Toby Dorsey has posted this draft on SSRN (forthcoming, University of Richmond Law Review:

When the Supreme Court holds that part of an act is unconstitutional and must be stricken, what does it do with the rest? In almost every case, there is an implicit assumption — shared by every party and every Justice — that the rest of the act should stand, preserving as much of the functionality of the act as possible. In politically charged cases, however, the Court applies this functional assumption to most of the act, but then isolates a few provisions and applies a political preference test to those provisions — preserving those provisions only if Congress would have wanted them preserved. The Court’s political preference test is a test that has not been applied consistently and cannot be applied with any certainty. Accordingly, it may be unconstitutionally vague. Regardless, it is unseemly and unreliable. It is also unwise, because the test as formulated by the Court is so open-ended — its consequences so potentially broad — that every case not only allows, but actually invites, an argument that the Court should strike down an entire act. This leaves the entire statute book fraught with uncertainty, because the Court has left open the possibility in every case that it may strike far more broadly than the Constitution requires if there is a political justification for doing so. The Court should discard the political preference test. If part of an act is unconstitutional, the Court should strike that part, and leave the rest alone.