Share this:
07/16/2001 07:00:00

Up to 6 million votes lost in 2000 presidential election, Voting Technology Project reveals

Though over 100 million Americans went to the polls on election day 2000, as many as 6 million might just have well have spent the day fishing. Researchers at Caltech and MIT call these "lost votes" and think the number of uncounted votes could easily be cut by more than half in the 2004 election with just three simple reforms.

"This study shows that the voting problem is much worse than we expected," said Caltech president David Baltimore, who initiated the nonpartisan study after the November election debacle.

"It is remarkable that we in America put up with a system where as many as six out of every hundred voters are unable to get their vote counted. Twenty-first-century technology should be able to do much better than this," Baltimore said.

According to the comprehensive Caltech-MIT study, faulty and outdated voting technology together with registration problems were largely to blame for many of the 4-to-6 million votes lost during the 2000 election.

With respect to the votes that simply weren't counted, the researchers found that punch-card methods and some direct recording electronic (DRE) voting machines were especially prone to error. Lever machines, optically scanned, and hand-counted paper ballots were somewhat less likely to result in spoiled or "residual" votes. Optical scanning, moreover, was better than lever machines.

As for voter registration problems, lost votes resulted primarily from inadequate registration data available at the polling places, and the widespread absence of provisional ballot methods to allow people to vote when ambiguities could not be resolved at the voting precinct.


The three most immediate ways to reduce the number of residual votes would be to:

· replace punch cards, lever machines, and some underperforming electronic machines with optical scanning systems;

· make countywide or even statewide voter registration data available at polling places;

· make provisional ballots available.

The first method, it is estimated, would save up to 1.5 million votes in a presidential election, while the second and third would combine to rescue as many as 2 million votes.

"We could bring about these reforms by spending around $3 per registered voter, at a total cost of about $400 million," says Tom Palfrey, a professor of economics and political science who headed the Caltech effort. "We think the price of these reforms is a small price to pay for insurance against a reprise of November 2000."

Approximately half the cost would go toward equipment upgrades, while the remainder would be used to implement improvements at the precinct level, in order to resolve registration problems on the spot. The $400 million would be a 40 percent increase over the money currently spent annually on election administration in the United States.

In addition to these quick fixes, the report identifies five long-run recommendations.

· First, institute a program of federal matching grants for equipment and registration system upgrades, and for polling-place improvement.

· Second, create an information clearinghouse and data-bank for election equipment and system performance, precinct-level election reporting, recounts, and election finance and administration.

· Third, develop a research grant program to field-test new equipment, develop better ballot designs, and analyze data on election system performance.

· Fourth, set more stringent and more uniform standards on performance and testing.

· Fifth, create an election administration agency, independent of the Federal Election Commission. The agency would be an expanded version of the current Office of Election Administration, and would oversee the grants program, serve as an information clearinghouse and databank, set standards for certification and recertification of equipment, and administer research grants.

The report also proposes a new modular voting architecture that could serve as a model for future voting technology. The Caltech-MIT team concludes that this modular architecture offers greater opportunity for innovation in ballot design and security.

Despite the fact that there is strong pressure to develop Internet voting, the team recommends a go-slow approach in that direction. The prospect of fraud and coercion, as well as hacking and service disruption, led the team to recommend a cautious approach to Internet voting. Also, many Americans are still unfamiliar with the technology.

"The Voting Technology Project is part of a larger effort currently underway—involving many dedicated election officials, researchers, and policy makers—to restore confidence in our election system," commented Steve Ansolabehere, a professor of political science who headed up the MIT team. "We are hopeful that the report will become a valuable resource, and that it will help to bring about real change in the near future."

Baltimore and MIT president Charles Vest announced the study on December 15, two days after the outcome of the presidential election was finally resolved. Funded by a $250,000 grant from the Carnegie Corporation, the study was intended to "minimize the possibility of confusion about how to vote, and offer clear verification of what vote is to be recorded," and "decrease to near zero the probability of miscounting votes."

The report is publicly available on the Caltech-MIT Voting Technology Project Website:

Written by Robert Tindol