There is basically nothing about Iowa’s Democratic caucuses that has a significant parallel elsewhere in the presidential nominating process.

Yes, it’s a selection process like any primary, but one in which caucus-goers are often expected to change their minds — a sort of low-tech instant-runoff-style contest. Sure, other states have caucuses, but none play as significant a role in determining where the race goes. Groups of Iowans commit hours of their time not just to vote but to cajole and argue and celebrate in hundreds of locations across the state.

And this year, figuring out who actually won may well be significantly harder than you think.

How the Iowa caucuses work

Before we explain why, it’s worth a quick review of how the caucuses work. Let’s do so by looking at it from the precinct and the statewide levels. An important note: This is how only the Democratic caucuses work. The Republican contest, in years where it matters, is a lot more traditional.

So imagine a fictional caucus location outside Des Moines. There are 200 people there, ready to weigh in on which candidates should earn the four county delegates assigned to that location.

The caucus-goers are asked to go to different parts of the room to indicate their support for a candidate. After they do so, we see that 50 people are in the section for Sen. Bernie Sanders (I-Vt.), and then 45 for former vice president Joe Biden, 40 for Sen. Elizabeth Warren (D-Mass.) and 25 for former South Bend, Ind., mayor Pete Buttigieg. Thirty more are divvied up among other candidates, and 10 people haven’t made up their minds.

Candidates have to have the support of at least 15 percent of the caucus attendees at this caucus site to be considered viable. (Some smaller caucus sites are allotted fewer than four delegates and therefore have slightly higher viability thresholds.) In this case, that means that only Sanders, Biden and Warren can continue on. Everyone who supported Buttigieg or any of the other candidates has to decide whether they want to now support one of the three viable candidates, join to boost a nonviable candidate — or back no one at all.

Supporters of Biden, Sanders and Warren can’t change their minds, a new rule this year. Supporters of the nonviable candidates pick a team. Let’s say that 25 go to Biden — most of Buttigieg’s, for example — 15 to Sanders and 20 to Warren. Five supporters of other candidates decide not to back anyone.

Now we zoom out. Each of the three viable candidates in this caucus location will be awarded county delegates — but, as with everything else, it’s not that simple. A formula determines how many county delegates are awarded at a caucus location — the number of supporters of a candidate times the number of county delegates at stake, divided by the number of people attending the caucus.

So in this case, Biden gets 1.54 county delegates (75 times 4 divided by 195), Sanders gets 1.33 and Warren gets 1.23. After rounding, that’s two county delegates for Biden and one each for Sanders and Warren.

We zoom out again — and things get complicated. Each county is allotted a number of state delegates. Let’s say, for the sake of our example, that Polk County, where Des Moines is located, gets 10 state delegates. Across all of the precinct sites in the county, there are 40 county delegates up for grabs, including the four in our example. Each county delegate is therefore worth one-quarter of a state delegate.

The math is fairly straightforward (since we used nice round numbers). Biden is awarded 0.5 state-delegate equivalents from our precinct. Sanders and Warren each get 0.25 state-delegate equivalents.

Those fractions are odd, certainly, since we tend to think of delegates as, you know, people and we tend to think of people as, you know, whole units. Across all of the caucus locations and all of the counties, these fractions are added up for each of the candidates. The 0.5 Biden got here in Des Moines is added to the 1.3 he got on the east side of Ottumwa and so on. Eventually we get statewide totals for the number of state-delegate equivalents each candidate earned.

Those fractions are eventually translated into the number of people who will attend the convention where national delegates to the Democratic convention are identified. Most of those national delegates are divvied up by congressional district, with the final total across the state roughly matching the distribution of state-delegate equivalents that emerge from the caucuses.

See how things can get wonky?

Even after just the realignment in our imaginary Des Moines caucus location, we can see a potential conflict.

Who would you say won? In a normal vote, it would be Sanders, since he got more votes than anyone else in the first voting. But Biden’s team will note that this isn’t a normal vote and that, after the alignment, Biden came out on top.

Once we start adding up numbers across the state, more points of conflict can be introduced. In 2016, for example, Hillary Clinton got 49.84 percent of the total votes (after realignment) and 49.85 percent of the state-delegate equivalents. The race between Clinton and Sanders was close that year, but with far more candidates, this year it’s possible that differences in candidates’ support could be closer still. A small shift in the conversion of final support percentages to delegate equivalents could easily flip the positions of two candidates.

Clinton ended up with 52.3 percent of the pledged national delegates from Iowa, thanks to changes at the state convention. Things can shift.

What will be reported that night?

We’re used to knowing who won a race by looking at who got more votes. But, as should be clear from the numbers above, Iowa won’t be that simple.

Nor is the state Democratic Party making things easier. It will release not one number on Monday but four:

  1. The results of the first alignment (which, in our example, Sanders won).
  2. The results of the final alignment (which we had Biden winning).
  3. The number of state-delegate equivalents each candidate earned.
  4. And, a bit later, the estimated number of national delegates each candidate won.

That’s four numbers all aimed at measuring the same thing — whom Iowa Democrats want to be the next president. It’s likely that all four metrics will point to the same candidate. But if there’s a candidate who’s the second choice for a lot of voters, it’s easy to see how the first two numbers could be in conflict. Add in the wonkiness of the calculations here and you can see how the other two numbers could throw another name into the mix, especially if the race is close.

Who will the media say won?

It may depend on whom you listen to. The Associated Press will use the statewide state-delegate equivalents as its metric for victory. Most outlets will join the AP in doing so. But there’s nothing preventing a media outlet from focusing instead on one of the other metrics released by the state party, such as the final vote.

People may be able to pick and choose which result is most pleasing to them. Which, of course, is what the candidates will do.

Who will the candidates say won?

Themselves, naturally.

What’s important to remember is that, for all of the attention paid to Iowa, the actual impact on the race is small. We’re talking about only 49 delegates being at stake, of the nearly 2,000 needed to clinch the nomination — making how campaigns talk about the Iowa results perhaps more useful to them than the delegates themselves.

All the candidates have an incentive to leverage the results in Iowa to their advantage. Every campaign will want to say that it won the state or, at least, that it did better than expected. If Sanders wins the first-alignment vote across the state but not the final alignment, look for him to tout his victory as the first choice of Iowa voters. If Biden wins after alignment, his team will no doubt celebrate being the consensus pick of the state. Mix in the tendency of campaigns to celebrate even second- and third-place finishes as something important and we get some very muddy water.

To some extent, the decision by the party to sort of throw up its hands at the idea of a clear, numeric winner does us all a favor. We tend to place too much emphasis on the results in Iowa, after all, tracking how they might trickle out through the rest of the contests. It’s probably safer for us to approach Iowa with a default position of messiness, that the results are murkier than we like to present them. That’s probably a better picture to present to voters.

Or maybe a candidate will clearly win across all four metrics presented by the party. In that case, let a thousand takes bloom.