A French wave. (REUTERS/Stephane Mahe)

Bernie Sanders fans, like the fans of other surging outsider candidates for every election cycle since polls existed, promise any outside observers that the level of enthusiasm for their candidate means that voters young and old will swarm caucus sites on Monday, delivering the Vermont senator a staggering victory which will be the first domino in his march to the White House. It's possible! Surges from previously apathetic voters do happen, and sometimes that results in a victory. But far more often, in campaigns at every level of government, the promised surge never materializes and the candidate loses. It's the activist's version of "don't trust the polls"; we know how that works out.

We will use this video of a giant wave sweeping in on a beach as a metaphor. It can be awfully hard to know if such a wave -- from the ocean or in politics -- is imminent and if some candidate is about to get washed out to sea. (Is that the right analogy? I don't know. It's a good video is all.) We look at polls and examine other metrics in an attempt to make predictions, but we could still be wrong.

Take voter registration in Iowa. The New York Times took a look at how registration looks in the state so far this year, since a huge surge of voter registration for one party over the other would generally mean that more people are preparing to go caucus. They found that registration overall was climbing at the same rate as in 2012, not as in 2008, when then Senator Obama saw a wave of support nationally.

We can break that down by party, thanks to data from the Iowa secretary of state. Registration increases relative to three months before the caucus are indeed modest. In 2004, the biggest jump was among those not affiliated with a party. In 2008 it was more Democrats; in 2012, more Republicans. This year it's not much of anyone, really.

But there's a big grain of salt here: People can register at the caucus site right before they participate. In Iowa, there were big surges in registration as recorded in the beginning of the month after the caucus, because the state is then tallying everyone added that month. We can't see that yet.

A weirder metric, mentioned on Twitter on Wednesday by FiveThirtyEight's Nate Silver: Google searches for "caucus" in Iowa are at about the point we saw the month prior to the 2008 decision and well ahead of 2012.

But we'd expect that, since both Democrats and Republicans care about this year's outcome, unlike in 2012. Who knows!

So what about polls? In the new NBC/Wall Street Journal/Marist poll we discussed earlier on Thursday, the pollsters asked people to state how strongly they felt about top candidates. Here's what people said about Donald Trump, Ted Cruz, Hillary Clinton and Sanders in the first three states to vote, Iowa, New Hampshire and South Carolina.

Trump has much stronger support than Cruz in all three states, and his supporters are less likely to say they're willing to change their minds. On the Democratic side, though, both candidates are about even, suggesting that support for Sanders isn't really that much stronger than support for Clinton.

In the Democratic caucus process, voters have a chance to argue, to duke it out over candidates. More Sanders backers say they might be willing to change their minds in that state than do Clinton backers, but by a narrow margin. Should we read into this that Clinton has an advantage in the state? Should we dismiss the claims that Sanders will surf his way to victory?

A better metric -- maybe the best metric -- from that new poll is this one. The pollsters separated out respondents that were known to be registered voters in the states. And among those voters, voters with a demonstrated track record of having voted, Clinton is the only one that sees more support than she sees overall. In other words, a larger percentage of Clinton's support comes from people who have voted before, compared to Sanders -- and compared to Trump and Cruz, who are also running as outsider candidates.

This is important because there is no better predictor of who will vote than people who have voted before. The reason that so many promised insurrections fall flat is that they necessarily depend on people who don't vote as much, and people who don't vote as much don't vote as much. It's right there in the description. Hillary Clinton has a big advantage among voters who do vote, which is a good advantage to have.

That said, it is possible that Sanders (and Trump) are those sneaky waves that it's hard for us to see coming. (Beyond the anecdotal shouting on social media or people crowding small arenas.) We'll know on Monday.