Slate and Vice want to shake up the way that news organizations report on election results. So far, no one is quite sure whether that’s a good thing.

With VoteCastr, the two companies are experimenting with offering real-time projections on who is leading the race in over 100 battleground state precincts. It’s a wholesale departure from the traditional approach of news organizations, which have waited until polls have closed before they reported results and called states.

As of 12:30 p.m. EST, VoteCastr estimated that Hillary Clinton leads in five of the six states for which it’s collecting data.

Slate and Vice say that the goal is to break up the monopoly on information that campaign teams and reporters have on turnout information. (Here’s a more detailed explanation of how the data collection and predictions work.)

Since it was announced this summer, the project has been contentious, particularly among poll watchers and political reporters who are concerned that reporting on election results in real-time could suppress voter turnout. The concern goes back to 1980, where a clear early Reagan win was blamed for downballot Democratic losses in Western states. Those claims have never been backed up by hard data, though.

My guide to today:
1. Ignore early projections
-Exit polls can be wrong https://t.co/H6BXVzA5Sr
-Votecastr toohttps://t.co/g4vy5Ilkqz

— Brendan Nyhan (@BrendanNyhan) November 8, 2016

The votecastr show seems a lot like conventional vote-count analysis, only about 8 hours too early, and with lots more swearing.

— Brad Heath (@bradheath) November 8, 2016

The @Slate & @votecastr experiment today is something I don’t really agree with, but I think lots of people will be checking. #ElectionDay

— Matthew Lumby (@MatthewLumby) November 8, 2016

While it’s still too early to say what effect, if any, VoteCastr’s turnout modeling will have on voters, news organizations like Bloomberg haven’t been shy about reporting on Votecastr’s predictions. Markets, too, are moving based on the project’s early numbers.

Clinton has an edge in key states from early voting, Slate’s @votecastr finds https://t.co/ohxy1DuDhR pic.twitter.com/yuaw3QKM10

— Bloomberg (@business) November 8, 2016

.@votecastr @Slate @juliaturner This is irresponsible and looks like a grab for ad traffic revenue. Definitely losing credibility fast.

— OJ Hornung (@JamBottoms) November 8, 2016

Looks like MXN traders just looked at https://t.co/KYH3P0RBdY pic.twitter.com/G8QsHyiok1

— Toby Nangle (@toby_n) November 8, 2016

So did the peso rally on this VoteCastr data? And if so, why?

— Joe Weisenthal (@TheStalwart) November 8, 2016

On the other hand, critics of the project say that the real-time projections only add to the confusion of election day, particularly if Slate and Vice do a poor job of untangling and contextualizing the numbers.

@votecastr Delete your account

— Warriors Fan (@warriorsfannba) November 8, 2016

Others have spotted some errors with the effort’s results and methodology, which is understandable given the experiment nature of the project. In one case, observers noted that VoteCastr is including Jill Stein in its Nevada predictions, despite Stein not appearing on the ballot in that state.

@RalstonReports And these bugs impact media reports, voter participation & potentially, outcomes. @votecastr = blinkered technophilia

— Edward G. Robinson (@O__Robinson) November 8, 2016

Let’s bring on exit polls, add to @votecastr and really confuse people! I suppose we better let them actually count votes. Seems burdensome.

— Jon Ralston (@RalstonReports) November 8, 2016

I know it’s early but after all the whoopla, I’m a little underwhelmed by watching the @votecastr paint dry on @vice and @slate.

— Jared Gilbert (@jaredagilbert) November 8, 2016

Haha “analysis being provided by VoteCastr is built around an unproven technique that hasn’t before been used in campaign forecasting” https://t.co/Q7IH2UVOx5

— Hipster (@Hipster_Trader) November 8, 2016