Just how wrong does someone have to be to lose their job in Washington, D.C.? For John McLaughlin, even historically wrong wasn’t enough.

If the name doesn’t ring a bell, maybe you know his work. He’s the Republican pollster who predicted just weeks before the June election that then-House Majority Leader Eric Cantor (R-Va.) would win his primary by 34 points. This week, Cantor packed up his Capitol Hill office, having lost that election by 10 points.

“The worst part of it is, you build a relationship with a longtime friend and you never want to lose,” McLaughlin says now, noting that he has not seen Cantor — a client for almost 15 years — since before that fateful night. “That’s a loss that you never get over.”

That might be especially true if you were a young Republican congressman climbing the ranks of leadership, as Cantor was. For McLaughlin, the anguish may be real, the embarrassment may keep him up at night, but his employment status? That hasn’t changed.

“We got attacked right after it,” he says, adding that he didn’t lose any of his current clients running in the 2014 cycle. “There was a feeding frenzy of people calling up all our clients asking if they would continue to use us. But they stuck by us.”

Eric Cantor (R-Va.) was expected to win his congressional primary by 34 points, but he lost to a newcomer by 10 points and has vacated Capitol Hill. (Mandel Ngan/AFP/Getty Images)

To be fair, the Cantor race was a historic upset, one that caught everyone by surprise. Not everyone, however, had been paid more than $70,000 to see it coming. In theory, pollsters should be some of the most accountable people in politics. You can see exactly how right or wrong they are with each campaign. And yet . . .

“I’m trying to think of anybody put to such desperate shame that they never worked again, but I can’t think of it,” says Edward Lazarus, a former Democratic pollster. “People were fired from campaigns, but I can’t think of it ever ending their career.”

It may not be shocking to hear that behind-the-scenes D.C. isn’t a pure meritocracy. It’s the land of second opportunity. So what if Vice President Dick Cheney miscalculated the situation in Iraq last time? It didn’t stop him from co-writing a recent op-ed in the Wall Street Journal about the situation there this summer. Karl Rove will appear on television until the end of time despite his election-night kerfuffle with his own network about whether President Obama had won Ohio. Get into the upper echelon of official Washington, and it’s hard to completely fall out of favor.

McLaughlin gets it right a lot more often than he is wrong. But he’s had a bad string of it lately. In 2012, he had former senator George Allen (R-Va.) up 47 percent to 44 percent just weeks before he lost his Senate race by six points to former Virginia governor Timothy M. Kaine (D), and he had Republican presidential nominee Mitt Romney beating Obama in Virginia by seven points, only for him to lose by four. As for the Cantor race, McLaughlin says that he was blindsided by people who don’t normally vote in Republican primaries (including Democrats) and that no one could have seen it coming. The explanation will probably be debated for a long time, but meanwhile the job offers will probably keep coming in.

“To err is human, but in the political consulting world much relies on personal relationships, not just performance,” says Alan Secrest, a former Democratic pollster. “It’s like being an NBA coach — you’re never completely out of it.”

For a good example, look no further than Secrest himself, a man famous for being on the receiving end of a dead fish from Rahm Emanuel, now the mayor of Chicago. At the time, Emanuel was working at the Democratic Congressional Campaign Committee. By some accounts, Emanuel blamed the loss of a high-profile race partially on Secrest’s polling work and sent the putrefying package with an accompanying note: “It’s been awful working with you. Love, Rahm.”

But was it a death knell for the pollster, who worked closely with the DCCC? Of course not. He worked for two more decades on hundreds of campaigns before becoming the executive director of the Tennessee Democratic Party in January of this year.

“It had no negative impact whatsoever in terms of our marketing,” he said about the fish incident, noting that in his opinion it had nothing to do with the accuracy of the polling and was the symptom of a soured working relationship. “We had lots of name recognition at the time, and it added to that.”

A pollster is one of those jobs — like a football lineman or an oil-tanker captain — that normal people tend to notice only when one of these specialized professionals messes up. In that sense, 2012 was a banner year for Republican pollsters. Romney may have lost handily in his quest to become president, but he famously thought he was going to win right up until the last minute.

A lot of that blame fell upon a polling firm called Public Opinion Strategies. Neil Newhouse, who acted as Romney’s top pollster, still doesn’t like to talk about 2012. It’s too fresh. He would rather talk about the work he does now, of which there is still plenty.

“We changed everything, from how we sample to the screeners we use,” he says.

So why is it that in a time when there is an increasing appetite for data (at least based on how many new data-driven Web sites have sprung up in the past year) that there can be so much bad polling? Part of the reason: It’s getting much more expensive.

The number of people reachable only by cellphone has shot up to about 4 in 10 adults. Interviewing people on cellphones is more costly, and many pollsters can cut expenses by compromising on the quality of their methods. Many new pollsters rely on automated voice recordings to interview people or turn to panels of professional poll-takers on the Internet. With the industry at a turning point, it’s possible that by 2020 phone surveys may not be the primary means for figuring out how the public thinks.

This is one of the cases for why pollsters should keep their jobs after getting things wrong: They are the most likely to be able to learn from those mistakes. Newhouse has twice been named the pollster of the year by the American Association of Political Consultants, most recently for the work he did on Scott Brown’s winning 2010 Senate campaign in Massachusetts. McLaughlin got to keep his jobs because his clients presumably liked the work he was doing for them.

But even some partisan pollsters say there isn’t enough oversight in the business. “Some pollsters are just too busy,” says Kellyanne Conway, a Republican pollster who has been at it for nearly three decades. “If you’re going to traffic in volume, there are going to be screw-ups.”

There’s some bipartisan agreement on this.

“The level of accountability in our industry is lower than it should be,” said Mark Mellman, a leading Democratic pollster. In 2002, Mellman served as a pollster for the campaign arm of the Senate Democrats. It was a rare midterm in which the party that controlled the White House, in this case the Republicans, gained seats in Congress. Amy Sullivan, in a Washington Monthly feature titled “Fire the Consultants,” said Mellman was “perhaps more than anyone else, the architect of that defeat.” The piece referenced a New Republic article that said Mellman essentially told Democrats what they wanted to hear, not what they needed to know to win.

“Those articles didn’t stop the presidential nominee from hiring me,” Mellman said, referring to the fact that he went on to work for then-Sen. John F. Kerry’s failed presidential run.

In an editorial Mellman wrote for the Hill after Cantor’s loss, he said the key lesson wasn’t that the pollster was wrong but that you can never assume a lead will hold in a primary. “If you want to be accurate, keep polling,” he wrote.

One hundred percent of pollsters, with a margin of error of zero percent, would probably agree.

Peyton Craighill contributed to this report.