Did those controversial electronic voting machines properly record Americans" votes yesterday? We"ll never really know. Without a paper trail, there"s simply nothing to check against in order to verify accuracy.

For a vendor, it"s great work if you can get it: Install systems where your users have no way to measure achieving the desired results. For our democracy, though, it"s unacceptable.

Once again, we ended Election Day with an Electoral College cliffhanger in a highly polarized nation, complete with a tight state election that decided the outcome (this time Ohio replaced Florida). And ominously, there"s already talk around the Internet from disgruntled Democrats that exit polls differed substantially from reported results in both Ohio and Florida.

The difference between exit polling and final reported numbers was an early warning signal of the fiasco to come in Florida 2000. Whatever one thinks the outcome of a full recount in Florida would have been four years ago, the exit poll gap certainly highlighted the fact that some Palm Beach County voters thought they voted for Al Gore (and would have told exit pollers that) but actually voted for Pat Buchanan, thanks to a confusing "butterfly ballot."

It"s that history which fuels suspicions that there may be something more than bad exit polling methodology to explain this year"s exit poll/vote result differential in Florida and Ohio.

Whichever presidential candidate you prefer, it"s simply not good for the nation when voters lose confidence in the system. Democracy can only work when all sides -- especially those who lose -- have faith that votes are counted accurately. The inability to verify votes could prove poisonous next time in a nearly evenly divided, highly polarized election.

Accuracy and security aside, though, Election Day 2004 highlighted another critical flaw in electronic voting systems: They don"t scale.

Despite countless reports that turnout would be heavy this year, many polling places were simply overwhelmed. Some people waited three or four hours or more to cast their votes. Let that sink in: People spent half a working day standing in line in order to take five minutes selecting the candidates they prefer. That"s absurd!

Unless you assume a deliberate attempt to discourage voter turnout by making voting as inconvenient as possible, the problem was lack of resources. At least one judge tried to get lines moving by ordering poll workers at e-voting sites to provide voters with alternate paper ballots, but state officials balked. Yet as one harried Ohio official said, e-voting systems aren"t like a loaf of bread: You can"t simply run to the store and buy another one.

In theory, though, they are like Web servers. You can buy enough for peak demand, have some in reserve and put them online during those times when you need to. In practice, though, cash-strapped state and local governments won"t shell out money for peak-load systems.

It"s inexcusable to commit to a system that can"t properly function at peak times. We"re not talking about impatient online holiday shoppers here. We"re talking about the foundation of our democracy.

Mechanical voting machines with levers have the same scalability problems, of course. But paper ballots read by optical scanners do not.

Massachusetts had a record turnout yesterday, and voting was quite heavy in my precinct. Yet the parking lot at my polling place was worse than the line for voting; we use paper ballots, and I was in and out in less than 10 minutes. No doubt it took workers longer to count ballots after voting ended because of the volume, but citizens were able to exercise their right to vote quickly, easily and painlessly.

If a paper-ballot polling site becomes swamped with voters, officials can call in extra volunteers, set up additional tables and pass out ballots down the line. It"s a lot tougher to add machines -- mechanical or e-voting -- that don"t exist. And with hard-copy ballots, if anyone questions the results, votes are easily recounted.

Sometimes, simpler is better. Especially when you can prove that it works.