Someday Americans may die because of Congress’s decision earlier this month to cripple a Defense Department program designed to catch future Mohamed Attas before they strike. That’s not a prediction. But it is a fear.

The program seeks to develop software to make intelligence-sharing more effective by making it instantaneous, the better to learn more about suspected terrorists and identify people who might be terrorists. It would link computerized government databases to one another and to some nongovernment databases to which investigators already have legal access. If feasible, it would also fish through billions of transactions for patterns of activities in which terrorists might engage.

But now these goals are all in jeopardy, because of a stunningly irresponsible congressional rush to hobble the Pentagon program in ways that are far from necessary to protect privacy. This is not to deny that, absent stringent safeguards and oversight, the ineptly named "Total Information Awareness" program might present serious threats to privacy. It might, for example, subject thousands of innocent citizens and noncitizens alike to unwelcome scrutiny, and might even expose political dissenters to harassment by rogue officials.

But some curbs on potentially dangerous (and potentially life-saving) government powers in the name of civil liberties are not necessary to protect privacy and exact too high a price in terms of endangered lives. Congress’s rush to strangle TIA in its infancy is such a case. It makes little more sense than would a flat ban on any and all wiretapping of phones that might be used by U.S. citizens. Like TIA, wiretapping poses grave risks to privacy if not carefully restricted. So we restrict it. We don’t ban it.

The problem with the near-ban on TIA-sponsored by Sens. Ron Wyden, D-Ore., and Charles Grassley, R-Iowa, and known as the Wyden amendment-is that rather than weighing the hoped-for security benefits against the feared privacy costs, and devising ways to minimize those costs, Congress was stampeded by civil-libertarian hysteria into adopting severe and unwarranted restrictions. The Bush administration shares the blame because the person it put in charge of TIA research is Adm. John M. Poindexter, whose record of lying to Congress about the Iran-Contra affair does not inspire trust.

"There are risks to TIA, but in the end I think the risks of not trying TIA are greater, and we should at least try to construct systems for [minimizing] abuse before we discard all potential benefits from technological innovation," says Paul Rosenzweig, a legal analyst at the Heritage Foundation who has co-authored a thoughtful 25-page analysis of the TIA program, including a list of muscular safeguards that Congress could adopt to protect privacy and prevent abuses. Instead of weighing such factors, Rosenzweig says, Congress has "deliberately and without much thought decided to discard the greatest advantage we have over our foes-our technological superiority."

The Wyden amendment seems reasonable enough at first blush. That may be why all 100 senators and the House conferees voted to attach it to the omnibus spending bill that cleared Congress on February 13. The amendment allows pure TIA research to continue, if the administration files a detailed report within 90 days or the president invokes national security needs. And the amendment’s restrictions on TIA deployment have been sold as a temporary move to allow time for congressional oversight.

But such measures, once adopted, are a good bet to become permanent in today’s habitually gridlocked Congress, where determined minorities have great power to block any change in the status quo. And the Wyden amendment’s impact is likely to be far broader than advertised. It flatly bars any deployment of TIA-derived technology, by any agency, with exceptions only for military operations outside the U.S. and "lawful foreign intelligence activities conducted wholly [my emphasis] against non-United States persons" (defined to mean nonresident aliens).

The scope of the latter exception is ambiguous. But Rosenzweig fears that it will be read narrowly, and that the Wyden amendment will be read broadly-especially by officials fearful of congressional wrath-as barring virtually all uses of TIA technology, even to search the government’s own databases for suspected foreign terrorists. This is because virtually all large databases are "mixed": They contain information about U.S. citizens, resident aliens, and nonresident aliens alike.

In any event, the Wyden amendment quite clearly prohibits any use of TIA technology to pursue the unknown but apparently substantial number of U.S. citizens and resident aliens who may be loyal to Al Qaeda, such as suspected dirty-bomb plotter Jose Padilla and the six suspected Yemeni-American "sleepers" arrested in Lackawanna, N.Y., last year. As a technical matter, the FBI, the CIA, and the Department of Homeland Security remain free to develop and deploy similar technology on their own. But they will hesitate to risk charges of evading Congress’s will. Not to mention the wastefulness of barring these agencies from building on the TIA technology already developed by the Pentagon.

How did TIA become such a dreaded symbol of Big Brotherism? Part of the reason was well-founded concern that unless strictly controlled, the more exotic uses of TIA, such as surveying billions of transactions involving hundreds of millions of people for patterns deemed indicative of possible terrorist activities, could subject huge numbers of innocent Americans to scrutiny as potential terrorists. But Rosenzweig and others who share these concerns, including officials of the TIA program itself, have already been crafting safeguards. Among them are software designs and legal rules that would block human agents from learning the identities of people whose transactions are being "data-mined" by TIA computers unless the agents can obtain judicial warrants by showing something analogous to the "probable cause" that the law requires to justify a wiretap.

It was largely misinformation and overheated rhetoric from civil-libertarian zealots-on both the left and the right-that pushed the Wyden amendment through Congress. The misinformation included the false claim that Poindexter would preside over a domestic spying apparatus, and the false suggestion that TIA was poised to rummage through the most private of databases to compile dossiers on millions of Americans’ credit card, banking, business, travel, educational, and medical records and e-mails.

To the contrary, Poindexter’s job is limited to developing software. And even without the Wyden amendment, TIA would give investigators access only to databases and records-government and nongovernment-that they already have a right to access. Its most basic function would be simply to expedite the kinds of intelligence-sharing that might have thwarted the September 11 attacks, by linking the government’s own databases with one another and with any legally accessible private databases. The goal is to enable investigators to amass in minutes clues that now could take weeks or months to collect.

Here’s a hypothetical example (adapted from Rosenzweig’s analysis) of how as-yet-non-existent TIA technology might help stop terrorists-and how the Wyden amendment might prevent that.

Say the government learns from a reliable informant that the precursor elements of Sarin gas have been smuggled into the United States by unidentified Qaeda operatives via flights from Germany during the month of February. Its first investigative step might be a TIA-based "query" of foreign databases that might help generate a list of possible terrorists. (But the Wyden amendment would bar a TIA-based query for the names of any who might be Americans. And it could be construed as putting entirely off-limits any "mixed database" that includes Americans.)

A second step might be a pattern-based query to U.S. government databases to produce a list of all passengers, or perhaps all nonresident aliens, entering the U.S. on flights from Germany during February. (But the Wyden amendment would bar a query for all passengers, and would again pose the mixed-database problem.)

A third query might seek to find which of these passengers’ names are also in government databases of known or suspected terrorists. (But the Wyden amendment would pose the same obstacles.)

Fourth, with a list of subjects for further investigation based on these queries, TIA could be used-perhaps after obtaining a judicial warrant-to link to any legally accessible commercial databases to find out whether any of these subjects has bought canisters suitable for deployment of Sarin gas, or rented airplanes suitable for dispersing it, or stayed in the same motels as other subjects of investigation. And so on. (But for the Wyden amendment, that is.)

It is not yet clear whether it is even possible to develop technology powerful enough to do all of this. But it might be possible. Shouldn’t we be racing to find out?