ATM may be model for voting machine

Caltech and MIT, two of the nation's most prestigious scientific institutions, vow to develop a better way to tabulate election results

A country that can accurately dispense money from automated teller machines

ought to be able to reliably count votes, said Charles Vest, president of

the Massachusetts Institute of Technology.

Vest and David Baltimore, president of the California Institute of Technology,

vow to merge the brainpower of two of the nation's most prestigious scientific

institutions to develop a better way to tabulate election results.

Last month's vote count was "embarrassing," said Baltimore, who appeared

with Vest at a video news conference Dec. 14 to unveil a project to develop

an "easy-to-use, reliable, affordable and secure United States voting machine."

Disputes and lawsuits over voter confusion and voting machines that

undercounted or overcounted votes led to a five-week delay in the election

of Texas Gov. George W. Bush as the next president. "We must find a solution,"

Vest said. "Each of us must be confident that his or her vote has been reliably

recorded and counted. A country that has put a man on the moon and an ATM

machine on every corner has no excuse."

In fact, the ATM is likely to serve as a model for the new voting machine.

"Most people have been able to figure out ATMs," Baltimore said.

Currently, the most common voting machines in polling places across

the nation are punch card machines and optical scanners. But both are plagued

with high error rates.

In Florida last month, voting machines that used optical scanners failed

to count about three of every 1,000 votes. Punch card machines failed to

count 15 out of 1,000 votes, according to Vest and Baltimore. The voting

machine envisioned by MIT and Caltech will be designed to minimize the possibility

of such confusion.

It should also:

* Indicate to the voter that his or her vote was recorded.

* Decrease to near zero the possibility of miscounting.

* Be tamper-resistant to minimize fraud.

"A solution is highly likely," Vest said, but it took "the truly bizarre

circumstances of the recent presidential election put it on the front burner."

The first phase of the project, which involves evaluating existing voting

technology, is expected to take six months. Replacing the nation's voting

machines could cost several billion dollars.