The debate on whether to centralize or decentralize our IT systems in the federal government is as hot as ever — even after 50 years or more. Frankly, I don't think the debate will ever end, as there is no simple organizational system for IT except in the case of small, public organizations. And in the federal sector, we don't do small.
So what is the solution? First, we could view our IT systems as a possible mirror of our organizational structures; in doing so, we might find some useful guidelines. After all, information systems — other than those that perform specific, tactical processes such as payroll or just-in-time ordering — should mirror the organization in which they are embedded. For example, information flows throughout the organization should be mirrored by our IT systems; so should decision processes, spans of control and headquarters-versus-field operations.
But is the solution really that simple? Is it only a matter of mirroring organization structures with IT structures? Another possible answer is provided by Paul R. Lawrence and Jay W. Lorsch in their seminal work, "Organization and Environment," published by Harvard Business School. In that book, Lawrence and Lorsch argue that organizations should be designed with the same or greater level of variety found in their external environment. In other words, variety in an organization is required to engage effectively with variety in the environment.
The Law of Requisite Variety is so important to the physical, social and biological sciences that we see its application in cybernetics, management, crowd control and information assurance, to name just a few areas of human endeavor. In each case, the effective and successful organizational design is one that has the same or greater variety than the system that is being engaged. This approach would allow for both centralized and decentralized IT systems in today's federal agencies.
When the external environment is pressing on the organization and the decision cycle is short, fast response times are required. For example, when a homeland or national security crisis emerges, decentralized processes and IT systems are critical to survival.
During both Gulf wars, the command and control systems designed for peacetime were cumbersome, slow and ineffective when boots were on the ground and moving. Young soldiers and Marines quickly configured their own computers and mobile devices to communicate with one another in real time and shorten the decision cycle.
During peacetime, when time is not of the essence and our national security organizations can focus on their internal processes, greater centralization of IT systems is the way to go. Centralized IT systems are simply cheaper and more efficient.
Thus, once again, we can find both centralized and decentralized IT systems in our federal agencies, depending to a great extent on the influence of the external environment and the permeability of our organizational and nation-state boundaries. And that is how it should be.
In the '60s, mainframes were behind doors, giving operators full, centralized control of their IT system. Next came the advent of dumb terminals and the beginning of decentralization. Everyone could access the mainframe. And finally, full-blown, decentralized "power to the user" arrived with the invention of Tandy's TRS-80, the Apple I and eventually the IBM PC.
Those were simple days, and we will not be returning to them. Today, our IT systems and architectures are complex — sometimes so complex that they challenge human understanding. Our federal organizations are also complex. And finally, the answer to the long-lived and recurring debate — "to centralize or decentralize?" — is complex and always will be.
Gerry Gingrich is a professor at the National Defense University and an expert on information technology strategy, leading and managing innovation, and technological and organizational change. She can be reached at gingrich@ndu.edu.
