Ah, remember that election night not so long ago when TV networks predicted Al Gore had won Florida?
That was largely because the networks (ABC, CBS, NBC, CNN, and Fox) and the Associated Press relied too heavily on the Voter News Service, a private consortium that asks voters in select precincts how they cast their ballots and then feeds the tally to its partners in the media who, in turn, predict winners in races.
The VNS method works pretty well in elections that aren't close. And it's tried to overhaul its methods since November 2000, in time for Tuesday's vote. But with so many close races, reflecting a nation that's largely split 50/50 politically, this semiscience of soothsaying deserves a closer look before it rattles the foundation of American democracy again.
One problem is that VNS hasn't finished its overhaul. Its demographic information may not be complete or available to the members on election night. And VNS's new tabulating system hasn't been tested in a real election. Extra rigor in vote predictions is required in tight races, and especially with a rising trend in early voting and mail-in ballots.
The good news: The Associated Press plans a separate count to be made available to all the members if the VNS system fails.
The once-burned TV networks will likely be twice shy in their competitive urges to provide predictions on election night before most Americans start to snooze off. They should also better explain to viewers how they make such predictions and why they're even necessary when the final vote count is usually just hours away.
And they should honor their commitments not to project winners until all the polls are closed in a state. In prime TV no results should always trump bad results.