The Obama administration’s brave new disclosure policy.

On his first full day in office, President Obama promised a “new era of transparency” for government. But that promise has gone largely unfulfilled, as the executive branch launched a gajillion clunky Web sites and a nifty Flickr photo stream. Today, the administration finally pinned down its transparency policy: The White House is not just making things public. It’s making public the process of making things public. This is not just transparency; it’s metatransparency.

That’s the theory behind the administration’s new Open Government Directive (pdf). The initiative, unveiled Tuesday by the Office of Management and Budget, requires every federal agency to come up with three “high value data sets” and release them in the next 45 days. Agencies then have to set up their own open government Web sites—http://www.[agency].gov/open—to showcase their transparency efforts. OMB will then regularly check in about new ways the agencies are improving in the areas of “transparency, participation, and collaboration.”

How do we know the agencies will play ball? The White House is setting up what it’s calling an “Open Government Dashboard” to monitor how the different agencies are doing. Think of it as a combination of data.gov and a high-school hall monitor. On the dashboard will be the best data sets and evaluations of each agency’s progress at meeting the new transparency standards. If your agency is doing well, everyone will know. If you’re not doing well, everyone will know. Meanwhile, on each agency’s Open Government site, readers will be able to browse the newly disclosed data sets, discuss them, manipulate them, rearrange them, and request new ones.

The executive branch already has a bunch of data sites and several blogs. But none of them is particularly revolutionary. For one thing, the changes weren’t governmentwide. For another, they were fairly superficial. It’s fun to see Obama shooting hoops. But it doesn’t change the public’s relationship with government.

Tuesday’s directive does—or it will, if the White House follows through. “It is going to take work for this to be significant and have an impact,” says John Wonderlich of the Sunlight Foundation. The White House has already signaled to agencies that it’s serious about getting good data out of them. Now it has to prove it.

The first problem is figuring out what kind of data the public cares about. For example, what exactly is a “high-quality data set”? As any researcher knows, there are good data and there are bad data. Good data means accurate, timely, well-formatted (spreadsheets) digital information that can be stored, sorted, reorganized, graphed, and combined with other good data. Bad data means old, poorly organized, crappily formatted (paper or PDF) information that takes hundreds of man-hours to organize properly, let alone analyze.

Here’s one example of “good data” just announced by the Department of Housing and Urban Development. (Every Cabinet department had to announce a new data set as part of the launch; see an appendix here.) Inspectors across the country examine public housing projects and give them a rating based on how they comply with building standards and housing codes. You can then rank the buildings and cross reference that list with owners of the buildings. “That would answer, ‘Who are worst slum lords in the United States?’ ” says Bill Allison, also from the Sunlight Foundation. You could then cross-reference that list with campaign donations and see whether big contributors are getting HUD contracts.

Then there’s the risk of oversharing. In 2007, the private site FedSpending.org published old records from the United States Department of Agriculture that included Social Security numbers and names. When the USDA found out, it removed the data from the Web. Similar snafus are possible under the new initiative. But despite the rhetoric about changing the culture of governance, chances are agencies will err on the side of nondisclosure rather than overdisclosure.

Some skeptics argue that transparency and efficiency are inversely related—that the time an agency spends prepping data could be better spent actually administrating. (Lawrence Lessig, meanwhile, argues that it makes us see skeeziness where skeeziness doesn’t exist.) The Obama administration rejects that framework. Quite the opposite, it says: Transparency and efficiency go hand in hand. The data are already there. They just need to be better organized. Making them public makes better organization possible—both by the agency and by academics, businesses, and activists. And better information means greater ability to figure out what works and what doesn’t. The tricky part will be getting agencies to comply. At least we’ll all have the chance to watch.