Slate's Josh Voorhees explained the holdup in 2 p.m. post on Tuesday:
There have been two issues ... in getting this experiment off the ground. The first is that while VoteCastr’s trackers have been (and continue to be) out in the field today, VoteCastr’s data team ran into technical difficulties that prevented us from presenting that data on Slate. The second is that our data visualizations were designed to show a combination of early-vote and live turnout data. The unexpected delay before we started receiving that live turnout data meant our visualizations showed only early vote totals, which remained largely static this morning.
Voorhees went on to say that readers should expect live projections to pick up as the afternoon goes on. Estimates from Florida appeared first.
Our original post previewing Slate's predictions follows:
On Election Day, Slate and a Silicon Valley startup called VoteCastr plan to break one of journalism’s unwritten rules. Together, they will publish real-time vote projections in seven swing states while polls are still open, using calculations they believe to be even more accurate than exit surveys.
The fear is that an effort to predict presidential election winners in key battlegrounds could suppress turnout in those states. Might a voter in Ohio who planned to cast her ballot after work decide not to bother, upon learning that voting patterns from earlier in the day suggest either Hillary Clinton or Donald Trump is already expected to carry the state?
It’s a thought that veteran political analyst Mark Plotkin called “downright scary and disruptive” in a column in the Hill in September. A belief that too much Election Day information could influence voter behavior is why news outlets typically resist the temptation to project the winner of a state until polls close, even when exit surveys could enable them to make accurate calls earlier.
Slate and VoteCastr will use demographic and turnout data to make educated guesses throughout Election Day about which major-party nominee is likely to win a given state. The projections will work like this: VoteCastr already has conducted large surveys that have allowed it to gauge support for Clinton and Trump on a precinct-by-precinct level in key states.
“The survey data will predict how exactly support in, say, Cleveland’s precinct 14D should break down between Clinton and Trump,” Sasha Issenberg, a former Slate columnist who is now VoteCastr’s editorial director, wrote in September.
That’s the demographic data. Then, on Election Day, the firm will station vote counters at polling stations who will keep track of how many people are casting ballots in each precinct.
“So when a tracker visits Denison Elementary School and reports via a smartphone app that 78 people have voted by 11:15 a.m., VoteCastr can determine how well each candidate is faring,” Issenberg explained. “By blending this information with reports from dozens of other precincts across the state, VoteCastr’s statistical models can predict who is, at that very moment, winning Ohio and by what margin.”
Slate will publish VoteCastr’s predictions, though it will not use them to declare winners.
While Slate and VoteCastr are breaking new ground with their real-time projections, they are reviving a debate that is decades old. The style of Election Day reporting that seems traditional and restrained today was actually highly controversial in the early 1980s, when Congress held hearings in which they pressured television networks to impose blackouts on vote results in Eastern states until polls had closed in other time zones.
On Election Night in 1980, Ronald Reagan’s victories in the East made him the clear winner, even before voting had finished in the West. TV networks declared that Reagan had defeated Jimmy Carter — and Carter had even conceded — while polls were still open in parts of the country.
“There was an uproar in the West ... when NBC News led its two rivals in calling the election for Ronald Reagan at 8:15 p.m. Eastern time, almost three hours before the voting ended in California and other Pacific Coast states,” David Broder wrote in The Washington Post in April 1981. “There were reports of people leaving the lines outside polling places, of drivers on their way to the polls going home instead. There were accusations from losers of close local races that the broadcasts had cost them the election.”
A month after publication of Broder’s article, network executives were called to explain their coverage decisions on Capitol Hill. Here’s how The Post described the proceedings in May 1981:
A Senate committee yesterday opened the first hearings on legislative proposals to limit television networks' projections of presidential election results before polls close in the West, making voting hours uniform across the country, and move presidential elections to Sunday. …[Sen. S. I. Hayakawa] (R-Calif.) has also proposed that election officials be barred from releasing any vote results until all polls are closed around the country. Predictably, news executives from the major television networks strongly opposed any attempt to limit their ability to project election results. They also insisted such projections did not influence the election.William J. Small, president of NBC News, called Hayakawa’s proposal “distressing and potentially dangerous. Broadcast projections have never been demonstrated to have any measurable effect on either voter turnout or voter choice,” he said.“I would like to go on record as expressing the vehement opposition of CBS News to any proposals which would require any news organization to suppress information in its possession, or would deny to us access to any information that would otherwise be available,” said William A. Leonard, president of CBS News.Richard Wald, senior vice president of ABC News, argued that “our free press traditions” and First Amendment freedom of press provisions require Congress to avoid restricting news coverage.
None of the proposals floated in 1981 were ever enacted.
The arguments made by TV executives 35 years ago sound a lot like those presented in 2016 by Slate editor-in-chief Julia Turner, who wrote in September that “the news blackout that usually prevails is premised on the idea that publishing information about voter behavior may depress turnout. But such fears are unsupported by research. Academics examining the question have found no consistent effects on voter behavior.”
Of course, there is no research yet on the effect of exactly what Slate and VoteCastr are doing — because no one has done it before. But in the debate about whether it will suppress turnout and take a toll on democracy, it is worth remembering that we have had a similar debate before — and that what was once deemed irresponsible by some soon became the norm.