3 – Virtual Organizational Capacity (VOC) & Capacity Building Strategies Organizational capacity: stable funding, empowering policy, domain knowledge, computational expertise, technology training, high performance computing resources, collaboration technology skills, communication competence, personnel and human resources, critical support, etc. Effective big data project methodologies needs to be sensitive to the range of factors that constitute VOC, so the methodologies can be designed to cater to the unique needs, challenges, and situations of the 3 user groups. 9

10. Three Critical Matters 3 Different User Groups Better identify their needs and unique situations for designing project methodologies. Mutually Constitutive Perspective Identify strategies to develop innovative team methodologies and improve big data projects. Virtual Organizational Capacity The lack of organizational capacity in various areas presents serious threats to big data projects’ success. 10

11. Conclusion & Future Research How may agile practices (between technology development and use) be adapted to support big data projects? How can ‘mutual constitutiveness’ between technology and organizing in big data projects be empirically characterized? What specific types/factors of organizational capacity must be developed in order to enable successful operation of big data projects? How can specific types of organizational capacity be conceptualized and operationalized for organizational assessment? How can virtual organizations for big data projects be helped in understanding where they are lacking in capacity? 11

12. Three Critical Matters in Big Data Projects for e- Science Kerk F. Kee, Ph.D. Assistant Professor, Chapman University Orange, California www.ekerk.com Kerk.kee@gmail.com A CKNOWLEDGMENT The research reported in this paper is supported by the National Science Foundation, Grants ACI #1453864 and ACI # 1322305. 12