Atrios is
upset about "the fact that the media called the election 15 minutes before the polls closed on the Florida panhandle, causing literally millions of panhandle voters to turn their cars around and drive home."
A fair amount of research has purported to show that premature annunciation has no measurable effect on turnout, but why take the risk? If one voter decides not to vote, that's enough for me.
What has happened is that the Voter News Service, which was formed as a consortium of the AP and the major networks to conduct and compile exit polling data, declined in power and eventually fell precipitously as a result of its technological meltdown in the 2002 midterm elections.
Some folks blame the very concept of the VNS for its failure--they say that depending upon a single source for exit polling is dangerous. Having mutiple competing media outlets, they say, magically ensures accurate projections.
But my take is that the decline began in the early nineties when the networks, in a childish effort to beat out their rivals by being first, started developing their own more aggressive models to call elections (using the VNS data).
I find this bizarre.
I mean, does anyone choose a news network based on the speed with which it announces the victor? Do advertisers care? More generally, why can't we just wait until, you know, they actually count the votes? Why does it matter at all? In 2000, we didn't know who the "real winner" was until the Supreme Court decided that partisanship was more important than principle. It was a debacle, but the government kept functioning, Bush was able to pull his transition operation together, and Americans mostly just did their thing.
The embarrassing back-and-forth over Florida in 2000 ought to have been a wake-up call for the networks, but I think it will actually get worse before it gets better.