A new voting heuristic?

I'm thinking of testing a new voting heuristic based on Daryl Levinson & Richard Pildes, Separation of Parties, Not Powers, 119 Harv. L. Rev. 2311 (2006), which builds on earlier work suggesting that the political parties have essentially obliterated the traditional notion of separation of branches.  We do not have political branches that compete for power, as Madison envisioned, due to party loyalty.  Thus, Congress' response to Hamdan v. Rumsfeld, 126 S. Ct. 2749 (2006), was to pass the Military Commission Act of 2006, which gave the President about 90% (estimate) of what he was asking for.  This occurred because, Levinson & Pildes would suggest, Congress was controlled by the same party as the President.

Thus, my new voting heuristic to be tested is to vote for divided government, not because it leads to gridlock, which is good; but because it functionally replicates the original separation of powers with branches checking one another.

In the off-year elections, this is an easy rule to apply: vote for the congressional candidate (and senate candidate, if a Senate seat is up for grabs) who belongs to the opposite party from the President.

In the Presidential election years, this is more difficult to apply.  Perhaps Congress will be firmly in the hands of one party, in which case the rule dictates which Presidential candidate to vote for.  Otherwise, one might have to look to polls and see if there is a clear expected winner of the White House and vote in the congressional races acccordingly.

Of course, this can be disconcerting -- after all, what if the heuristic leads me to vote against a Presidential candidate that I actually like?  Well, I have free will, so I guess I would vote for the candidate I liked (if there is such a person).  But the implications of doing so, assuming that Levinson & Pildes make a persuasive case, are interesting.  After Hamdan, the Court was roundly congratulated for upholding the principles of separation of powers and for ensuring that branches would check one another.  To the extent that one agrees with that view, and to the extent one agrees with Levinson & Pildes, should we be voting for divided government?

It doesn't seem to be an adequate response to say that, "Yes, but I would like to have my party in control of both political branches, because we would do the right thing."  After all, that's just a question of disagreement about results, not structure.  No doubt the Republicans believe that they are doing the "right thing" currently.  The separation of powers criticism is structural in that it argues that decisions reached when the branches aren't checking each other are suspect.  That structural defect is present no matter which party is in control of the political branches.

Thoughts?  It seems like a crazy idea, but perhaps it's just one of those Tragedy of the Commons situations where we would all be better off doing this, but can't trust the "other side" to do the same thing, so we end up collectively making things worse.  (Of course, belonging to neither party makes it easier for me to contemplate this heuristic.)