1. Principles for Success in Collaboration Experiments

2.  Among NSF-funded efforts, Unidata has been an unusually collaborative success story  This success derives, at least in part, from the application of principles that may be critical in other collaborative contexts.

3.  A sense of community ownership ◦ Earliest plans thru latest “Strategic Outlook”  Traits like those defining a “collaboratory” ◦ See Fulker, Bates, Jacobs in BAMS 1997  Decentralized organization & architecture ◦ Multiple data sources to multiple recipients with no data center ◦ Reliance on cooperating tool builders

4. Jim Collins Good to Great (2001) : – Disciplined thought – Disciplined action – Persistent core Treacy & Wiersema Harvard Business Review (1993) : – Primary alignment with 1 of 3 value disciplines Operational Excellence Customer Relations Product Leadership

5.  Willingness to confront reality  A simple, coherent strategic concept  Unidata manifestation ◦ Constant reminders (Policy & Users Committees) about real customer needs & fiscal constraints ◦ Program renewal via peer-reviewed proposals ◦ Strategic focus on building/providing tools (catalysts, not completed reactions)

6.  Willingness to say “no”  Clear assignment of responsibilities  Advancement as a cumulative process  Unidata manifestation ◦ Simple governance model (under UCAR) ◦ NSF exerts (strong) leadership & (patient) support ◦ Successfully conducting long-term development alongside (immediate) user support

7.  An enduring purpose  Immutable core values (as basis for advances)  Unidata manifestation ◦ Current strategy likely resonates with founders ◦ Remarkable core values, such as  Continuous rebalancing of development vs. support  Centrifugal force (distributed functionality…)  Smart use of abstractions

8.  Primary alignment with one of three value disciplines (while attending to the other two) ◦ Operational Excellence (Dell, e.g.) ◦ Customer Relations (IBM, e.g.) ◦ Product Leadership (Apple, e.g.)  Opinions may differ on which is primary for Unidata…

9.  Unidata manifestation ◦ Data access & sharing via community-owned data relays (IDD) & servers (THREDDS…) ◦ Data analysis & visualization via free (mostly open- source) software ◦ Well established reliability record  Is this the primary discipline? ◦ May be most important factor for users with small budgets but some computing experience

10.  Unidata manifestation ◦ High expertise dedicated to support ◦ Training workshops ◦ User meetings  Is this the primary discipline? ◦ May be most important factor for users with the least computing experience (or heavy reliance on student operation)

11.  Unidata manifestation ◦ Truly unique software (important abstractions)  netCDF  IDD/LDM  IDV, Bundles (on VisAD)  RAMADA, THREDDS, LEAD  Is this the primary discipline? ◦ May be most important factor for expanding into new communities

12.  While Unidata success clearly depends on ◦ Operational excellence ◦ Customer intimacy  Unidata has evolved to be a design-centered organization, and the primary value discipline (defining Unidata’s future) is ◦ Product leadership

13.  My WAG: Unidata’s product leadership will significantly affect scientific “publication” (as that notion evolves…) ◦ Data references will be common in bibliographies ◦ Increasingly complex ideas (requiring data and “bundles” to describe them) will take shape collaboratively in wiki- or blog-like contexts ◦ The speed of idea propagation will increase ◦ Libraries will become Unidata users

14.  Unidata probably is reproducible (in other fields, e.g.) with attention to principles  Key principles underlie Unidata because the individuals who shaped the program (many of them in this room) applied them instinctively  Thanks for a fun ride!