Idea: Fixing the US primary election system with rotating county primaries

by Alan Cohen

John Tillinghast recently wrote me with a clever idea to fix the US primary election system. Since it’s very much in the spirit of this blog, I’ll present it for your comments. The text is a mix of what he sent me and some broader framing of the problem.

The problem: US presidential primary elections (and caucuses – let’s ignore the difference for now) are held first in Iowa and New Hampshire every year. The ostensible reason for this is to choose two small states that will give the politicians a chance to meet many voters face-to-face and not just saturate major media markets with ads. While the record of these states at choosing the eventual nominees of the parties is mixed, it is certainly true that many candidates drop out at this stage of the race. The problem is that these two small, highly non-representative states thus have quite disproportionate power over a national process. So the challenge is to find a way to preserve the need for politicians to succeed at convincing voters face-to-face to choose them, while removing the disproportionate role of these two states.

The idea: Choose 10 counties from around the country randomly for the first round of primaries, two months before the primary date. These counties could be chosen to represent both urban and rural areas (though in the urban case it might be necessary to have a smaller geographic unit to limit the population), and could have a joint primary on the same day. Counties chosen once would be excluded from being chosen again for the next 20 years. One month later there would be a national primary. This would mean that just two primaries would occur, one month apart, allowing the whole election calendar to be moved back substantially (to, say, June or July). This would have the great effect of delaying the start of campaigning and prolonging the period of actual governing! (Alternative: have three counties vote per month for three months before the national primary.)


Have the initial barrier to entry be more representative.
See how the candidates do when they have to choose strategy quickly.
Not give any small regions a permanent disproportionate influence on
our elections.
More fun to watch early events if the location changes every four years.

How best to choose. I think we would want to have some stratification,
say 1-2 locations in each of the country’s main regions. We might also
want some guarantee that there would be different income levels,
population density, or demographics.
Why we don’t do this already. Big states have tried to get more early
primaries but have been stopped. Is the existing system too entrenched
to change at all, or was this because of worry that CA, NY and TX
would choose the nominees themselves?

Why it won’t happen: The current primary system is not a result of clear national laws, but of a mix of party rules and state law. As we saw in 2008 when Michigan and Florida tried to move up their primary calendars in a way the Democratic party didn’t like, efforts to punish the states were only moderately successful. In other words, in order to change the current system in a coordinated way, all 50 states and both parties would need to agree. Even if one party comes up with a new system, a legislature controlled by the other party in one state could play havoc with the implementation. So the irony is that the current system, once started, can evolve a bit at a time through piecemeal changes, but cannot be completely restructured in a coherent fashion barring a national constitutional amendment.