One reason Americans dislike Washington is that its denizens work so hard to obscure what they do. So it’s welcome news that someone wants to bring more transparency to the numbers and methods that are supposedly the basis for policy-making.

The American Enterprise Institute will soon unveil its Open Source Policy Center in an effort to crack the codes used by government bean counters. The think tank’s goal is to produce open-source economic modeling to give outside academics, experts and average Americans the tools to test, check and improve the hidden calculations that government uses to design policy. This is wonky stuff, and therefore it won’t make the cable TV shows, but it is an essential step toward holding accountable the increasingly powerful administrative state.

As it stands, federal agencies base their policy judgments on assumptions and statistical models that are rarely disclosed. These can then be used to promote, or doom, a tax bill, spending proposal or regulation. The numbers elite jealously guard their methodology so no one can question their findings.

When the Joint Committee on Taxation “scores” the revenue impact of a tax bill, for instance, Members of Congress, staffers and the general public see only the final estimates. Even Members of the tax-writing House Ways and Means Committee aren’t privy to the methods behind the calculations. Ditto the scores that flow from the Office of Tax Analysis at the Treasury or Congressional Budget Office.

We recall once asking visitors from CBO how they arrived at their estimates that higher government spending always increases GDP and job creation. The answer was the models. Which models? The Keynesian economic models that assume that $1 of spending produces some “multiplier” of higher GDP. And, no, CBO doesn’t share those assumptions with mere journalists.

Republicans have made progress by requiring that CBO and Joint Tax provide a “dynamic” revenue estimate so that their analyses look at how tax-policy changes affect real-world behavior and economic growth. But static scores are still used officially for legislation, and the methods the bureaucracies use to calculate the economic impact of tax cuts remain opaque.

Do they distribute the cost of the corporate income tax to the consumers who actually pay it? Do they assume that a tax is borne more by capital or labor? The current opacity helps those on the left who claim that dynamic scoring is rigged to help the rich.

The methods used by the executive branch to estimate the costs and benefits of new regulations are also hidden. The Obama Administration has turned this into a black-box art, routinely underestimating the cost of environmental, labor and financial mandates. Veterans of the ObamaCare wars will recall how then budget director Peter Orszag estimated that the new entitlement would somehow bend the “cost curve” of health spending.

Then there’s the Administration’s fictional estimate of the “social cost of carbon,” which it claims measures the damage per ton of carbon-dioxide emissions. The Environmental Protection Agency’s cost estimate is huge, and it serves as the basis for justifying as a benefit essentially any carbon rule the Administration wants to impose. Congress has tried to examine the assumptions behind EPA’s estimate but has met resistance at every turn.

The bureaucratic defense is that secrecy is essential to deter lobbyists who might influence the government’s analysis. But as everyone who wasn’t born last night understands about Washington, lobbyists have a way of influencing the bureaucracy anyway. Depending on who’s in power, the bureaucracy may favor the Sierra Club over the American Petroleum Institute. But rest assured the agencies listen to lobbyists.

Matthew Jensen, the managing director of AEI’s new open-source center, notes that the better way to guard against influence is “to make the entire process transparent, so that everybody knows what is behind the numbers.” He adds that while the government often employs good modelers, “there is a universe of cutting-edge computer scientists and economists at universities across the country who could be lending their expertise to help make their government better.”

The AEI center will test this with its initial Web application, Tax Brain, that will allow the public and experts to study the effects of individual and payroll tax policy using open-source simulation models. As with open-source software, the goal is to attract a crowd of like-minded tech and policy wonks to develop new tools or refine old tools for analyzing tax, spending and regulatory policies. The hope is that over time this will move policy makers to a common methodology—or if not, then at least everyone can see what the difference of opinion is about.

Federal agencies are grabbing ever-more power over ever-more of the economy, and most of the media find policy details too boring to cover. Someone has to hold the bureaucracy accountable, and AEI’s new effort is a welcome attempt to do so.