Toward an Earth Science Enterprise Federation: Results from a Workshop (1998)

Prior to the workshop, participants and speakers were provided with a paper that applied political principles (federalism) to business management (Handy, 1992). Many aspects of these principles have application to scientific data management. To educate participants about these principles, the workshop began with a set of presentations on different federation models. Then workshop participants were divided into working groups. These groups considered a broad range of issues pertaining to an ESE federation, which may be summarized under four headings: (1) objectives, (2) governance, (3) potential costs and benefits, and (4) measures of success.

Before establishing an ESE federation, workshop participants agreed that a number of issues should be addressed. The following issues do not represent an exhaustive list, but they were deemed to be of high importance:

• Goals. A statement of purpose (e.g., foster effective stewardship of data and data products, provide easy access to useful data products, create innovative high-quality information products) states succinctly the central commitment of members to the guiding principles of the organization. Typically, the goals of an organization begin with a vision and mission statement. The wording should be agreed upon by all the prospective founding members and amended only after a deliberative process. An example of a vision and mission statement crafted at the workshop follows.

• Constituents. The federation should identify its stakeholders. As described in Chapter 1, the

• constituents for an ESE federation include data producers, global change scientists, knowledge brokers, and for-profit businesses. These correspond roughly to NASA's ESIP Types 1, 2, and 3.
• Common needs/values. To function coherently, federation partners should identify common needs and values, including what sources of EOS and non-EOS data are available and whether data are needed in real time or at some point after collection.
• Codes of conduct. The members must establish codes of conduct, particularly expectations of behavior. For example, among the many issues to be decided is whether partners are free to cooperate or compete with the federation.
• Legal standing. To be effective, the federation needs to employ personnel and enter into legally binding commitments. This implies that the federation should be constituted as a corporation (e.g., non-profit corporation) under state laws. All the rules, behaviors (including disclosure of conflicts of interest), outcomes, and intervention mechanisms of the federated structure need to be crafted in "plain English" before lawyers craft formal articles of federation or bylaws.

Governance means the mechanisms by which participants--funding agencies, system operators, data system managers, research and operational users, application specialists--share in the design, implementation, management, and operation of the information system on behalf of the broad constituency of users. There are many examples of federations that could serve as role models for an ESE federation (see Appendix A), but only the partners can recommend a structure and approach to suit their purposes. They can choose between tighter and looser forms of federation. They can also choose between more competitive and more cooperative styles, although science is generally a cooperative venture. The choice of what sort of federation to develop depends ultimately on how prepared members are to sacrifice a degree of autonomy in order to achieve ends (e.g., scientific advances, commercial advantage) they could not achieve alone.

The governance mechanisms implemented for the prototype federation should be viewed as an evolutionary process. The mechanisms adopted in a mature ESE federation may well be different from those of a federation in its infancy.

• Management style. There is a continuum of management styles, ranging from hierarchical to consensual. The former implies a high degree of centralized control, whereas the latter, which is more typical of federations, implies the ceding of power to the lowest possible level, consistent with getting the job done.
• Role of NASA. NASA has several potential roles in an ESE federation: to ensure that the goals of the Earth Science Enterprise are met; to provide funds, and therefore a voice in the federation; to nurture the development of the federation; and to evaluate the success of the federation. At the same time, it should encourage bottom-up management. As pointed out by Charles Handy, federalist centers are meant to be minimalist; they exist to coordinate, not to control (Handy, 1992).
• Autonomy. The choice of the type of federation to develop depends ultimately on the degree of autonomy members are prepared to sacrifice.
• Sharing authority. A key element of governance is the distribution of power and operating functions. In a federation, authority comes from the bottom, not the top. Adherence to the principal of subsidiarity is critical to an effective federation of partners (Handy, 1992). The central organization provides vision and leadership to the subordinate units, not micro-management. Ideally in a federation, subsidiary units should have the power to discharge the manager(s).
• Priorities. A federation exists to meet the needs of its constituents. This implies that the highest priorities of the earth science community at large will come to the fore.
• Resources. The partners must decide how financial resources are distributed throughout

• the federation. Are resources allocated to centralized services to which federation members must then subscribe, or are they allocated directly to the members, who then decide whether or not to subscribe to the central services? In particular, can the members opt out of subscriptions to centralized services? To support these decisions about centralized vs. decentralized services, the federation must put accounting systems in place that track resource flows and relative performance.
• Interoperability. Currently interoperability among EOSDIS participants is provided through the EOSDIS Core System, which allows users to enter the system at any point and find data of interest. Although a common program interface may not be necessary in an ESE federation, some capabilities, such as search and retrieval, will remain important. It is therefore likely that the partners will need to choose protocols, technology, and algorithms to work together.
• Leadership. Visionary leadership is a prerequisite for organizations that have multiple missions. A champion--either from within the federation or from NASA--who works on behalf of the federation, would help ensure its success.
• Accountability. Because the ESIPs were chosen through a competitive process, a NASA project officer will be responsible for ensuring that they meet their contractual obligations. But NASA will be held accountable by Congress, not for the successful fulfillment of the contracts, but for furthering the science.
• Communications. Pathways must be multidirectional: between partners and management, and among partners. Pathways should also be defined for communicating with NASA, the broader community, and international organizations.

• Qualifications. Openness and inclusiveness are desirable attributes of a scientific information system. Before joining, members should give evidence of substantial commitment to the purposes of the federation, including the need to represent and serve the broader community. Similarly, inadequate commitment, and therefore performance, would be grounds for ejection. In the prototype federation, membership is equivalent to receipt of funding following competitive selection by NASA. This criterion, however, precludes active representation of any entity, however committed, that is ineligible from such competition (e.g., a data center responsible to a foreign government).
• Classes of membership. The partners could consider different classes of membership, which would imply different roles and responsibilities among federation partners. Rules for who belongs in what class would also have to be devised. Alternatively, members could have equal status, although this policy could lead to problems as the federation matures.
• Evolution. Initially the prototype federation will be modest in size, and partners will need to define a process for adding new members. Technological opportunities and new scientific needs will also create a demand for new members. As the federation matures, its form and function will undoubtedly change.
• Responsibilities. Although the primary task of members in the prototype federation is to fulfill the terms of their contracts, they should avoid self-interest, while promoting the overall goals and needs of the organization and the broader constituency. It is also important to build in incentives that reinforce rather than discourage good communication and teamwork, and that link the self-interest of members to the needs of their constituents.

• Models. The partners should choose a suitable model for governing the federation, noting that different styles have different implications for the operations of the federation. For example, partners may elect a council or board of trustees, or they may choose a more distributed model, such as the Interact Engineering Task Force, a loosely self-organized group of individuals who make technical contributions to the evolution of the Internet. Working groups to address such issues as intellectual property, technology, and standards may also be established. Finally, the partners should consider the

• relative advantages of having rotating versus fixed headquarters.
• Procedures and processes. After a consensus on the vision and mission statements is reached, the partners should draft the articles of federation, including bylaws and guidelines. Day-to-day procedures for the federation, its sub-units, and the interactions among them need to be established. Processes for accomplishing the tasks of the federation, such as vetting products or defining metrics, will also have to be established.
• Policies. Federation members have diverse interests and will want different policies on such issues as quality control, intellectual property, and standards. With regard to the first, different constituents need different levels of quality control. Intellectual property issues, such as data privacy and commercialization, should be clarified and set in the context of federal law. For example, who has intellectual property rights to products produced by the federation: the federation or an individual partner? Finally, the partners should define a strategy for evaluating, selecting, rejecting, and evolving standards. They should also address whether federal standards (e.g., Federal Geographic Data Committee standards) or EOSDIS standards apply to federation partners.

The costs and benefits listed in Table 2.1 are expressed in general terms only, because they involve assertions about expected human behavior in the context of a "well-designed" federation, contrasted with an equally undefined centralized alternative. Such predictions can only be verified by actual experience, and the arguments supporting them are inevitably largely hypothetical. In addition, the costs and benefits are different for NASA, the federation, the partners, and users.

For the federation to be successful, it is important to decide at the outset how the federation should be judged. To do this, federation partners should develop a baseline from which to measure achievements of the federation, define metrics (e.g., number of new innovations), and obtain feedback from peers. Since success begets success, it is also important for the federation to get off to a good start with some early successful activities. These will instill confidence in the members and may defuse potential critics.

Success may mean different things to NASA, the federation, and individual partners. Some possible measures of success are given below.

Measures of success for NASA include:

• increased productivity of the science in the Earth Science Enterprise;
• wider dissemination of innovative information products;
• lower costs;
• ability to be self-sustaining; and
• reduced dependence on NASA funding. Measures of success for the federation include:
• increased productivity of the science in the Earth Science Enterprise;
• maintenance or advancement of position as the primary source of ESE information;
• satisfaction of the constituents;
• development of new information and capabilties;
• demonstrated commitment of the constituents and NASA;
• attraction of new members;
• ability to be self-sustaining;
• increased size and diversity of user community;
• effective stewardship of holdings (i.e., ensuring the scientific quality and integrity of the information products for the benefit of future generations of scientists); and
• reduced dependence on NASA funding.

Measures of success for the partners include:

• constituent satisfaction with the federation's data sets, capabilities, or technology;
• meeting the terms of the contract by advancing the science;
• easier generation of data products and other information; and
• increased market share.