The political dissection of Sept. 11 continues, but for the intelligence community, one lesson is already obvious: Analysts must do a better job of "connecting the dots," seeing patterns in disparate pieces of information and alerting policymakers forcefully to the dangers.

If the intelligence community keeps doing what it was doing before Sept. 11, 2001, it risks making more mistakes that could have disastrous effects. The danger of the status quo was underlined by the intelligence failure in assessing Iraq's weapons of mass destruction; analysts signed off on estimates that provided a rationale for war but that, in hindsight, appear to have been wrong.

So what is the CIA doing to shake up its people and procedures? Can it break down the walls of secrecy and bureaucracy that prevent analysts from communicating more effectively with colleagues at other U.S. intelligence agencies -- let alone with foreign intelligence and security services? Can the CIA reinvent itself for the threats of the 21st century -- and bring outsiders into that process?

Last week I attended a CIA-sponsored conference here that was described as an "experiment" with these disruptive ideas. The conference, "New Frontiers of Intelligence Analysis," brought together officials from intelligence and police agencies of nearly 30 countries. The sessions were all unclassified, and I was one of several journalists and a dozen or so academics who were invited.

"We are in a world where transnational terrorist threats exceed the ability of any one nation to deal with them," explained Carol Dumaine, the conference organizer, in opening the meeting.

Dumaine, at 45 already a 24-year veteran of the agency, heads a small group called the Global Futures Partnership that spent two years arranging the conference with the Link Campus of the University of Malta in Rome. Her team is also organizing similar seminars at Harvard, Stanford and other leading American universities. Its members see their role as in-house dissidents and agents of change, and the very fact that they are in business suggests that top CIA officials know they have a problem and want to fix it.

As at any conference, the quality of the presentations varied, but the best of them were mind-bending. An American technologist described how to use the tools of computer gaming to give intelligence analysts the same advantage that card counters have at blackjack. An American anthropologist who conducted a two-year study for the CIA of its analytical culture explained the cult of the "expert" -- and the danger that the more expertise analysts have, the more likely they are to reject anomalies that don't fit the patterns they've learned to discern.

The anthropologist contrasted the cautious world of the analysts with changes in what was once a similarly conservative culture among surgeons. Today status goes to the surgeons who are innovators and risk-takers, and the profession accepts the idea that it will lose some patients in practicing good, aggressive medicine. "That change has not occurred in the intelligence community," he said.

What I liked was that many of the speakers told the analysts things they probably didn't want to hear. A British scholar said he feared a culture of mediocrity in the intelligence world; he contrasted it with the 1940s, when the brilliant but eccentric code-breaker Alan Turing pedaled to Bletchley Park each day wearing a gas mask. A German warned bluntly: "Too few analysts know enough about Islamic culture to penetrate the mind of a jihadi terrorist." An American professor said that to understand today's threats, "we need more tolerance for views that are not our own."

The analysts' need for more openness and sharing collides with the limits that have traditionally been imposed by the collectors of secret information. But the Rome conference left me wondering if the world can afford to play by the old secrecy rules as it deals with global terrorist networks. An Italian summed up the lesson this way: "To fight a network, you need another network." Personally, I'd prefer the risk of putting too much information into that network rather than too little.

At the end of the three days, the CIA's deputy director for intelligence, Jami Miscik, talked about some of the criticism she had heard of "the cognitive rigidity of analysts" and "the parochial effects of experts." She said she warns her analysts that the "the biggest danger is inherited assumptions that get handed down from generation to generation."

The CIA, working with other intelligence services around the world, needs to break the old culture that allowed Sept. 11 and the WMD mistake to happen. They made a good start last week in Rome. We're living in a world where business as usual is too dangerous. Change is good, especially in the normally closed and bureaucratic world of intelligence.