Friday, October 08, 2004

Electoral Triage

A Politics Blog post (linked by Eugene) on the potential for an electoral debacle in Colorado reminded me of something I’ve been thinking for a while: the Electoral College system, while created for reasons abstrusely related to federalism, has an ancillary benefit that was not (at least to my knowledge) imagined by the framers of the constitution. Specifically, it minimizes ordeals attributable to voting problems such as miscounts, missing ballots, ballot-stuffing, and so on.

This might seem an odd claim in light of Florida 2000. But notice that the problem was limited to that one state. It’s unlikely that Florida was the only state in which voting problems occurred; indeed, I seem to recall some news reports about one or two other states that might require recounts. But Florida was the only state for which both (a) the size of the margin of error from voting anomalies exceeded the margin of victory for the state, and (b) the number of electors at stake could have altered the outcome in the Electoral College.

Imagine if presidential elections were decided by popular vote. In that case, recounts (and the associated meshugas) would be demanded in every county that had any voting anomalies, so long as the magnitude of all potential errors exceeded the national margin of victory. Suppose, for instance, that California voted Democratic by a 60-40 margin, with in-state errors equivalent to a potential 2% of the vote, or a 4% swing. Under the Electoral College, those errors could safely be ignored, because California’s electoral votes would go to the Democrat regardless. Under a popular vote, those errors would suddenly become relevant.

Allocating electoral votes on a proportional rather than winner-take-all basis, as Colorado might choose to do, would tend to increase the number of states for which the margin of error mattered, since a small margin of error could still correspond to an elector or two. But demanding recounts would still only be worthwhile if the number of electors at stake could affect the national result.

In short, the Electoral College creates a modular voting structure that insulates the system against large-scale impacts from small-scale errors. Errors within modules (that is, states) need only be examined and corrected when the errors are sufficiently large and the module is sufficiently important to the system-level outcome.


Anonymous said...

Its unclear whether there is a cost or a benefit. Insensitivity to error goes hand-in-hand with insensitivity to actual changes. Yeah, the fact that the system produces the same result if there is a 2% margin of error in California as if there is a 0% margin means that error doesn't cause a need for a recount. But it also means if there is an *actual* 2% shift in CA sentiment, it also has no effect on the election.

It insulates the system against errors by making many people's votes irrelevant.

-Patri Friedman

Anonymous said...

Yes, a proportional allocation of electorates on a state by state basis would increase the number of states where the margin of error mattered. However, if we abandonded the presidential form of the executive branch and instead replaced it with a proportional representative form of government (such as is found in Germany with the chancellorship and in France as well) would be better because then instead of having 50 states to divide the populace, we might have 435 (the number of districts) instead. It seems to me that PR government would be much more representative as well as causing less chances for an electoral debacle like we saw in 2000.