With Iowa set to vote tomorrow, after years of anticipation, the nation will finally start the process of selecting our next president. Until now, all we’ve had to go on are the opinion polls. But the question that should be on our minds isn’t just which candidates will win the first-in-the-nation caucus but whether the polls that have been the focus of our fascination ever since President Obama was re-elected will prove to be mistaken again. After a steady series of drubbings in recent elections both here and abroad, 2016 will be a crucial test for our poll-driven political world. If, as has happened so often, Iowa voters don’t reaffirm the final verdicts of the opinion surveys that have given Donald Trump the mantle of frontrunner, the public will have a right to ask why the media and politicians take them so seriously.

Let’s start this discussion by acknowledging the obvious about any discussion of polling. Those candidates that are trailing always say the only poll that counts is the one on Election Day. They’ll point to the size of their crowds or voter enthusiasm as proof that the polls are wrong. But those are generally just excuses that will soon be followed by concession statements and then, in presidential campaigns, by an announcement about them “suspending” operations, a euphemism that will allow them to keep raising money to pay off their debts even though they’ve thrown in the towel.

Moreover, when polls show a consistent pattern and the average of polls (supplied by the essential Real Clear Politics website) point in one direction, then sensible people are right to think that they can’t all be wrong.

But let’s also remember some other things about Iowa, in particular, and polling in general.

The first is that the caucus system is not quite the same thing as a primary, let alone a general election. Predicting whether people will both turn out and stick around for what could be a tedious evening of politics is not as easy as asking whether they will just show up to vote. Voter enthusiasm is crucial to winning in Iowa, as is organization on the ground. In the case of Donald Trump, polls show that his voters are the most enthusiastic but he has little in the way of a ground game. It’s impossible to predict whether the former will overcome the latter.

In the past few cycles, there have also always been surprise winners in Iowa. Though some polls were able to show a late surge for particular candidates, like Mike Huckabee in 2008 or Rick Santorum in 2012, voters there and in New Hampshire seem to have taken a particular pleasure in proving polls wrong and there is a theoretical chance that could happen again this year.

But the questions about polls these days aren’t so much about the vagaries of the voters in some states or the potential strengths of the candidates but a crisis in the industry. At their best, polls only provide a snapshot of the moment and can be rendered obsolete by any late-break development. But the problems go deeper than that. The fact that increasingly few Americans can be reached via landlines has altered the polling landscape. Fewer voters want to take phone surveys and many others may not be telling them the truth about their preferences for all sorts of complicated reasons, especially when they may vote for candidates that are unpopular with the media. Nor can turning to the obvious alternative — the Internet — help because online surveys aren’t random and are inherently unreliable. Moreover, the small size of the samples used in national polls and even in some state polls undermines their credibility.

These factors may explain the series of stunning polling debacles in recent years. Pollsters underestimated the size of Barack Obama’s victory in 2012 and that of the Republicans two years later when they swept Congress in a midterm landslide. Their extent of their failure in the Kentucky governor’s race last November when Republican Matt Bevin won easily rather than being in a tossup his Democratic opponent might win was enormous. Polls also failed to predict the Conservative victory in Britain last year as well as Prime Minister Benjamin Netanyahu’s decisive win in the Israeli elections last March. Pollsters have had their success too but the size and the importance of their misses must inject an element of skepticism into the mix.

So it is entirely possible that much of the data on which the media, the public and the politicians themselves, have been relying could be garbage.

The point is, over the course of the next nine days all the assumptions we’ve been operating under in the past few months about Trump and his rivals as well as the Democrats may be overturned. Or not. But the pollsters should be sweating out the results as much as the candidates. If they turn it out to be wrong, then perhaps it would be wise to remember in future election cycles that media-driven candidacies that are ratified by unreliable polls are not sure things. Many Republicans may be looking for a savior to prevent Trump from hijacking their party. But pollsters that have come up with data that show him winning in Iowa and possibly sweeping the table in the primaries that follow may seem him as the man whose triumph will break their losing streak and ensure that we will spend the next few years chewing over polls that may or may not be accurate.