Giving Democracy the Old College Try

IS THE ELECTORAL COLLEGE (EC) democratic?
Yes, insofar as it is a means by
which the declared preferences of the
voting public are translated into selection
of an elected representative. If the
crux of democracy is that policies or
representatives who make policies are
chosen by a broad electorate in free
elections, the EC clears the bar.
The E.C. is nonetheless a target of much criticism, and
there are at least four prongs to the attack that it is insufficiently
democratic. First, the public’s role is indirect: we,
the people, do not actually choose presidents, but only
slates of anonymous ciphers to whom the actual job of
choosing a president falls. Second, the complex design
wherein the presidency is won in a vote by 538 electors,
themselves chosen in 51 elections (in 50 states plus DC)
confers unequal voting power on American citizens.
Third, the EC system is not guaranteed to be decisive, and
tie-breaking procedures are even more indirect, taking
place in the US House (for President) and US Senate (for
Vice President). Fourth, the 2000 election reminded us
that our current system allows a candidate who won fewer
votes than a rival to secure the presidency.
The first point is true, but its importance is unclear.
People almost universally talk as though they are voting
directly for presidential candidates. Only pedants say, “I’m
voting for the slate of electors pledge to choose Obama
(McCain).” Commentators, pollsters, pundits, academics,
and probably even electors talk about candidates winning
vote and states. Shouldn’t we be bothered, nonetheless, by
these anachronistic middlemen? The gravest danger, from
the democrat’s point of view, is “faithless” electors, individuals
pledged to back a given candidate who surprise the
nation by voting otherwise. There have been 158 faithless
electors, but only 9 (out of 8,048) in the last 60 years. In
the hyper-close 2000 race, Bush beat Gore by 5, rather
than 4, electoral votes because a Democrat from DC
abstained, a symbolic protest she would surely have foregone
had the electoral vote been tied or had Gore won by
one. In 2004, one Democratic Minnesota elector voted for
John Edwards for president, apparently by mistake.
Optimists note that those chosen to be electors are generally
party loyalists, and that the very rare divergences
from pledged votes have not mattered in modern contests.
No one seriously argues that today’s electors should be
accorded discretion to vote as they like, without regard to
their state’s popular vote tallies. Pessimists fume that
cabals and bungles are possible as long as electors are
human, and that the very existence of the electors is an
affront. On balance, though, they seem more a curiosity
than a threat. Some states constrain them to vote as directed
with legislation, and a more radical reform would be to
automate the College so that electors are tabulation
devices, not humans (such a change probably could not
pass constitutional muster absent an amendment).
On the second point, the appeal for votes to “count
equally” is mostly illusory. True, there are substantial discrepancies
in number-of voters-per-elector across states: in
2004, values ranged from about 75,000 in DC to nearly
300,000 in Wisconsin. Such variation arises from multiple
sources, including: large turnout differences; a bias favoring
small states in the EC, due to every state being apportioned
one elector per Senator; apportionment of House
seats (and, thus, electors) never matching population
shares exactly, since the House is fixed at 435 members
whose districts cannot cross states lines; and the fact that
apportionments are adjusted only once per decade, even
though populations shift constantly. But electoral rules that
accommodate some malapportionment of this sort are
common elsewhere, and were typical in the US before
Reynolds v. Sims and related cases of the mid-1960s. Moreover,
computing “power” for individual voters is more
complicated still. Realistically, it depends on the size of the
state voting electorate, the closeness of the state contest,
and tricky permutations involved in constructing all possible
winning coalitions (combinations of states). DC is the
most over-represented presidential-election unit in the simple
count above, but it is also lopsidedly Democratic, and
DC voters are the least powerful by some calculations.
In any case, in large-scale elections, all votes are exceedingly
unlikely to matter, in the sense of making or breaking
a tie. A rational cost-benefit-oriented voter expecting
even a few thousand others to turnout would never bother
to cast a ballot. Voting is largely an expressive activity: we
vote from a sense of duty, because we were asked to do so
and would feel guilty about not following through, or
because we enjoy the sensation of being part of a movement.
A voter who thinks her ballot will be decisive is kidding
herself, even if she lives in a small, evenly split state
like New Mexico or New Hampshire.
The third complaint is more worrisome. It has been
184 years since the House chose a president, but foes of
the EC like to highlight the near-misses, elections in which
the EC could have failed to pick a winner had a few thousand
voters chosen differently. There is little doubt that
most Americans would be aghast to see a presidential election
resolved by the US House; but it is hard to know just
how alarming are these counterfactual histories.
The elite-level tie-breaking procedure of the EC is unattractive,
but non-resolution is possible even in a national
plurality election. An exact tie in popular vote is not necessary
for deadlock: if a result is close enough for a recount, a
battle distinct from the initial contest ensues, over how to
deal with the inherent messiness in large-scale elections that
is usually safely out-of-sight. Democrats will recall Florida
in 2000 with rage, and the US Supreme Court’s role in the
resolution. But recounts almost always turn up messes. In
2004, for example, Republicans in Washington saw a series
of recounts marred by irregularities (e.g. the late appearance
of new ballots, somehow overlooked in earlier tabulations)
turn a win by their gubernatorial candidate into a loss.
It is thus well to remember that the Electoral College is
not uniquely prone to chaos. In a national direct election,
if the margin were sufficiently close, there would be no
limit to the domain of the conflict: we could see Floridastyle
recounts and court fights in 50 states (3,000 counties).
Granted, we’ve had few presidential elections with
extremely close national vote totals, and simple mathematics
ensures that very, very close totals are much more likely
in individual states than in the national sums. Still, the
“recounts everywhere” scenario, though quite unlikely, is a
serious worry on par with those counterfactuals wherein
the House might have had to choose the winner if a few
states had voted differently.
“But Gore won more votes!” It is arbitrary, rather than
non-democratic, to employ an electoral system that does
not necessarily select the candidate who won the most
votes. When both candidates know in advance how the
election will be determined, there is nothing undemocratic
about not being majoritarian. Gore wasn’t even the only
modern VP to be foiled by the EC: Nixon outpolled
Kennedy in popular votes while losing the presidency in
1960 (a point obscured by most textbooks, which assign
to Kennedy votes cast for electors who openly opposed
him and cast their ballots for Harry Flood Byrd).
Ultimately, there is no such thing as a perfect, error-free
electoral system. Specialists have proven complicated theorems
establishing that all voting rules are prone to some
manner of manipulation. The Electoral College is quirky,
creaky, and can fail. But that’s also true of democracy, alas.