The news that the Gallup organization won’t be conducting any polls about the upcoming presidential primaries would have been unimaginable only a few years ago. But after their disastrous performance in 2012, the decision is being treated as a verdict on what is not unreasonably seen as the mistakes Gallup made when it failed to predict President Obama’s re-election victory. But Gallup’s memorable and disastrously wrong polls that were conducted in the last weeks of the 2012 race weren’t the only ones that were mistaken in recent years. Time after time, we have seen polls proven to be completely wrong. Just as Gallup got the last presidential race wrong, in both Great Britain and Israel the consensus of pollsters predicted outcomes that were contradicted by results. Exit polls, which are treated by the media as information gold, are also increasingly seen as unreliable indicators of what the voters think. But while presidential candidates are pouring millions into the pockets of those conducting surveys on his behalf and Donald Trump seems to view his current standing in them as proof of his worth, the day of reckoning for all pollsters may not be far off.

The problem facing pollsters is obvious. Changes in American behavior have made it more difficult to assemble a representative sample of public opinion. Many of us no longer use landlines. The rise of cell phones and caller ID, as well as the fact that many younger voters use phones more texting than calling, has created an environment in which traditional polling methods are increasingly obsolete. There is also the question of whether respondents are influenced by what they think is the politically correct or fashionable answer to give which may not dovetail with their actual beliefs or how they will vote. That is surely a factor in understanding why, with the conspicuous exception of Gallup’s belief that Mitt Romney was ahead late in the 2012 race, conservatives tend to be undercounted in horse race polls.

While some pollsters did better than Gallup in 2012, that was not so much a product of superior technique as it was a decision to guess that Democratic turnout in 2012 would match that of Obama’s first election in 2008. But smarter weighting of samples won’t make it possible for pollsters to accurately gauge what voters are thinking. As Politico notes, Gallup isn’t the only pollster scaling back. Horse race polling as opposed to those surveys that dig deeper into opinion about issues seems to be getting less reliable and that’s why Gallup and even as respected an organization as Pew are doing less of it.

At best, polls are snapshots of public opinion in any given moment and can even if they are accurate, be outdated before they are published. But the failures in several major elections of pollsters to predict massive trends such as the conservative surges in the 2010 and 2014 midterms, the big turnout of pro-Obama minorities and women in 2012 or the equally unexpected surges for Conservatives in Britain and the Likud in Israel this year illustrates that the models pollsters are using are deeply flawed.

That’s significant and not just because horse race polling may be coming up with the wrong outcomes. The early months of the GOP presidential race have been dominated by polls. It was the pollsters who established Donald Trump as the overwhelming frontrunner and are at one and the same time, his bona fides as a candidate and proof of his seriousness. Polls have also and continue to be used to decide who is a first-tier candidate and thus worthy of inclusion in the main event at Republican presidential debates. Bad polls are causing candidates to run out of money and drop out even though the margins of error on most of the polls being used are higher than that of the percentage of support for many of the candidates. It’s also not clear just how much, especially at this stage of the race, they are telling us about name recognition and how much about the actual rise or fall of the individual candidates.

All of that is not to assert that polls are meaningless but Gallup’s decision should sober up many pundits who regard them as gospel. Trump may ultimately prevail in the battle for the Republican nomination or he may, as I insisted he would early on, ultimately fail. But those of us who have been figuratively preparing his inaugural address or consigning him to the dustbin of history need to understand that the numbers they are feeding us are still relatively meaningless. Every poll will ultimately be eclipsed by actual vote totals.

Until a next generation of pollsters can come up with better methods that those now being employed, there is no reason to regard them as anything more than a sidebar to serious political reporting. Gallup may be the first of the major pollster to wave the white flag when it comes to predicting elections, but it won’t be the last.