The notoriety of picking Trump the winner

"Conflicted."

That's the word political scientist Helmut Norpoth used to describe his current lot in life, one that finds his professional credibility intertwined with a politician who appears to have less and less of it: Donald J. Trump .

With mind-boggling confidence, Norpoth's "Primary Model," his heretofore successful election forecast system, projects Trump to win the presidential election in less than two weeks. The model puts the odds of this above 87 percent.

And despite a blizzard of public polls to the contrary, Norpoth isn't hedging his bet. Instead, with invocations of "Dewey Beats Truman," he's calling into question the consensus of survey research that shows Hillary Clinton on the glide path to 270 electoral votes.

Norpoth's contrarianism comes with the perks of publicity.

At the moment, his results are being regularly heralded by an adoring audience on the right.

"I have never gotten anywhere near (this) attention for my other forecasts," Norpoth told CNBC.com in a telephone interview this week. He added that if given the choice between "a forecast that is right and gives you almost no publicity (and) a forecast that turns out to be wrong in the end that gets a lot of notoriety, I would certainly take the latter."

But what is the cost of riding this tiger? For some, very little, but for a respected academic like Norpoth there's certainly a risk. And some of his colleagues are already preparing the post-election woodshed for his comeuppance. In January, Norpoth is scheduled to be part of a panel discussion at the Southern Political Science Association meeting in New Orleans.

"He is going to have some 'splaining to do, as they say," said Alan Abramowitz, a political scientist at Emory University. Like Norpoth, Abramowitz also has stewarded a successful election forecasting model, "Time for Change," which likewise predicts Trump will win the election. But while their projections are similar, their reactions have differed greatly.

At a Labor Day gathering of political scientists in Philadelphia, where both men presented their findings, Abramowitz immediately backed away from his. He has since argued that the unique self-destructiveness of the Republican nominee will likely overwhelm the other variables of his model.

"My forecasting model may be the least significant wreckage created by Trump," Abramowitz told CNBC.com.

In a way, Norpoth and Abramowitz are serving as case studies in a separate kind of social science experiment, one that tests the exigencies of academics, presidential politics and modern media. What is clear now is that nobody is talking about Abramowitz's Trump-picking model, while many are talking about Norpoth's.

This time four years ago, similar buzz surrounded the work of two political scientists at the University of Colorado, Kenneth Bickers and Michael Berry, whose forecasting model projected that Mitt Romney would win the presidency with 330 electoral votes. Their model was one of the few that year to use state-level economic data to project voting booth behavior, which seemed compelling on its face.

However, after the votes were counted, and President Barack Obama won, Bickers acknowledged that the model's assumptions about the domestic economy's role in voter behavior were fundamentally flawed. Although he initially indicated he might try to revise the model for future elections, he ended up scrapping it.

"I think there is a misconception of what forecasting is and why forecasters do it," Berry told CNBC.com. "I received a lot of communication from the public [saying] that I was an operative of the Romney campaign and we were trying to manipulate the results, and that is not what forecasters are doing at all."

"There is tension there that speaks to some of the issues in doing this and how to engage with the public and the media," added Berry. Bickers did not respond for multiple requests for comment.

For his part, while Norpoth is basking in the attention, Abramowitz is playing the expectations game.

"You have to evaluate these sorts of models based on their long-term track record and it is always possible to have an aberrant election," he said.

But Michael Lewis-Beck, a political scientist at the University of Iowa, and one of the godfathers of election forecasting, thinks that Abramowitz's hedge may prove more costly than Norpoth's lunge.

"I was shocked he was so blunt about discarding his model," said Lewis-Beck, whose "Political Economy Model" projects a Clinton win. "(Abramowitz) doesn't hide his talents under a bushel. He tells the world his model is the best, to the point of being pleasantly annoying. ... He had so loudly proclaimed that it was strong science and he is really turning away from the science."

Abramowitz, on the other hand, charges that Lewis-Beck has been far too liberal when it comes to tweaking his model over the years.

"The problem with doing that is it allows you to tweak your model based on what you see in the polling," said Abramowitz, "and you can always come up with something that will produce a forecast that is more in line with the polls. I have stuck with the same model, and it has always worked up until now." (That said, Abramowitz added a "polarization" variable to his model this cycle.)

Though he gives no indication he'll vote for Trump, Norpoth says he nevertheless finds himself "rooting for (Trump) to behave himself and not make stupid mistakes. I can see fully that what he is doing is not helping him and not helping [the forecast]."

If the polls are wrong, and Trump prevails, all glory goes to Norpoth. And if not, he pledges to take his licking at the political science panel in January. It will be duly assessed.

"I will not go into hiding if I fall on my face," Norpoth said. "I will go there with a black eye and a cast."