For political watchers on Twitter on election days, following The New York Times election "needle" is an addiction -- before it's then derided for swinging too wildly or simply being wrong. During the Pennsylvania special congressional election, the needle went down entirely, causing a storm of tweets.
Some background first: In the waning days of the 2016 election, The New York Times' Upshot team used a graphic "needle" to predict the outcome of the race. According to The New York Times, the needle "predicts the outcome of an election based on incoming results, prior election results and demographic data."
At 7:30 p.m. on Election Night, the needle had Hillary Clinton's probability of winning at 85 percent based off pre-election polls. As the probability of her victory dropped throughout the night, her chances went down until eventually, it swung toward Donald Trump having 95 percent chance of winning.
The swings of the needle caused many on Twitter to complain, even though the Upshot had warned it could swing wildly. After President Trump's victory, The New York Times published a story explaining that "data failed us" during the 2016 election.
Despite that, the New York Times brought back the needle for the Georgia and Virginia elections, in much the same form, and the Times noted that it predicted those results accurately. During the Alabama special election in December, Upshot correspondent Nate Cohn predicted early on that Cohn and Josh Katz later admitted in an article.had over a 95 percent chance of victory over Roy Moore. But the Upshot team had another prediction: If Moore did eek out a victory, it would mean the end of the needle,
As all political junkies know, Jones did win in Alabama, and thus, the needle was back again for Pennsylvania's special election. This time, it started out saying that Lamb had a slight chance of victory. Then it swung to Saccone. Then it swung back to even. Political Twitter noticed.
And then it disappeared. In a tweet, Cohn said the Times took it down because Westmoreland county, a GOP-heavy county, would not be providing precinct results, thus making the needle ineffective. In a follow-up, Cohn said the needle would come back when Westmoreland posted its results.
Political Twitter noticed the absence of the needle immediately.
There were Twitter users who were more even-headed about the needle's demise. Decision Desk HQ's Jeff Blehar tweeted that it was "very responsible" to take down the needle.
As the night wore on, the winner of the race was still uncertain but one thing was clear: The needle wasn't coming back. Not for this race, at least.