18 Association of Research Libraries Research Library Issues 298 — 2019 courses available via Coursera at https://escience.washington.edu/ education/mooc/, and all three institutions contributed to the open access book The Practice of Reproducible Research: Case Studies and Lessons from the Data-Intensive Sciences, available at https://www. practicereproducibleresearch.org/.3 Core to the development of data science at the three MSDSE universities was a posture of non- competitiveness with existing programs and disciplines, particularly in computer science, engineering, and statistics, and a belief that the set of tools and methods that comprise data science are critical to unlocking new discoveries across all research domains. Abt Associates conducted an assessment of the grant in 2015 (at the midway point), and published its findings in 2019. Abt found that the MSDSE funding strategy was effective at increasing positive conditions for data-driven discovery at academic institutions, and noted that “the culture of partnership and experimentation adopted by the program facilitated mutual learning and growth.”4 The Abt report also stressed the importance of strong management and staffing, the success at establishing promising career tracks, and new programs that led to collaboration. In a secondary review of 20 other data science entities, Abt also found that there is no one way to create a data science center: culture, administration, physical space, funding, and people all combine in different ways to work toward supporting data-intensive science on campuses. Finally, a concentration on ethics was not a formal objective of the funding agencies providing MSDSE funding, but NYU, UC Berkeley, and UW have all developed an emphasis on data science and ethics, or data science and the public good. As the sites developed informal and formal curricula, and started working with students, it was immediately clear how important ethics education is to these efforts. There is no one way to create a data science center: culture, administration, physical space, funding, and people all combine in different ways to work toward supporting data- intensive science on campuses.