I would like to address three themes that run through these illuminating
and challenging comments. First, should election administration and
technology rank high on the list of political reforms in the United
States? Second, does the system of election administration in the United
States disproportionately affect racial groups and lower income Americans,
and if so how can that be addressed? Third, do we want to keep our decentralized
system of election administration?
Before November 2000, improving our methods of casting and counting
votes was not a high priority, except perhaps among local election administrators.
For years, progressive political reformers have pushed for proportional
representation (as Robert Richie does here) and other broad systemic
changes, such as those expounded in Joshua Rosenkranz's, Marc Strassman's,
and Heather Gerken's commentaries. The problems that election administrators
deal with seem much more mundane. But perhaps they are no less important.
Spain, Italy, and Japan have switched electoral systems in recent years
without profound political transformations. Brazil has invested in more
reliable methods of counting votes, and has improved greatly the public's
confidence in its elections.
To have confidence in election outcomes, every democracy needs an accurate
and secure method of casting and counting votes. The voting technologies
used in the United States were inadequate in four states—Florida,
Iowa, New Mexico, and Wisconsin, where the number of spoiled and uncounted
ballots exceeded the vote margins. Nor is this problem new. Massachusetts
rid itself of punch cards like those used in Florida following a challenged
election in 1996 that strongly resembled the controversy in Florida.
But the challenge we face is not about past elections. The proposals
of the Caltech/MIT group and many others are about the future. The way
we vote is changing. New computing and communication technologies and
the demand for greater convenience drive those changes. Internet voting
(as Strassman and Cornfield indicate) is here, with the promise of convenient
and more accessible voting, especially for many voters with physical
disabilities. Still, secure Internet voting appears to be a ways off.
The focus of new voting technologies will differ from past innovations.
Past innovations have been driven by the demand among election administrators
for a faster count. Further improvements in speed are marginal at best.
Today, the questions before policy makers and engineers are about the
experience of voting. How can we guide the development of voting technology
to make voting easy and obvious, to remove barriers to voting, to give
people confidence that their own votes are counted and that elections
are conducted fairly, accurately, and securely?
Race and Class
One possible justification for greater national control is the guarantee
of voting rights. It is unclear how equal protection, as it is embodied
in Bush v. Gore and the Voting Rights Act, applies to voting
technology. The courts will have the opportunity to clarify these matters
in cases now unfolding, such as Common Cause v. Jones in California.
Several respondents (Smith, Rosenkranz, and Gerken) raise the prospect
that inequalities in access to the polls and the counting of votes arose
because of race and income. But is there evidence of systematic racial
discrimination in election administration?
I looked at two pieces of evidence and was surprised to find that the
answer is likely no. First, consider voting technology. The residual
vote rate of a county does depend on the percent of non-white residents
in the county. However, the racial composition of the county does not
magnify or interact with the extent to which ballots are uncounted or
spoiled. This suggests that race is unlikely to be a factor in the operation
of voting equipment. Second, consider registration. The Current Population
Survey asked registered voters in 2000 why they did not vote. Most express
some form of disinterest in the election. A surprisingly large number
(7.4 percent) state that problems with registration prevented them from
voting, as might occur with an incorrect purge of the registration roll.
This does not, however, depend on race. Seven percent of black respondents
said that they could not vote because of registration problems, compared
with 7.5 percent of white respondents.
This does not mean that problems do not exist in some areas, as investigations
such as the House Government Reform Committee study have found, or that
other factors, such as literacy, pose systematic barriers, as Richard
Posner has argued. But the problems encountered in the 2000 election
are not problems experienced only by specific groups. They are experienced
Election administration is perhaps the most localized public service
in America. Federal involvement in voting technology is minimal—a
set of voluntary standards under the aegis of the Federal Election Commission
but run by the National Association of State Election Directors. The
federal government pays for none of the election administration in the
U.S.. One may view the problems in Florida as a great example of the
failings of federalism. The federal government has left this responsibility
to the states, and most states have pushed it off on the counties and
municipalities. That's the case against decentralization, and Dan Ortiz,
Rogers Smith, and Donna Brazile spell out the case quite convincingly.
Nonetheless, I believe that U.S. elections are decentralized for a
simple reason: Americans like to have a lot of elections. We prefer
democratic control over administrative appointments wherever possible.
On one election day, a voter in California may cast more votes than
a voter in the United Kingdom casts in a lifetime. Almost all of these
votes are for state and local offices and propositions.
It would not be easy to centralize local elections, because local governments
represent many different jurisdictions. A typical county conducts elections
for county offices, multiple city governments, and special districts
such as water districts. Ballots must be tailored to each local circumstance,
and I would argue against Smith that it is indeed "hard to add offices
to ballots for different locales." Los Angeles county alone must format
about 5,000 different ballots to accommodate all of the jurisdictions
and languages within the county. When the county asked five electronic
voting machine manufacturers to prepare their machines for an experiment
with early voting, only one of the manufacturers was able to accommodate.
Even with the existing level of decentralization, many technologies
are approaching their limits. Cook County, Illinois, for example, had
to increase the number of chads on its punch cards by 30 percent in
order to accommodate all of the judicial elections in that locale. The
rate of uncounted and spoiled ballots doubled when the county moved
to these "high density" punch cards. The problems faced in formatting
ballots actually argue for further decentralization, down to the municipal
level, or for holding separate elections for each unique jurisdiction.
Canada, for example, has separate national and local elections.
But there is a tradeoff. By holding many elections at one time, we
benefit from economies of scale. Having separate elections for each
jurisdiction in the U.S. would require voting at least ten times during
an election year—including primary and general elections for city,
county, special district, state, and federal offices. Turnout is already
low voting twice a year.
Federal involvement is important, but not in the management of election
day itself. Instead, the national government needs to: (1) set mandatory
standards for the security and performance of equipment; (2) aid poorer
communities to update technology; (3) provide information on the performance
of equipment to help states and counties become better consumers of
voting technology; and (4) guarantee the rights of voters. These recommendations
are eloquently spelled out in the report of the National Commission
on Election Reform and in other reports. Beyond these proposals, however,
let's leave elections to the states.
Stephen Ansolabehere is co-director of the Caltech/MIT Voting
Technology Project, and professor of political science at MIT.
Return to the forum on machine
politics, with Stephen Ansolabehere and respondents.