Mon, 31 Mar 2008

Campaign Reform 2008

While it lacks the timeliness of those pundits who make daily or weekly entries, this Ironwriters site does have many of the same characteristics as other web blogs. And so, in that spirit, I offer some takes on this year's campaigns for President, specifically, on some modest proposals for improving the party primary election systems.

The first remarkable thing that strikes me about his year's campaigns is that they did manage to take place within this calendar year. The conventional wisdom prior to this year was that the earliest caucuses and primaries in Iowa and New Hampshire tended to give too much emphasis (in terms of media coverage and political momentum, not delagates) to these relatively small and allegedly under-representative states, so many other state parties engaged in games of political chicken to see just how early in the process they could schedule their primary/caucus in the hopes that that would make them more "relevant" to the process. The ridiculousness of the whole process reached its apex when the New Hampshire legislature passed a bill instructing the Secretary of State to ensure that their primary was always the earliest on the electoral calendar, even if that meant scheduling the vote in 2007! Fortunately, that turned out not to be necessary, although the election year was barely a week old when Iowa and New Hampshire voted this time around. Michigan followed a week later, with a primary that lacked the imprimatur of both national parties because the state had advanced its date so far forward. (The sanctions imposed by the Republican party turned out to have little effect on the later results, but the Democratic party's penalty of stripping Michigan of its delegates to the national convention may yet turn out to play a pivotal role in the race -- more on that later.)

Shortly thereafter, the so-called "Super Tuesday" primaries were held -- far earlier than usual. Traditionally a large slate of states would hold votes some time in March; due to the urgency to more to the front of the calendar, this year's Super Tuesday vote was held in 24 states (three times the usual number) on February 5th. While the number of delegates at stake (as a percentage of the total needed) was roughly the same for both parties, the impact of the results could not have been more different.

The Republican party primaries were structured as winner-take-all contests, with whoever achieved a majority (or even a plurality) of the vote being awarded all the delegates at stake. Coming off an upset victory in New Hampshire, John McCain managed to surge into a virtually insurmountable lead for the Republican nomination after the Super Tuesday voting. While winning nine states to Mitt Romney's seven and Mike Huckabee's five, McCain captured nearly twice as many delegates as Romney and Huckabee combined, thanks primarily to victories in large states like New York and California. While not yet official, McCain had effectively assured himself of the nomination; Romney and Huckabee dropped out of the race soon thereafter.

The Democratic picture was at once simpler and yet also more complicated. The party's primaries were not winner-take-all, but instead awarded delegates on a roughly proportional basis, roughly approximating the candidate's percentage of the popular vote. Even after the winnowing out of the Democratic field to the two front-runners, Hillary Clinton and Barack Obama, the primaries to date have failed to give either candidate the majority needed for the nomination. Instead, the contest has turned into a slugfest, with Clinton and Obama trading "victories" back and forth, but both sides capturing delegates in approximately equal shares. Like two exhausted boxers holding on for one round after another, baring a series of landslide victories for one side or the other, the future seems to hold more of the same, even going so far as to make it likely that neither candidate will be able to achieve a clear majority of the delegates before the nominating convention in late August.

Now the Democratic nomination process is seemingly fraught with peril: Obama has won more contests, but Clinton has (albeit narrowly) carried larger states like New York, California, and Texas. Obama has accumulated more delegates, but Clinton (supposedly) has greater support among the so-called "superdelegates" (essentially senior state and national party officials) whose votes at the convention may decide the outcome. At one point Obama won eleven contests in a row, but Clinton finally stopped that momentum, and now looks to do well in the upcoming contests in the half-dozen or so remaining elections in April and May.

What has ended up happening is that the Democrats have ended up with two candidates who appeal to different segments of the party's base, but in roughly equal numbers. Combined with the proportional awarding of delegates, it's no wonder that after the initial sound and fury, the overall result will probably turn out to be a rough and inconclusive tie.

It's not clear to me how the Democratic nomination struggle will play itself out. It would be absurd for Obama to consider dropping out, given his lead in delegates and number of primaries and caucuses won. Clinton will no doubt feel increasing pressure to drop out "in the interests of the party", but she gives no indication of doing so -- the Clinton's (Hillary and Bill) political organization is probably second only to that of the Kennedys in terms of skills, savvy, and out-and-out ruthlessness. As long as she is not mathematically eliminated, and as long as she can make use of her political capital as a Senator and former First Lady, Clinton will likely stay in the race.

I am too young to remember when the outcomes of political conventions could be decided by politicos working in the "smoke filled rooms" off of the convention floor, but the Democrats may find themselves in just such a situation when they meet in Denver this August. Similarly, I have only vague recollections of normally arcane things like which delegates and delegations would be seated ever being significant -- mostly having to do with civil rights issues and agendas. But since the convention gets to decide its own rules about which delegates can be admitted, the currently in absentia delegations elected in the "renegade" Michigan and Florida primaries -- both contests won by Clinton -- could turn out to be critically important for her, both in absolute terms of the number of delegates, but also in playing her trump cards of "experience" and "electability" compared to Obama. For those who enjoy political theater, the prospect of the first "wide open" convention in more than a generation is something to be gleefully anticipated. For those whose priority is ensuring a Democratic victory in the general election in November, the prospect of a Clinton/Obama knife fight can only bode well for the eventual election of John McCain, the Democratic vote hopelessly fragmented and split.

So how did the Democrats get themselves into this impasse? Mostly, I think, it was an outgrowth of the "democratizing" of the party's nominating process starting in the 1970s. Before that, most states chose their delegates at state conventions managed by party leaders; after that, there was an enormous growth in the number of states choosing their delegates in primaries or caucuses. This was thought to make the process more democratic (small "d") by giving the average voter a greater voice in the selection of the candidate. The other great democratizing change that originated in this period was the notion of eliminating winner-take-all votes and awarding delegates in rough proportions to the popular vote. For decades, the Democrats managed to avoid a protracted primary struggle, and a consensus candidate generally emerged before the convention. But this year, as mentioned, two candidates emerged from the early primaries, neither able to deliver a knockout blow to the other.

Could this stalemate have been avoided?

One wonders if a winner-take-all system like the Republicans use would have affected the outcome. Similar to the Republican field this year, a winner-takes-all system would have probably resulted in a winnowing of the Democratic field faster than eventually happened. And it's impossible to predict the impact of Obama's victory in the Iowa caucuses followed by Clinton's "comeback" in New Hampshire under such a scheme. I suspect things would have played out much as they did anyway -- the delegate totals would probably be different, but Clinton and Obama would probably have wound up stalemated much like they are now.

There is so much tied up in the traditional mentality of "first-past-the-post" that little consideration seems to have been given to other approaches to the nominating process. Perhaps the Democratic standoff of 2008 will serve as the catalyst for a more rational analysis of the nominating process. One place to begin is looking at alternatives to the actual "choose one" voting system currently used.

The problem with "choose one" voting schemes is that they don't really allow people to express preferences or even degrees of preference. In a field of four candidates, I may be equally willing to support Candidates A or B, would prefer either of those to C, and would definitely not support D. But the best I can do is to try to choose one of A or B. Tactical considerations may also now enter my mental calculations: for example, I may vote for C, my least preferred candidate, if I suspect neither A nor B has a chance to win and I want to prevent the election of D. The numbers of such permutations and considerations are virtually endless; it's no wonder people feel that fixing voting systems is a warren of virtually infinite complexity.

One possible solution would be to hold a runoff election for the top two vote-getters if no candidate achieves a clear majority. Runoff systems are popular in many countries and in a number of state and local elections in the U.S. While not immune from tactical voting considerations (such as voting for a weaker candidate in the first round so they survive to oppose the preferred, stronger candidate in the runoff), runoff systems do ensure that the eventual winner does have the support of an absolute majority of voters, and in a country where "majority rules", that is no small thing. They are also familiar and involve no real changes in well understood voting methods.

The biggest drawback of runoff systems involves the amount of time and expense involved in holding two (or more) rounds of voting. It can result in lower turnout (aptly named "voter fatigue") in the runoff rounds. And a large field of candidates can sometimes result in odd effects, such as the large number of left-wing candidates for French President in 2002 who split the votes of the left among themselves, resulting in two right-wing candidates contesting the runoff election. But runoff voting schemes have much to recommend them, and such schemes would not be impossible to use in party primaries -- perhaps in conjunction with a system of regional primaries: a round of regional voting could be followed by the runoff vote a week or two later. Each regional primary could be completed in a month or so, and the whole primary process would only take about four or five months. (More on that later.)

Another potential solution involves the use of Condorcet voting schemes (named for the eighteenth century mathematician Marie Jean Antoine Nicolas Caritat, the Marquis de Condorcet). Condorcet systems are those that select the so-called Condorcet winner, the candidate that would defeat each of the other candidates in a head-to-head election. The typical Condorcet system involves the voter ranking the candidates in order of preference. Votes are then tallied by first listing all the possible pairs of candidates, then examining each ballot to see how the voters rank each member of each pair; the higher-ranked candidate on each ballot receives one vote in that matchup's tally. Once all possible pairings have been tallied, if one candidate beats all the others they are declared the winner.

Unfortunately, Condorcet systems by themselves do not ensure a single Condorcet winner. Often Condorcet voting schemes include a mechanism for resolving these situations. Another consequence of Condorcet voting schemes is that it is possible for a candidate to emerge as the Condorcet winner without being the first preference of any voter. In that sense, Condorcet systems can produce a "least objectionable" winner, usually when the first choices are spread among many candidates, but a reasonable consensus exists on the second- or third-ranked choice. (It's this aspect of Condorcet systems that make them appealing in party primary elections.) Finally, there is an interesting mathematical possibility in Condorcet voting systems, where no clear winner is possible -- the election result can be cyclic (more people prefer A to B, and B to C, but they also prefer C to A) even though each voter expressed a specific, non-cyclical order of preference. Again, in these cases, non-Condorcet voting methods (for example, a runoff) must be used to resolve the outcome. Finally, there is the possibility of an ordinary tie: two candidates tie with each other but each defeats the remaining candidates. Ties can be resolved again with runoffs, or with more traditional tie-breakers like tossing a coin.

Condorcet voting schemes are relatively simple for voters to use. Traditional ballots listing all the candidates would be relatively easy to modify to allow the entry of a ranking instead of a single selection. (This is also a scheme to which computerized voting machines lend themselves more readily.) But the counting is far more complicated than in simple "choose one" schemes, and in large-scale elections like party primaries, some form of automated or computerized tabulation would almost certainly be needed -- since local results don't necessarily aggregate up to higher levels, there can only be a single tabulation at the highest level.

Finally, a voting scheme that is often useful at winnowing a large number of candidates to a manageable subset is called Cumulative voting. In a typical Cumulative scheme, each voter has as many votes as there are candidates. The voter can then allocate those votes across the candidates as they wish, even to the point of casting all their votes for one candidate. Cumulative voting is roughly analogous to the candidate ranking used in Condorcet systems, except that Condorcet schemes involve an absolute preference ranking (ordinality) hwereas Cumulative schemes allow the voter to express a relative indifference to two or more candidates by allocating them an equal number of their votes.

Cumulative voting systems are relatively straightforward for voters to understand, but, like Condorcet schemes, are more complicated to tally; there is also a higher likelihood of voter mistakes by accidentally overvoting, especially if the number of candidates is large (again, computerized voting machines can be useful here). Because Cumulative systems by their nature are designed to promote proportional representation, they could be used as a first-round tally to reduce a large field of candidates to a more manageable number for a subsequent runoff using another voting system.

So here's how the party primary election system should be revised.

Instead of the current rush to the front of the electoral calendar, we should start by leaving Iowa and New Hampshire their traditional roles as first primary and first caucus. Tradition does have some value, and, by themselves, that's more than enough campaigning for the month of January -- and besides, the number of delegates at stake in these two states is pretty small, and almost certainly won't affect the overall outcome of the nominating process. The biggest outcomes will be media attention and declarations of "momentum" to the winners, whoever they are.

The real fun starts in February. Instead of individual state-by-state primaries, the country would be divided into four geographic regions (say, Northeast, Southeast, Northwest, and Southwest) roughly equal in population, and each region would be alloted one fourth of the available delegates.

Each regional primary campaign would last one month, and the order in which the regional votes would be held would be rotated from one region to the next with each election cycle. The initial vote would use a Cumulative voting system; if there were no clear majority vote-getter, the top three candidates based on their Cumulative results would engage in a runoff election two weeks later. Then the next region would take its turn for another four week/two week cycle, and so on, covering the time from February through July. Hold the conventions in August, and kick off the general election campaign on Labor Day. (Reforming the general election campaign is a subject for another article.)

One plus for this scheme is that no candidate need drop out even if they do not do well in one regional contest, as the wider geographic areas involved and the smaller number of "campaigns" needed could encourage candidates to stay in the race. Rather than winnowing the field early on so that votes in later states don't have the opportunity to vote for a Fred Thompson or a Bill Richardson, the slate of candidates would be "full" each time, and a poorer showing in one region could be made up for in another. There would also be a lesser likelihood of candidates trying to focus on the largest states, since a valid strategy could involve carrying voters living in the smaller states of the region, since their votes count just as much as those of voters in the larger states.

Second, this system would eliminate the quadrennial scrambles to reach the front of the primary calendar, as there would be no advantage in "going first" -- each region would be given that opportunity once each electoral cycle; individual states would be unable to leapfrog one another. (And, as noted, Iowa and New Hampshire could keep their traditional "first in the nation" status.)

Finally, a regional primary system would shift the focus of the candidates and the voters from narrow, parochial, state-based interests to a broader, regional focus. After all, does New Jersey really have many unique and different issues that New York doesn't have? Is Virginia all that different from West Virginia? North Dakota from South Dakota? (In this case, definitely not: after all, North Dakota's new license plate motto is, "A Lot Like South Dakota".) Granted, what's important to a resident of Down East Maine is probably different from those issues of concern to a resident of Downtown Baltimore, but overall, I expect that there are probably more similarities than differences.

The notion of "think globally, act locally" applies just as well to large regions of the United States as it does to the Earth as a whole. It wouldn't be a bad mindset for voters to begin to adopt as we undertake the initial steps in electing the next President of the United States.

Posted Mar 31, 2008 at 03:31 UTC, 3166 words,  [/rich]  Permalink