Paul Edwards, a keynote speaker at the EarthCube All-Hands Meeting, shares an interesting viewpoint, sharing what social scientists have learned about governance in cyberinfrastructure and how those lessons may apply to EarthCube.
2. EarthCube goal
“…to design, build, and maintain an easy-to-use
system based on existing resources that embraces
open-source culture and methods to align
technology development with scientific needs.”
Richard et al. “Community‐developed Geoscience Cyberinfrastructure.” Eos 95, no. 20 (2014):
165-166
4. The Tower of Babel…
Heritage of multiple
disciplines, sensors, data
analysis methods
Cacophony of formats,
metadata, software
Earthcube survey of ~175
scientists (2011): need…
Common data formats
Better metadata and
metadata standards
Better ways to find data
Coupled web-based
services, such as
visualization tools
5. Cyberinfrastructure and
climate change informatics (Rood & Edwards 2014)
R. B. Rood & P. N. Edwards, “Climate Informatics: Human Experts and the End-to-End System,” Earthzine, May 2014
6. The loading dock model of cyberinfrastructure
Data Models Services
Loading Dock Model
7. Access is not the main problem
Beyond the loading dock model
Need for translational information for (many)
particular users and uses
Human communication — often informal — remains
the most basic process for effective data sharing
Metadata as product vs. metadata as process
Always provide for communication with data creators
8. This morning
A little history of infrastructure
… and of governance in meteorology
What is governance?
Governance and software in Earth system science
9. This morning
A little history of infrastructure
… and of governance in meteorology
What is governance?
Governance and software in Earth system science
10. Infrastructure: a historical model
Paul N. Edwards
System building: designed,
coherent, centrally organized
Proliferation of systems;
variation
Networks: dedicated gateways
link heterogeneous systems
Internetworks: generic gateways
link heterogeneous networks
Decentralization, fragmentation
Abandonment, substitution
time
Edwards et al. 2007
11. Dedicated or improvised gateways (Egyedi 2001)
Paul N. Edwards
Whose responsibility?
Who sets standards?
Who pays?
14. Internetworks link networks
Paul N. Edwards
Routers are gateways
connect computers to
each other (network)
… and connect the local
network to other
networks
“The” Internet connects
millions of networks
15. This morning
A little history of infrastructure
… and of governance in meteorology
What is governance?
Governance and software in Earth system science
18. 1870 1900
1930 1960
Surface station coverage: evolution
Source: J. Hansen and S. Lebedeff, “Global Trends of Measured Surface Air Temperature,” Journal of
Geophysical Research 92, no. D11 (1987), 13,346-13,347. Diameter of circles drawn around each station
19.
20. Stages in the history of weather forecasting
20
Systems: national
weather services
Set own standards
Networks: national and
international
The Réseau Mondial
Internetworks
Integrating
heterogeneous data
sources
Surface stations
Air bases and airports
Marine data
Satellites
Governance
International Meteorological
Organization (1873-1949)
World Meteorological
Organization (WMO, founded
1950)
Set standards, assisted
coordination — but
lightweight relative to
national services
22. This morning
A little history of infrastructure
… and of governance in meteorology
What is governance?
Governance and software in Earth system science
23. What is governance?
Aligning an organization’s practices and procedures
with its goals, purposes, and values
Oversight, steering, and articulating organizational
norms and processes
vs. management: detailed planning, supervision of work,
allocation of effort
24. Modes of governance
Hierarchy Network (of
firms)
Market or firm Bazaar
Contractual
framework
Employment
contract
Neoclassical
contract
Property
contract
Open source
license
Incentives
intensity
Low Medium High Low
Control
intensity
High Medium Low Low
Social
relations
Strong ties Strong ties Anonymous Mostly
anonymous or
weak ties
Membership Employees
selected
Members
select each
other
Buyer selected
by seller
Open; many
free riders
Timeframe Long-term
commitment
Long-term
commitment
Transaction or
contract
Variable; no
commitment
Source: adapted from B. Demil and X. Lecocq, “Neither Market Nor Hierarchy Nor Network: The
Emergence of Bazaar Governance,” Organization studies 27, no. 10 (2006): 1447-66
25. Open source culture: bazaar governance
E. Raymond, “The Cathedral and the Bazaar”
Linux is ‘a great babbling bazaar of differing agendas and
approaches’
Characteristic: chaotic market, huge variations in quality
“Low levels of control and weak incentives intensity
are distinctive features of bazaar [governance],
lending a high uncertainty to governed transactions.”
Source: B. Demil and X. Lecocq, “Neither Market Nor Hierarchy Nor Network: The Emergence of Bazaar
Governance,” Organization Studies 27, no. 10 (2006): 1447-1466.
26. …but how does governance really work?
Highly competent groups
can get a lot done without
much management from
above —
but there are limits to
leaderless teams,
especially when work is
time-sensitive and
requires coordinating
complex, interdependent
activity.
27. This morning
A little history of infrastructure
… and of governance in meteorology
What is governance?
Governance and software in Earth system science
28. Organizations in science…
Organizations provide space, equipment, money, and
support
Stable, long-lasting (decades)
Well-defined roles and routines
Have boundaries, hierarchies, and entrenched cultures
Research (NCAR, GFDL, universities) vs. operational (NOAA,
NASA, DOE)
National laboratories and military research
Funding agencies (NSF, NIH) and foundations
They strongly structure work incentives and disincentives
29. … vs. projects
...but most scientific work takes place in projects,
teams, and working groups
Varying sizes
Lifespans vary, but mostly short (1-5 years)
Depend heavily on funding cycles
Often cross organizational boundaries
Many scientists are involved in several projects at
once
Overlapping membership
Funding is an ongoing concern
34. Operational norms and rules
Expectations that govern everyday interaction
among project members
Largely informal and tacit (unarticulated)
May be embedded in organizational routines or tools
Usually surface only during crisis or conflict
Difficult to change without a forcing factor
Tools can embody operational norms — but usually can’t
force changes
35. Cyberinfrastructure pitfalls
35
Software makes it seem easy to build gateways
between systems and networks…
“You just…”
… but social, institutional, and security gateways are
even more important
Multiple institutional cultures
Complex projects with many working groups
Multiple security and legal standards can block
interchange
36. Conclusions: some lessons from history
36
Centralized design and control is not the primary
path to working infrastructure
Instead, build gateways (couplers)
Standards technologies, institutions
Must be lightweight, readily understood, easily transferred
across regions and cultures (including disciplinary
cultures)
International governance of data standardization and
exchange in meteorology was achieved by the
1960s
in the face of enormous technical obstacles
(communication channels) and social obstacles (Cold
War, decolonization)
37. EarthCube goal
“…to design, build, and maintain an easy-to-use
system based on existing resources that embraces
open-source culture and methods to align
technology development with scientific needs.”
Richard et al. “Community‐developed Geoscience Cyberinfrastructure.” Eos 95, no. 20 (2014):
165-166
38. Conclusions: some lessons from history
The tensions between hierarchy, network, and
bazaar modes of governance will be difficult to
resolve
Cyberinfrastructure can help, but it can also hinder
Social and organizational issues must be addressed
along with technology
The EarthCube experiment is enormously important,
and worth doing!
39. 25 July 2014Paul N. Edwards , University of Michigan
School of Information
Edwards et al., Knowledge
Infrastructures: Intellectual
Frameworks and Research
Challenges (2013)
knowledgeinfrastructures.or
g