The Democratic Party should set the first Presidential primaries and caucuses in a sequence most likely to produce a Democratic President in November 2008. The criteria should be: swing states we need to win, representativeness (demographically, geographically), and diversity in size/cost to campaign (to give opportunities to small as well as big campaigns). These criteria give a wide range of candidates an opportunity to succeed, while emphasizing the preferences of the swing voters we most need to win over.
There is no evidence that the current sequence (Iowa Caucus followed by the New Hampshire primary) has helped to elect Democratic Presidents, nor is there some implicit virtue of the fine citizens of those two states that justifies giving them an influence greater than Democratic voters in other states.
Iowa and New Hampshire will resist - however, a Democratic Party able to take on the Republicans needs to have enough backbone to push back against pressure from two states if it is detrimental to the interests of the Democratic Party.
A strong set of potential states for 2008 would be Arizona, Ohio and Florida. Arizona has relatively low cost media markets, Ohio and Florida are key states we need to capture. The combination provides both demographic and geographic diversity. But ultimately the specific states selected are not as important as whether or not they fit the criteria that I have identified above.
Another aspect of states competing with each other for prominence that is detrimental to the Democratic Party is the stretching of the campaign season farther than in any other major country in the world. Why not start this sequence in March or April? People would not complain about a shorter political season, fund-raising money would go farther, and there would be less time for negative attacks on candidates.
Let's optimize our selection process for a Democratic Presidential nominee around what is most likely to bring success.