A recent NYT op-ed by Mark Taylor, which opens by declaring graduate education to be "the Detroit of higher learning," has been generating lots of comment in the blogosphere (see here, just for starters). Taylor suggests, among other things, abolishing permanent departments (even at the undergrad level) and replacing them with "problem-focused programs." These two measures, however, need not be connected: you can create problem-focused programs without getting rid of departments. Indeed, I suspect many universities are doing just that.
It is well to remember that, as a group of social scientists wrote more than a decade ago, "the U.S. has had a long history of structural experimentation in [its] university systems" (I. Wallerstein et al., Open the Social Sciences , p.99). Perhaps it is time for a new round of such experimentation, given economic pressures, dwindling resources, overproduction of doctorates in many fields, and increasing (and increasingly exploitative) use of adjunct and other temporary instructors. What form such experimentation should take is the key question, one to which yours truly has neither the time nor expertise to propose an answer at the moment.