Most of the report (which I analyzed here) concerns an alleged resurgence of al-Qaida. But toward the end, the authors mention a small but expanding number of “violent Islamic extremists inside the United States” who are “becoming more connected ideologically, virtually, and/or in a physical sense to the global extremist movement.”

The report also notes dangers posed by “non-Muslim terrorist groups,” “single-issue groups,” and “even small numbers of alienated people” who might “find and connect with one another, justify and intensify their anger, and mobilize resources to attack—all without requiring a centralized terrorist organization, training camp, or leader.”

Then comes the point:

The ability to detect broader and more diverse terrorist plotting in this environment will challenge current U.S. defensive efforts and the tools we use to detect and disrupt plots. It will also require greater understanding of how suspect activities at the local level relate to strategic threat information and how best to identify indicators of terrorist activity in the midst of legitimate interactions.

One aspect of this is uncontroversial—the need to integrate city and state law enforcement (”suspect activities at the local level”) to national watch lists and other databases (”strategic threat information”). This effort falls under the Department of Homeland Security, and it’s a disgrace that, with such a large budget, nobody has yet devised a systematic method of doing this.

However, the other part of this passage (”how best to identify indicators of terrorist activity in the midst of legitimate interactions”) pushes a very hot button.

A year and a half ago, passionate arguments broke out over reports about the National Security Agency’s “data-mining” technologies. A couple of years earlier, the Pentagon attempted to fund a still-more intrusive program called the “Total Information Awareness” network. (The latter effort failed, but the concept was almost certainly re-routed to the NSA or elsewhere.) Concerns were raised about privacy rights, the abuse of power, and the worth of such networks to begin with.

Judging from the NIE (not just from these key passages but from its general assessment of a “persistent and evolving terrorist threat”), the debate over these vast surveillance systems will soon be renewed. So, it’s worth making some distinctions that tended not to be made the last time around, at least in much of the public discussion.

The key distinction is the one between data-mining and wiretapping.

In data-mining, the NSA casts a vast “net” across the global communications system, encompassing, at least in theory, every phone call, e-mail, electronic signal, and so forth. The idea is not to monitor all these calls (there’s not enough time or manpower to do that, in any case) but rather to detect patterns. If someone makes or receives a number of calls to or from, say, Pakistan, Iraq, or Syria—or if someone is in frequent contact with someone else who makes or receives such calls—this pattern would set off alarm bells. At that point, an intelligence officer might want to start monitoring that person or those people to see who they are, what they’re doing, and whether anything about them or their behavior seems ominous.

I don’t think anyone, other than libertarians, should have a problem with this sort of data-mining. Nor do I think there should be a problem with the idea of monitoring people whose calling patterns set off alarm bells. I say this even though I have probably set off some alarm bells myself. For instance, I have been in e-mail correspondence with a friend and former professor who is helping to set up a Western-style liberal-arts college in Pakistan. (I should apologize to other friends who could have been caught up in this hypothetical sweep if I phoned or e-mailed them shortly after communicating with my old professor.)

However, let me add a caveat: I have no problem with extending the probe from data-mining to monitoring if certain lawful procedures are followed.

Now, here’s the catch: I don’t know—and I don’t think anybody else without a specially compartmentalized security clearance knows—whether the NSA or other intelligence agencies are currently following lawful procedures. More to the point, I don’t know whether there are lawful procedures.

When the New York Times revealed in late 2005 that the NSA was data-mining inside U.S. territory, some critics demanded an end to the practice unless it had been approved in advance by the “FISA courts.” These are highly classified panels of judges, set up under the Foreign Intelligence Surveillance Act of 1978, who approve (or decline, though they almost always approve) warrants to place foreign powers or their suspected agents under surveillance.

Bush administration officials responded, at the time, that warrants did not—and could not—apply to data-mining. And they were right. Warrants apply to specific individuals, entities, locations, or phone numbers. It’s not possible to request a warrant to track the calling patterns of an entire country’s population.

What was never settled in that controversy, at least not publicly, was whether FISA warrants apply to any subsequent monitoring that takes place as a result of these vast sweeps. If someone wants to monitor my calls because of the e-mails with my friend in Pakistan, does a FISA court have to give permission first? And are there limits to which of my calls can be monitored, what can and cannot be done with the resulting “intelligence,” or how long it can be stored?

Again, I don’t know the answers, nor do I know anyone who does (and I’ve asked around).

This isn’t a trivial matter. Whatever the justification for this sort of surveillance in theory, the opportunity for abuse in practice is enormous—and even likely, as long as the current bunch is in power. The congressional leaders and intelligence committees can find out where these matters stand. They should ask and legislate new safeguards before the next “surge”—a surge in domestic spying—gets under way, because, by all indications, it’s coming soon.