The Post has an op-ed by Max Stier arguing it's more important to get good people in government than to worry about government reorganization, because most mergers/reorganizations fail. He's got a point. I think I've noted the ASCS/FmHA reorg in 1994 still hasn't erased all past lines. But...
Back in the day, Harry Truman thought it was nonsense for us to have 3 air forces (Army, Navy, Marines), two armies (Army and Marines), etc. so he was pushing for one armed service. Of course he got shot down. For 30 years or so the Joint Chiefs were rather powerless. In 1986 Goldwater and someone else got reform legislation passed, essentially saying to the four services--if you want to hit the top ranks, you've got to spend time on the Joint Chief staff. (All my details are suspect, but the general idea is right.) That apparently has, over time, improved the coordination among the services.
The 9-11 Commission noted the divide between the intelligence and law enforcement communities, which their recommendations hoped to redress. The divide reminds me of the divides among the services.
My point is leadership needs a long range perspective. In the short term, Mr. Stier is right--focus on the people, not the organization. But for the long term it's important how you're structured, more so than who the people are. For example, look at GM. It was formed by the combination of different companies (i.e., Chevrolet, Buick, Cadillac, etc.), most of which became divisions of the company, each with its own dealerships, supply chains, etc. Although GM worked toward consolidation, in the long run that organization wasn't able to compete with companies like Toyota, with just two lines. Certainly the organization wasn't the only problem, but it was a big part of it. There were good people in GM (the company, UAW, dealers, etc.), but they were handicapped by the organization.