Remember this graphic I posted yesterday?

Getting the numbers there drove me a little mad. Allow me to explain.

On paper, "how much does Obama cut Medicare by?" should be a simple question to ask. The government was going to spend amount X on Medicare, and instead under the Obama budget they spent Y. Subtract Y from X and you get the amount of the cut. It's just arithmetic.

Sure enough, getting Y is simple. Just open up the budget's summary tables and go to table S-5. There you see numbers for how much is spent from 2014 to 2023 on each category of the budget: Social Security, defense, non-defense discretionary, and, yes, Medicare. Over 10 years, Obama's budget spends $6.668 trillion on Medicare. So far so good.

The trouble comes when you try to get X. The Congressional Budget Office put out a baseline in February detailing how much it expects the government to tax and spend over the next 10 years. That should get us our X, right? Nope. The problem is that the CBO relies on different economic projections than the Office of Management and Budget, which compiles the president's budget request.

For example, the OMB expects the unemployment rate in 2015 to be 6.7 percent. The CBO expects it to be 7.1 percent. The OMB expects GDP that year to be $18.2 trillion. The CBO expects it to be $17.9 trillion. Those aren't hugely different numbers, but they're different enough that you can't compare budget numbers prepared according to one set of assumptions to those prepared according to another set.

That means that if you want to compare the Obama budget to a budget compiled using the CBO's economic assumptions, such as Paul Ryan's budget or the Congressional Progressive Caucus's, you need to use different baselines for each. But surely the OMB puts out a baseline against which to compare Obama, right? Sure it does. Table S-4 in the summary tables.

Problem is that it's an "adjusted baseline." It doesn't project spending based on the policies currently enacted, but based on policies that it expects will be enacted. For example, the adjusted baseline assumes a "doc fix" will be passed, preventing cuts to Medicare provider payments. The CBO baseline doesn't assume that. So if you want to compare Obama to Ryan, say, you can't just compare Ryan to a CBO baseline and then Obama to an OMB baseline. You need to amend the OMB baseline so it's predicting the same policies that the CBO budget baseline does.

elmendorf_zients CBO Director Doug Elmendorf, left, and OMB Director Jeff Zients use different baselines. Why? (Chip Somodevilla/Getty Images and Larry Downing/Reuters)

Does the OMB amend it for you? Of course not. That would be too convenient. Instead, it provides table S-9, which explains how to "bridge" the OMB baseline to a more conventional baseline. But table S-9 doesn't explain what exact categories are getting changed. You have to make an educated guess of what adjustment goes where.

That's what we did when we put out yesterday's comparison graphic. And we added a note explaining we did this. Commenters were understandably confused. In an ideal world, everyone would use the same economic assumptions. For that matter, they'd use the exact same categories. But budget world is not an ideal world.

Paul Ryan breaks mandatory spending into Social Security, Medicare, Medicaid and other health, and other. Obama breaks it into Social Security, Medicare, Medicaid, and other. The best you can do to compare is to merge Medicaid into "other," but that loses information about how much each person is cutting in health care as opposed to, say, unemployment insurance.

All of which is to say is that seemingly simple questions about the budget, like "Does Obama or Ryan cut the Medicare budget more over 10 years?" are rather hard to answer honestly. And that's leaving out dumb mistakes that get made when you're trying to get the data out fast. For example, when I was entering numbers for Social Security yesterday, I took care to subtract out the $130 billion in cuts due to chained CPI. Only I cut them from the baseline, not the actual program spending. So what is actually a cut looked like a spending increase. Derek Thompson alerted me to the issue and it got fixed, but the damage was done.

That was a stupid mistake on my part, and I should have done better. But policymakers should do better, too. The reason we have institutions like the CBO is so we can have a common set of assumptions about what policies are going to take effect, what the economy is going to look like and so forth, to enable easier comparisons between policies. So it's mildly infuriating when institutions like the OMB work at cross-purposes and make it harder for citizens to understand exactly which politicians want to cut what from what.

Partly, I selfishly want to spend less time trying to reconcile summary tables so I can get the relevant numbers out quickly. Partly I want to spare people like Marc Goldwein and Adam Rosenberg at the Committee for a Responsible Federal Budget and Shai Akabas and Loren Adler at the Bipartisan Policy Center the frantic e-mails I send them whenever budgets come out, trying to make sure our math checks out, that we're comparing apples to apples and not apples to oranges (Marc gets credit for pointing out the economic assumptions problem to me).

But mostly I think it shouldn't be this hard to compare basic statements of public policy. The OMB shouldn't use an adjusted baseline, the House and Senate budget committees shouldn't construct their own baselines, and they certainly shouldn't use different economic assumptions. It doesn't have to be this hard.