When the GAO Gets In Over Its Head

he General Accounting Office does a good job by and large of crunching numbers and adding up balance sheets. But it diminishes its credibility when it advocates public policy, says a report released in mid-October by the nonprofit National Academy of Public Administration.

Yes and no. There's a thin line between advocating a public policy -- slash U.S. defense spending in the post-Cold War world, let's say -- and coming up with a conclusion based on sound auditing, i.e. spending more money on B-2 bombers is a dumb idea. What we've got here is classic Capitol Hill fallacies of logic. There's premise A: Congressional officials sic the GAO on politically expedient targets. And like news stories, audits have a wonderful tendency to turn out the way they were conceived before the facts were gathered.

Then there's premise B, raised by Republicans after this month's audit of the auditors: That the GAO does far more work for Democrats than Republicans, ipso facto GAO is a tool of the liberal Congress. Since GAO calls itself a "customer" of Congress, a terminology it should have left to the business process reengineering folks, it's easy to make a partisan connection -- even though it's flapdoodle.

Premises A or B won't get anyone anywhere. Here's the operant problem: the GAO auditors are generalists, as their name implies.

When it comes to complicated technologies like agency-wide computer systems -- in which even the creators of the technology aren't sure how to make it work -- your basic accountant is going to do what he or she was taught to do: Add up the numbers, apply the measurements, and come to a conclusion. Trouble is, that kind of evaluation is unsuited to arrangements like CIM, TAC-IV, CALS and others, which go by a sort of cost measurement all their own: "Logically, we think automating the Pentagon's pay record system makes sense; it's bound to be cheaper and more efficient than paper, but we're not sure how to measure that yet, and the software isn't written...but we think we ought to go ahead."

GAO audits don't have that sort of measurement built in. In fact, it's difficult to see how any classic technocratic assessment could. The curious case of Maine Republican Sen. William S. Cohen's brain-dead "investigative" Oct. 12 report, "Computer Chaos," is built on sundry GAO reports and doomed from the start: You have premise A at work whilst Cohen tries to validate premise B. What results is a waste of paper and time, a superficial treatment of an important subject, and a general public left with a false impression.

We have an answer to the operant problem: Combine the GAO with the Congressional Office of Technology Assessment, a piece of government whose true purpose and use seems to escape everyone. Plug the OTA technical types into GAO audits of complex business science problems. We can't guarantee more intelligent audits, but we would bet a combined GAO-OTA would be, let's say, deliberative and thoughtful enough to ward off quickie political hatchet jobs.