PolySciFi Blog

Wednesday, October 19, 2005


What makes for a legitimate election?

As part of our continuing discussion of the proposed Georgia identification law, which at least one reader thinks is interesting, Matt asked me the following question:
Do you believe that the more legitimate (net non-fraudulent) votes that are cast, the more legitimate the results of an election are? It's a yes or no. [net non-fraudulent added to reflect Matt's clarifier comment]

While I believe this metric is absurd, I think a full answer to this question requires answering the more philosophical (and less Hannitized) question of "What makes for a legitimate election?"

The 3 F's
In short, I believe any election which satisfies the "3 Fs" is a legitimate election - Fair, Free, and Fraud-free. Towards that end, the more effort that is made to minimize fraud, the more confident I am in the legitimacy of the election.

Clearly, by my definition increasing the number of net non-fraudulent votes (calculated as total number of non-fraudulent votes minus the number of fraudulent votes) permits the possibility of actually decreasing the legitimacy of an election. In fact, net non-fraudulent votes should be a non-starter for everyone just by considering elections where significant fraud occurs.

A problem with net non-fraudulent votes
For example, suppose we have a voting population of 100 people and the population is divided 55-45. Suppose that nine people typically vote but are generally representative of the population so we get a 5-4 result. (Republics are premised on using representative voters who we call representatives, natch.) The will of the people was clearly captured by this election (this is the criterion that I think Matt was really aiming for). Now suppose we held an election where there were 100 non-fraudulent votes and 20 fraudulent votes and all 20 fraudulent votes side with the 45. The outcome of this election is 55-65, and the will of the people has clearly not been captured by this election. However, the net non-fraudulent votes metric has gone from 9 to 80! (It is equally easy to construct an example where it doesn't go above 100 but fraudulent impostor votes result in an election I think we would all deem "illegitimate".)

As I stated above, the net non-fraudulent metric is an absurd metric to use to measure the legitimacy of an election. So I'm assuming I am either misinterpreting this comment, or Matt brain-farted.

Democratization as a metric and alternate election processes

A more plausible metric (which is in line with the way I first interpreted Matt's question before reading his clarifier) is the following:
Assuming no fraud or fraud is held constant, increasing the number of voters results in a more legitimate election.
Formally, we'll call this metric the democratization metric. I also don't think increases in the democratization metric increases the legitimacy of an election.

Further, I am not in favor of designing elections in an attempt to maximize the democratization metric particularly as an attempt to legitimize the election (which I measure by the 3 F's). In fact, I would be perfectly happy with many restrictions on voting. These restrictions range from limiting enfranchisement to only those who own property (I also held this position during the period when I didn't own any property) to limiting the election to only informed citizen voters (perhaps using the Heinlein test as Richard suggested). For that matter, I've even considered as a thought-experiment weighting votes by net taxes (if you like a progressive tax code, this one has interesting ramifications).

Assuming these elections were equally free, fair, and fraud-free, I would view each of these elections as equally legitimate. The reason why I prefer the first two alternate scenarios to current election approaches is because I believe they would lead to better and more responsible government. (Recall that I believe improving society is a time-varying multiobjective multidimensional nonlinear optimization problem so I am comfortable using criteria in addition to legitimacy when determining which is the best election style.)

Problems with the democratization metric
After making the shocking claim that I don't believe that increasing the number of non-fraudulent voters makes for a more legitimate election, let me make the even more shocking claim that you, dear reader, don't believe that increasing the number of non-fraudulent voters makes for a more legitimate election either. Further, we all consider objectives other than legitimacy in our preferences of different election styles.

I suspect that the vast majority (if not all) of polyscifi's readers agree with Matt (and me) that felons should not vote. So the vast majority of our readers believe that disenfranchising at least one class of people still results in a legitimate election. In fact, ya'll probably prefer an outcome where felons don't vote as it keeps people who have demonstrated a willingness to act against the interests of society from voting.

Considering an absurd example, I am certain that all of polyscifi's readers would agree that if we allowed all people in the world but US citizens to vote in the election that would not increase the legitimacy of the election (and in fact would dramatically decrease the legitimacy of the election because of its unfairness). Somewhat less absurdly, I figure that many of our readers feel that not enfranchising resident aliens (both the legal and illegal versions) does not negatively impact the legitimacy of an election.

So clearly increasing the number of votes, even assuming no fraud, does not increase the legitimacy of an election. It does, however, increase the democratic nature of the election, but now we're not discussing how an election should be held and are really discussing what kind of government we would like.

The participation metric
Just as my desire for elections participated in only by informed citizens is motivated by a desire to have a particular kind for government (intelligent, responsible government), a desire for increasing the democratization metric is similar to a desire for a particular kind of government (government with maximized participation) . This can then be formed into the following election metric which I sometimes see bandied about:
Maximizing the participation of those who are governed
An election that maximizes this metric can be said to maximize the legitimacy of the resulting government (everyone impacted by the government has a chance to influence the impact and is to some extent consenting to the government). This metric has the advantage of excluding voting by people who don't live in the US. However, only very few would argue that enfranchising resident aliens would lead to a more legitimate election and very few would view the outcome as desirable. Why is this outcome undesirable? Because the aliens do not have sufficient stake in the government (the same logic is the basis of my property ownership preference). Similarly, felons (who are governed perhaps more than most) remains problematic when using this metric for legitimacy. Likewise, I doubt that any of our readers would view an election that enfranchised children as more legitimate. Again we also wouldn't prefer this result because we believe that children cannot make an intelligent vote (using similar logic, you can see why I prefer that only the informed vote) .

So while it might be useful for furthering some other goal, increasing the number of voters (as evaluated by these three metrics) does not determine the legitimacy of an election.

Of course this discussion only disproves the contention that the number of voters is the sole arbiter of legitimacy and doesn't prove my belief that election legitimacy is determined by how fair (the election of Hussein where he was the only candidate is an example of an illegitimate election because it was unfair), how free (see situations where the "wrong" vote costs you your life), and how little fraud is committed (see Washington State 2004, or if you prefer, i.e., I don't buy it, Florida 2000).

But before I could proceed with any positive proof that the legitimacy of an election is increased by increasing the freedom and fairness of the election and minimizing the fraud, I would insist on us settling on a common definition of a "legitimate election". Because as it stands, any proof would be tautological as my 3F's definition (free, fair, and fraud-free) supplies the metric.

Post Script
In my original draft formulated over dinner at BW3, I had left "free" off my list of criteria for a legitimate election. Thason helpfully saved me from this embarrassing omission.


This page is powered by Blogger. Isn't yours?