David Leonhardt goes after the Iowa Caucuses and the New Hampshire Primary in a column today, citing a new article by a couple of economists (which I'm afraid I haven't read yet) that emphasizes the importance of those two states in the nomination process. Jamelle Bouie makes the conventional argument about small states and the opportunity for voters to examine candidates on a human-sized scale; Matt Yglesias suggests solving the problem of anti-urban bias by substituting Massachusetts for New Hampshire.
Let's see...the first thing I'd emphasize is that as the nomination process has evolved, Iowa and New Hamsphire -- indeed, the primaries and caucuses as a whole -- are less important than they once were, and support from party actors more important. Endorsements, money, favorable publicity, and other resources that party actors bring are only minimally state-dependent. Technology has probably also made location less important, as it's a lot easier to be a Romney or Pawlenty volunteer these days even if you live in a state that doesn't vote early.
The second thing that I'd emphasize is that while the frontloaded presidential nomination system and the malapportioned Senate do have a small-state and rural bias, the presidential general election generally has a bias the other way, since the states receiving the most attention are the big, close states: in recent cycles, Florida, Pennsylvania, and Ohio, each of which has plenty of urban and suburban areas.
I'd also stress, with Bouie, that changes in the system should be made carefully. Parties work best -- coalition-building works best -- under conditions of stable, well-known rules and procedures. So I tend to be pretty conservative about major structural reforms. On the other hand, it is certainly true that minor changes to the system have been a constant over the last forty years, and given the limited powers of national formal party organizations, continued constant change is a certainty as well. On the whole, I'm mostly ambivalent about Iowa and New Hampshire...someone has to go first and reap the benefits (a national primary is a terrible idea, so don't start up with that), and I'm not really convinced that a rotation system for those spots changes the fairness of the situation at all. And I do like tradition (although we shouldn't overdo it; Iowa dates back to 1972, and even New Hampshire only goes back 60 years, to 1952, as a significant event). All in all, I don't think changing it is worth the fight. There are other, more pressing process changes worth spending energy on.