Julia Azari is an associate professor and assistant chair in the Department of Political Science at Marquette University. This is the third op-ed in a series about how to improve the presidential nominating process.

Only a fraction of the Democratic primary electorate has voted so far, but the nomination season is off to a rocky start. Independent Bernie Sanders seems to be leading in popular votes, while upstart Pete Buttigieg is ahead in the delegate count. And there’s also the question of whether either one — or any of the other candidates — can bring the party together moving forward.

The current process is clearly flawed, but what would be better? Finding an answer means thinking about the purpose of presidential nominations, and about how the existing system falls short. It will require swimming against the tide of how we’ve thought about nominations for decades — as a contest between everyday voters and elites, or as a smaller version of a general election. A better primary system would empower elites to bargain and make decisions, instructed by voters.

One lesson from the 2020 and 2016 election cycles is that a lot of candidates, many of whom are highly qualified and attract substantial followings, will inevitably enter the race. The system as it works now — with a long informal primary, lots of attention to early contests and sequential primary season that unfolds over several months — is great at testing candidates to see whether they have the skills to run for president. What it’s not great at is choosing among the many candidates who clear that bar, or bringing their different ideological factions together, or reconciling competing priorities. A process in which intermediate representatives — elected delegates who understand the priorities of their constituents — can bargain without being bound to specific candidates might actually produce nominees that better reflect what voters want.

A nomination contest is not like a general election. They aren’t being fought to win, but to go on to November. But the kinds of processes that we associate with more open and high-quality democracy may not actually help parties produce nominees that really reflect the party’s overall concerns. Democracy thrives on uncertainty — outcomes that are not known at the beginning of the process. But uncertainty doesn’t help parties strategize for the general election.

The reforms that created the modern primary system in the 1970s opened the door to too much uncertainty — and to divisive nominees such as George McGovern in 1972. This spurred efforts by party leaders to take control informally through a system of endorsements and donations, narrowing the field down to acceptable candidates before the first caucuses and primaries took place. What’s emerged since then is a process that’s incredibly complicated. Different states jockey for influence in the official primary. Candidates strategize about delegate counts. Elites try to shape the decision early on. Everyone is doing guesswork about what others want. Reforms to the process should try to make that guessing a bit more informed.

Some critiques point to nominees such as Donald Trump — lacking in conventional qualifications and appreciation for democratic norms — as proof that nominations shouldn’t be too democratic. But the same system, more or less, produced candidates such as Barack Obama and Mitt Romney. The quality of the system can’t be measured solely in terms of the kinds of nominees it produces. Instead, we should think about how it reflects the preference and values of the different components of the party coalition.

For decades, the conversation about nominations has been about the conflicts between party elites and everyone else. Today, that conversation is counterproductive. A better approach is to think about how voters and elites could best play their different roles: to make their political parties more representative while ultimately narrowing the nomination choice down to one person. And the best way to do that would be through preference primaries.

Preference primaries could allow voters to rank their choices among candidates, as well as to register opinions about their issue priorities — like an exit poll, but more formal and with all the voters. The results would be public but not binding; a way to inform elites about voter preferences.

This process could accompany a primary of the sort we’re used to — in which voters’ first choices instruct the delegates, and preferences come into play only if there’s no clear winner. The primaries could also be held in combination with elections for convention delegates so that these representatives are informed by their constituents’ preferences. This would also help voters hold these delegates accountable in the future. The point is to build a way for party elites to understand what their base is thinking, and to allow them to bargain so that these different preferences and priorities can be balanced.

This might sound labor-intensive and a little risky, but the process is already lengthy and expensive. Candidates jockey for endorsements and donations for months leading up to the first contests. Why not invest some resources in finding out what voters really think, and then allow party delegates to figure out how those opinions can translate into a winning ticket?

Read more: