It's refreshing to have a big debate in Washington. Too often our debates are small and arcane. The Sept. 11 commission has touched off a much-needed debate of constitutional proportions: How do we best organize the intelligence functions of the government to protect the nation, yet oversee those functions to protect our citizens from the government?

The commission has rendered an enormous contribution to the nation. But its recommendations need to be the starting point for a great debate, not the final word. Political passions are rising, which portends danger. The American system of government is designed to move slowly, for good reason. Such a big and complex country needs to fully consider all the implications of major changes. We make mistakes when we move quickly, and we can't afford to make a mistake here.

Good as they are, the commission's recommendations are too narrowly centered on one problem. This is understandable. The commission was established to examine the problems the government had detecting and preventing the terrorist attacks on Sept. 11, 2001. By definition, that was a matter of coordination among elements of the government, both vertically within organizations and horizontally across institutions. This is often referred to as the "connect the dots" problem.

But that isn't the only trouble with the intelligence community. Before the war in Iraq, the policy and intelligence communities held the near-unanimous conviction that Iraq was chock full of chemical and biological weapons, yet we found nothing. We collectively embraced a uniform mind-set, which is every bit as serious a problem as connecting the dots.

The field of view of our intelligence community is too narrow. The community is relatively small and its component institutions isolated. It is understandably and necessarily preoccupied with protecting sources and methods. And bureaucracies naturally fight for resources. In that environment, intelligence bureaucrats, like bureaucrats in any organization, strive to please their policy bosses. Taken together, these factors contribute to an endemic narrowness of perspective. The shorthand label given to this problem is "groupthink."

We need to fight that narrowness by creating more competition for ideas in the intelligence assessment world. The competition among ideas is improved when different organizations reporting to different bosses compete for better insights and perspectives. Bringing together the entire intelligence community under a single boss who exercises budget and personnel control would further constrain the constructive competition we need within the intelligence community.

The two great problems -- connecting the dots and avoiding groupthink -- are in tension with each other. Implementing an organizational solution to just one of the problems will worsen the other.

The great debate underway in Washington has two camps. The Sept. 11 commission, Sen. John Kerry and many congressional leaders believe a new director of national intelligence (DNI) can succeed only if the person in that job controls the budgets and personnel of the intelligence agencies. People in this camp would leave the agencies with their host departments but give the budgets and control of personnel to the new director.

President Bush chose a different path. His plan would create a relatively weak DNI, whose power would come from managing a set of interagency processes and supervising a set of ill-defined new centers. Unfortunately, if unintentionally, this approach also diminishes the bureaucratic standing of the CIA.

In sum, both approaches are flawed. I know from personal experience in government that ambiguous command authority is dangerous. Keeping intelligence agencies within a department whose budgets and senior leadership depend on people outside the department won't work. Similarly, we have a long history to demonstrate that the power and standing of central coordinators of interagency processes -- Washington policy wonks now call them "czars" -- deteriorate rapidly with time.

More fundamentally, each of these two approaches solves one of the great problems but exacerbates the other. The Sept. 11 commission's proposal would improve "dot-connecting" but would threaten competition among ideas. The president's recommendation would better sustain idea competition but do little to solve the problem of interagency coordination.

Frankly, I didn't favor the idea of creating a DNI, but I understand politics. Both political leaders in a hotly contested campaign have endorsed it as a symbol. We will have a DNI. We now have to ensure that we get a good solution. There is a third path.

The new DNI should run the existing interagency intelligence centers or their successors and coordinate the tasking process. But the DNI needs to be undergirded with real institutional power. The technical collection agencies -- notably the National Reconnaissance Office, the National Security Agency and the National Geospatial-Intelligence Agency -- could be transferred to the DNI. The new director would manage the factories that provide raw material and support to the intelligence bureaus, which would remain within the Cabinet departments.

This approach would facilitate the integration of data collection while preserving diversity of perspective across the community for purposes of strategic assessment. Cabinet secretaries could devote their energies to demanding better analysis, rather than managing large bureaucracies that run machines to collect raw material for the intelligence process. This approach also would ensure that oversight of domestic surveillance on American citizens remained a responsibility of the attorney general, who is charged with protecting our civil liberties. Even here, however, the FBI could turn to the central collection agency, but under the attorney general's supervision.

My friends in the Defense Department are shocked that I have suggested this approach. Modern American war-fighting is more dependent on high-technology intelligence than ever before, they note. We cannot decouple the close working ties between our intelligence capabilities and our war fighters.

But there are ways to ensure that we sustain those close working ties. We should continue to send our best military personnel to work in these agencies and to support national collection efforts with tactical military intelligence systems. The DNI should have a board of directors made up of senior operators from the supported departments. And underlying it all is what I know to be true: that all civilian employees in these agencies consider it their highest priority to support the American warrior in combat. That will not change, even if these institutions report directly to a DNI.

Yes, there will be challenges and problems, but they are manageable. It is said that the intelligence community needs a reform like that of the Goldwater-Nichols Act, which transformed the Defense Department. In fact, Goldwater-Nichols changed the Defense Department because it institutionalized demand for better capabilities from the military services. The Pentagon fiercely fought Goldwater-Nichols when it was proposed by Congress. Now it swears by its results. We have proved in the Defense Department that we can bring competing institutions together for a common purpose without forcing people to wear a common uniform.

The writer is president and chief executive of the Center for Strategic & International Studies and a former deputy secretary of defense. The views expressed here are his own.