From the comments section at Ohio Voter Suppression News:
I was in Youngstown on Election Day and I visited a number of polling places with Election Protection. The Dep't of Elections was woefully unprepared for this election. Long lines caused waits which were at least 2 hours long during the midday period, traditionally the slowest period. They simply did not have enough machines. The ones they had often went down or mis-registered votes. People pressed Kerry and the confirmation screen showed Bush. Many people didn't realize there was a confirmation screen so who knows how many votes were lost that way. The polls were understaffed with workers (mostly older folk) who had little or no familiarity with computers or the election rules. There were Reupublican challengers at many polling places but they did not challenge many voters - they didn't need to. The voters were suppressed by the voting infrastructure which made it a great chore to vote.
While this election had the heaviest voter turnout since 1968, nationwide, we are really only talking about a 60% turnout instead of the 54% turnout in 2000, an increase of only 11% above the rate of growth of the population. Now admittedly, this increase was probably not uniformly distributed.
But consider this: the wait times for voting have got to figure into existing computer models of voting behavior. What these long wait times mean is not that vast legions of new voters came out of the woodwork, say, doubling the number of voters, but rather mostly that those voters who turned out in places where the infrastrucure is poor were much more reluctant to be discouraged from voting: they refused to leave. That a moderate increase in voter turnout can produce wait-times measured in hours means that there are some areas where voting is habitually discouraged.
This has got to figure into partisan strategies for influencing election infrastructure. The math is just too easy for it not to. X number of minutes to vote times Y number of voters times Z voting machines produces the total time necessary for a given numberof people to vote at a polling place. From there, you can calculate wait times. And from there, it's not too hard to plug in factors like how long the average voter is willing to wait. If no voter is willing to wait more than, say, 15 minutes, you can have relatively short lines. If the voter is willing to wait an hour, the lines get longer. If a significant percentage of voters are unwilling to leave without voting, you get the kind of catastrophic waits seen in Ohio. All of this depends on a willingness to exclude voters from participation. What I see from the wait times in this election is a process that relies on it.
Also, there was an awareness the age of the poll workers combined with th increasing complexity of the voting process would pose a problem. See this USA Today article from August:
Panel cites poll workers' age as problem
The biggest threat to November's presidential election is not balky voting machines or a terrorist attack, but the potential for confusion and mistakes by the nation's aging corps of 1.5 million precinct poll workers, federal election officials say.
The current corps of poll workers is well short of the 2 million needed for a national election. The average age of a U.S. poll worker is 72, according to the U.S. Election Assistance Commission.
"If they don't get it right, someone could be denied their opportunity to vote," says Paul DeGregorio, one of four members of the commission. The panel was created by Congress in 2002 to make sure federal elections run smoothly. Already this year, problems have cropped up during primary elections, he notes.
In addition, the Help America Vote Act of 2002 has imposed new procedures for elections, including ID requirements for first-time voters who registered by mail and provisional ballots for voters not immediately found on voter rolls.
"There's a growing complexity at the polling place," says DeForest Soaries, the commission chairman. "If all we do in November is what we did in 2000, that's going to be a problem."
Here is the Election Assistance Commission's Best Practices Tool Kit.
In July, the Caltech-MIT voting Technology Project issued a report, Immediate Steps to Avoid Lost Votes in the 2004 Presidential Election: Recommendations for the Election Assistance Commission (pdf; html), which contains this interesting sentence:
According to the U.S. Census in 2000, approximately one million registered voters said that they did not vote because the polling place lines were too long or the hours for voting were too short.