A lot of Americans responded with a shrug to the revelation that the National Security Agency may be searching for terrorists by tapping U.S. phone calls and e-mails without court permission and in violation of federal law.
To some degree, that response is understandable. The threat that government spying may run amok is pretty theoretical, while the threat of terrorist attack is very real in the wake of Sept. 11. None of us has forgotten watching those towers fall.
However, if you really want to understand what might be at stake in the NSA controversy in terms of our freedom and privacy, the Internal Revenue Service offers a compelling case study in what it calls its Questionable Refund Program.
Under this program, computer programs and data-mining techniques are used to sift through millions of tax refund requests and financial records, using the same basic approaches employed by the National Security Agency to analyze billions of phone calls, e-mails and other data. But while the security agency is looking for clues to potential terrorists, the IRS is mining its data for fraudulent tax-refund requests.
It's a useful technology, but according to a new government study, almost two-thirds of the refund requests identified as fraudulent through IRS data-mining are actually honest and clean. Of those returns categorized by the IRS as "conclusively fraudulent," 46 percent weren't fraudulent at all. In other words, IRS data-mining falsely identifies hundreds of thousands of American taxpayers as tax frauds every year.
Such misidentifications are called "false positives," and they're inherent in data mining. To cite another example, an experiment by the U.S. Army a few years ago used data mining to identify those who might help smuggle American military technology into the hands of the Chinese. Sifting through immense stores of public documents and secret government data, the experiment kicked out thousands of potential linkages, including the names of Condoleezza Rice and William Cohen, then U.S. secretary of defense.
Given that reality, the idea that the security agency's data mining of U.S. communications is capturing only the conversations of terrorists is naive in the extreme. The false positives in an operation of that scale — in other words, the numbers of Americans falsely identified by the program as possible terrorists — are undoubtedly enormous.
However, while false positives are inescapable, government can minimize the impact by handling them correctly, according to Mary DeRosa, a senior technology fellow at the Center for Strategic and International Studies.
The results of data mining should be used only as a tool to initiate more intensive investigation, DeRosa advises, so that "false positives can be discovered before there are significant negative consequences for the individual."
But in an analysis by the Center for Strategic and International Studies, she goes on to warn that without strict guidelines, "the stakes are so high when fighting catastrophic terrorism that . . . government actors will want to take action based on the results of data-analysis queries alone."
That may help explain how so many innocent Americans, including a 4-year-old boy, the police chief of Northfield, Minn., the co-author of the book "Bush's Brain" and U.S. Rep. John Lewis (D-Ga.) have found themselves on terrorist watch lists compiled by the federal government. Those folks have no idea how they got on the lists, and even though 30,000 of them have demanded their names be removed, the federal government refuses to do so.
The IRS data-mining program offers another chilling illustration of the danger. Under IRS policy, tax refunds identified as fraudulent are immediately frozen, sometimes permanently, based on no evidence except that their return matches the computer profile. And because those frozen refunds on average amount to 25 percent of the taxpayers' yearly income, freezing those refunds "often impose(s) a severe economic hardship" on hundreds of thousands of Americans who have done absolutely nothing wrong except turn up as a false positive, Taxpayer Advocate Nina Olson reported to Congress.
Even worse, the names of innocent taxpayers tagged by the program are retained and all future tax refund requests are frozen as well, and they're never told why.
"At a minimum, this procedure constitutes an extraordinary violation of fundamental taxpayer rights and fairness," Olson wrote.
Fortunately, Olson's report has been made public and has drawn the attention of Congress. Presumably, laws and policies will be changed to address that injustice. In fact, DeRosa warns, that kind of oversight is essential to ensure that data-mining technology is not abused and that innocent people are protected.
And in the end, that's the real problem with President Bush's insistence that the NSA spying program be immune to control by federal law, by Congress or the courts. Such unchecked government power will always — always — lead to abuse, and when the technology in question is as powerful as that controlled by NSA, the consequences for innocent individuals can be profound.
Jay Bookman is the deputy editorial page editor of the Atlanta Journal-Constitution. His column appears Mondays and Thursdays.
© 2006 Atlanta Journal-Constitution