One of the theories of the 1990's was "flatter is better". Reduce the number of supervisors and creativity will blossom and efficiency with flourish and good things will result. Al Gore adopted that theory as part of his "Reinventing Government" (it may have been part of "business process reengineering" as well, but I'm too lazy to check). So agencies were supposed to reorganize to cut management layers. My understanding of FSA's efforts was that it was mostly a paper exercise; first-level supervisors lost their personnel responsibilities but retained their day-to-day operational responsibilities; units formerly called "sections" became "work groups", branches became "sections", etc.
Of course, over the past 15 years, that particular reform may have gone by the wayside. Certainly the proliferation of titles at the upper levels of FSA and USDA seems to have continued.
Anyhow, Steven Kelman was involved with Gore, mostly on procurement reform as I remember. But now he's got a post in Federal Computer Week in which he seems to say Gore may have been wrong: "There was a period, especially in the 1990s, when the conventional wisdom was that first-line supervisors accomplished little. By contrast, Gittell’s finding is that those supervisors help broker communication across group boundaries."