Brendan Loy has posted this draft on SSRN. Here is the abstract:

This article strikes a blow in defense of America’s much-maligned system of choosing a president: the Electoral College. It argues that, although imperfect, the Electoral College is superior to the alternatives and should be retained because it does an excellent job of balancing the two central purposes of presidential elections: choosing the right winner, and doing so in a reasonably timely and undisputed fashion.
Whereas most literature defending the Electoral College focuses primarily on federalism, this article details the pragmatic and procedural case for maintaining the status quo. A direct national election would be a nightmare in a close race. Recounts and legal challenges would stretch from sea to shining sea, and with two centuries of legal precedent tossed aside, courts would have a very difficult time managing it all. Without the multiple layers of unappreciated procedural safeguards provided by the current system, post-election uncertainty could stretch well into January, raising doubt about whether we would have a clear winner by inauguration day.
No election system can be perfect, because an election is a measurement of the popular will, and all measurements have margins of error. That said, the current system, while imperfect, is pretty darn good. It gets the result right, at least arguably, almost every time — and it always gives us a definitive result every single time, as any presidential election system must. It always gives us a president.
The article’s introduction discusses a hypothetical 2012 election between President Barack Obama and Governor Jeb Bush. In this scenario, the “interstate compact” system, a proposal currently under consideration in various states, has come into effect. This means the Electoral College still exists, but now purports to represent the nationwide popular vote. The scenario briefly outlines some of the dire results this change could have in a very close election.
The remainder of the article is split into five parts. Part I argues that, as mentioned above, presidential elections have two main purposes: choosing the right winner, and doing so in a reasonably timely and undisputed fashion. Parts II, III and IV analyze how well these purposes are served by three different possible election systems: the current Electoral College system, a true direct popular-vote election, and the hybrid interstate-compact proposal. The article contends that the current system is, on balance, superior to either alternative; it more reliably fulfills both of the twin purposes of presidential elections than would either proposed reform. However, acknowledging the possibility that this balance could someday shift, Part V suggests a set of minimum requirements that would be absolutely necessary for a functional popular-vote system, should it ever become necessary to abandon the Electoral College.