The Primary Process is a Disaster. Here’s How We Build a Better One.

A scene from the disastrous 1968 Democratic Party Convention in Chicago. (Associated Press)

She stares at me with a half-smile, hand on her forehead, eyes wide and unblinking with disbelief. “Why would we ever decide to do anything this way?”

I teach political science, and one of my students from my “Parties and Polarization” class has come to my office hours to talk about the primary process. I am fortunate enough to be teaching such a course during the 2020 Democratic primary, which offers ample opportunity to apply class concepts to a real world event that is relevant, important and of immediate interest to my students. These students, largely political science majors, mostly already pay a great deal of attention to politics, and have followed the primary closely, but the sad reality is inescapable — despite their interest and intellect, most of them are utterly bewildered by the process. So they come to me, and I explain the rules in embarrassing detail.

Of course, even as a political scientist, I struggle to keep in mind the primary schedule, the complex rules for delegate allocation, and the process by which actual human beings are selected to become convention delegates. If it is my job to understand this process, and yet still I falter, it likely reveals that this process is too cumbersome for the average voter to understand or, more importantly, follow and trust. So when I finish my explanation, students without exception react as the student above. And I completely agree with their evaluation.

But it’s not just a matter of complexity; many students simply find the process to be biased. Many complain that a few small, highly demographically unrepresentative states get an enormous amount of say over which candidates succeed and how the process unfolds. Others express skepticism towards the power the process gives party leaders — mechanisms like superdelegates, who can in narrow races overturn the plurality will of the voting public. Still others are frustrated that candidates such as Bloomberg (and to a lesser extent Steyer) can apparently buy their way into electoral relevance without sufficient participation in the campaigning and debating process. I agree with all of these concerns.

So why do we do this to ourselves?

Far too often, Americans are content to build their institutions on the skeletons of earlier systems, rather than engaging in a process of total reform, and that explains why we have the ugly and complex primary system that we do today. The idea of a convention with delegates from the various states sprung out of the necessities of the 19th century — state party officials could not easily convene simultaneously in a single location to decide on a nominee, so delegates were sent on their behalf to centralized national meetings. At this point, the idea of the public having a say in the process was absurd to the parties. It was not until the early 20th century that progressive politicians and social activists had placed enough pressure on the system that formal mass primary voting was implemented. Still, these votes were often non-binding, delegates were usually party insiders, and the public’s vote was seen merely as playing an advisory role. It was not until the chaos of the 1968 Democratic Convention in Chicago that this process, too, fell apart, as the rift between party elites (who still saw the public as offering ‘advice’) and the public (who saw their votes as ultimately binding) became too large. The McGovern-Fraser reforms required that primary results all become binding but, still, the structure of delegates-and-conventions from the 19th century was left unchanged. Much of the complexity of the modern system arises from the corpses of those systems since abandoned.

Thus do primaries join the list of American institutions — the electoral college, the campaign finance system, the filibuster, and so on — that appear to be ill-built for the present moment. However, the primary process is distinct from these other systems in that it is, at least in theory, easier to reform. The electoral college, for instance, is built into the constitution and would require a constitutional amendment to change. Given that this would require ¾ of states to ratify, and that it benefits more than ¼ of states, short of more exotic reform schemes, it is difficult to imagine the removal of the electoral college anytime soon. Most major reforms to keep big money out of politics would likely be found by the Supreme Court to violate First Amendment free speech rights, and have in the past (see Buckley v. Valeo and Citizens United). Eliminating the filibuster, the source of our constant gridlock, would not violate the constitution, but it would require leaders from both parties to roll the dice on a completely new vision for what control of the Senate would mean, something they seem quite unwilling to do.

Primaries, on the other hand, are ultimately run by parties for party members, and the constitution says nothing about the means by which parties may conduct themselves internally. This means that the Democratic and Republican parties can largely unilaterally decide to change their rules without worrying about who has control over formal institutions, or what constitutional requirements might be violated. In other words, the primary process is ugly, but it is one that we could change with relatively fewer barriers!

In this piece, I argue that we can build a system that significantly improves on our current one. First, I will outline what the goals of a functioning primary system should be in 21st century America. Then, I will spend the remainder of my time outlining a blueprint for what a fair, informative, and relatively parsimonious primary process could look like.

What Do We Want in a Primary?

I argue that a successful presidential primary should accomplish the following:

1. Maximize Informativeness: Voters should have the greatest possible opportunity to learn about the advantages and disadvantages of each of the various candidates for office. By the end of the process, modestly attentive voters should be able to a) understand the major policy and ideological differences between the candidates, and b) evaluate which candidates are the most skilled campaigners.

The former is accomplished primarily through the debate system, which at the moment is conducted via an informal network of media and college participants, and run via ad-hoc rules created by central party committees. Because of this, the quality and format of debates is largely a function of the media organizations who serve as broadcasters and moderators (and, of course, tradition). Of course, the incentives of these actors are not necessarily aligned with the parties, and the debate process is not necessarily best left in their hands. First, the primary process benefits media orgs primarily though viewership, and they know viewership is likely to increase when candidate in-fighting is encouraged, so the questions they use are often designed to elicit combat. Sometimes this is useful in that it reveals useful ideological differences between candidates, but often it produces sound and fury signifying nothing. Media orgs also have some incentive to avoid asking questions that are too pointed, as upsetting prominent candidates may risk the ability of the networks to later get regular interviews with the candidate, and perhaps especially if they win the election. Finally, it is likely that media orgs have their own candidate preferences, and leaving the process in their hands may mean giving peripheral elites too much power. Even ignoring issues of bias and incentive, the format in its present form often does a poor job of forcing candidates to focus on their differences rather than their similarities, especially in early debates.

The latter is accomplished, in theory, by staggering the dates of state primaries. While many complain about the early voting states, the idea itself is not a bad one. Presidential candidates are the only types of candidates that must run a national campaign in order to win office, and in a country as large (in terms of both population and geography) and diverse as ours, that is an incredibly difficult task. Creating key early primary states incentivizes candidates to spend a lot of their early campaign time in just a few places, slowly winning over the trust of its people. As this typically requires candidates to get out of their home states and spend an enormous amount of time competing for the trust of a smaller population, the process should boost the chances of candidates who are resourceful, likeable, trustworthy, clever, and hard-working. Similarly, it punishes candidates who are wasteful, uncharismatic, uncreative, and lazy. For these reasons, a staggered primary should be supported, but modified heavily to avoid other disadvantages, as I later explain.

2. Minimize Superficial Candidate Advantages: Some candidates possess resources that significantly increase their likelihood of winning, but are often poor signals of key candidate qualities such as campaigning ability and policy mastery. In particular, these “superficial advantages” include name recognition and personal wealth. Sadly, a volume of political science research demonstrates that primary races turn heavily on money and familiarity — the candidates who do well are often those that simply lots of people know, or those who have the money to run advertising campaigns to raise their public salience. In general election matchup polls run during the primaries, candidate performance is almost entirely a function of name recognition. Biden, for instance, has benefited tremendously from his proximity to Barack Obama as Vice President. Sanders has benefited from his role as the only significant non-Clinton candidate in the 2016 race. Bloomberg has largely neglected to participate in the primaries outside of an advertising onslaught valued at over $500 million.

Not only are these qualities largely unlikely to predict candidate competence, they are also unlikely to predict general election success. Any candidate who wins their party’s nomination will automatically enjoy nearly universal name recognition in the general election. Moreover, in presidential races, spending differences between the parties are usually negligible, and it’s likely that even outrageous spending shifts (say, Bloomberg dumping $3 billion into the general) would have diminishing marginal returns. Therefore, the ideal primary process will find a way to de-emphasize these characteristics in favor of far better ones.

3. Minimize Distrust in the Process: The American public already regards politicians and parties with extremely low levels of trust, so the primary process should seek to avoid making this any worse. Doing so, I argue, requires two mechanisms.

First, the primary process should be as simple as it possibly can be while achieving all its goals, but no simpler. Complexity breeds mistrust. When the uninitiated are presented with a system with a number of arcane rules, such as delegate allocation and nomination, one might fairly assume that the complexity will allow those on the other side of the information asymmetry — party elites — to screw them over. Therefore, any working primary system should be simple enough that you could explain it to an interested but average member of the public in less than a couple minutes.

Second, and this may be controversial, but the party should probably stop trying to decide the primary on behalf of its constituents. I am not unsympathetic to arguments that party elites are vastly better informed about the realities of elections than the voting public, and that they should retain some say in the process to help guide the public away from disastrous candidates. However, in the modern era, in which the public generally does not recognize any right of party elites to meddle, and will in fact view even well-meaning machinations with deep suspicion, any possible elite intervention will result in a worse electoral outcome than if they had not messed with the process at all. Any action in a primary that a significant percentage of the public would interpret as “stealing the election from the rightful winner”, should be avoided at all costs.

4. Minimize Factional Bitterness: I choose my words carefully here. Primaries should reveal distinctions between factions within the party, and should encourage candidates with differing viewpoints to argue with one another. This is a healthy part of intraparty operations and is a welcome feature of primaries. However, the process should do so while generating as little “heat waste” as possible — that is, energy should be spent on meaningful, substantive differences, and not superficial disputes. Furthermore, the process should not generate so much animosity between different factions that the party has a difficult time bringing the entire coalition together in the general election. Of course, this is largely out of the hands of party elites; intra-party animosity will be a function of actual substantive preference disagreements among partisans. Still, rules can be designed in such a way that at least attenuate the perceived unfair treatment of particular candidates, and to improve the odds of selecting candidates who satisfy as large a majority of partisans as possible.

5. Maximize Representativeness: Finally, the primary process should unfold in a manner that does not unfairly advantage a particular demographic group over others, or the candidate(s) supported by said group. Because race, geography, and age play an enormous role in candidate selection, the power of each demographic subgroup in the primary should be in rough proportion to some benchmark of their overall power — their proportion of the vote in the general election, their proportion within the party, or their proportion within society as a whole. Furthermore, while primaries will be heavily slanted towards those with high levels of interest in politics, a fair primary process will avoid erecting barriers that make it difficult for lower-interest segments of the population to participate.

Towards A Better Primary

Having laid out the goals of a functioning primary system, I now offer a proposal for reforming the current system that I think fulfills these goals as much as can possibly be done. I will examine in turn four aspects of this new system: 1) ranked-choice voting; 2) restrictions on campaign spending; 3) the primary voting schedule; and 4) the debate system.

Ranked-Choice Voting
The first major reform I suggest is to completely replace the current voting system with a ranked-choice system across all 50 states. For those who are unfamiliar with this system, ranked-choice voting (RCV, and also known as Instant Runoff Voting) would have voters rank the candidates in their ballot in order of preference. When votes are tallied, the candidate who receives the least votes is dropped, and on ballots that had that candidate as the first choice, the next most preferred candidate is then used in their place. This process then repeats itself until one candidate is left with a majority of votes.

RCV would solve a number of problems with the modern primary process. First, it would obviate the need for actual human delegates. To the extent that delegates serve any useful purpose at the moment, it is to make decisions on behalf of the public in the case that no single candidate is able to win a majority of public support. The goal of achieving majority support is desirable, given we are trying to minimize factional bitterness, but allowing delegates to control this process also violates our desire to minimize distrust in the process. RCV solves this problem by automating it; no human beings are required to determine a candidate with majority support, the system cannot be gamed, and it will always produce a majority winner.

Second, it would spare voters a series of agonizing decisions over which candidate to vote for. In our present system, those whose favored candidates are unlikely to win often feel pressured to cast their vote for a less-preferred candidate who can win. This frustration can lead to voters simply walking away from the process at worst, or voting unhappily at best. It also forces voters to attempt to play games of strategy that they are likely incapable of playing well. Under RCV, voters would not have to do this at all. They would simply list candidates in order of preference. If their preferred candidate is unpopular, no worries! Once they are eliminated, their ballot will reflect their next most closely held preference.

Third, it would minimize the value of poor signals of candidate quality like wealth and familiarity. When voters attempt to vote strategically, they are disproportionately likely to select a candidate who “they think can win”. Of course, their perceptions of electability will be heavily tainted by name recognition and campaign spending, so the process of strategic voting inadvertently rewards candidates for their least useful qualities.

Finally, it would encourage candidates to seek broader support rather than attempting to win through building an intense minority faction. Sanders, for instance, has been criticized for relying upon a highly active minority of the base without attempting to make sufficient inroads with other party factions. If he knew that ultimately it would hurt him to be listed at the bottom of the RCV lists of voters, Sanders would be marginally more likely to be conciliatory towards skeptical partisans. Alternatively, if Sanders were to ignore this incentive, the process would likely end up eliminating him for failing to connect with enough voters.

In implementing the RCV system, the following should occur: 1) the RCV ballot must be adopted by all fifty states for the presidential primary. If sufficiently motivated, the party should be able to force the states to do this. The simplest way to do so would be to mandate that any state that does not change its rules accordingly would be completely ignored by the party in determining the eventual nominee, much as the party today threatens to strip the delegates of states who violate their rules. 2) the Caucus process must be eliminated. As many have argued, caucuses discourage many citizens from voting due to their complexity, required effort, and length of time. Similar to the above, any state which tries to use a caucus would have its results ignored by the party. 3) the entire delegate system must be abandoned. There will be no delegates, no superdelegates, and no voting at the convention. The nominee will be determined entirely by the results of a nationwide RCV, which I will describe in greater detail below.

The parties could use their clout in getting states to modify their primary laws by also mandating states to improve ballot access. While it may be too difficult to compel all states to create fully functioning vote-by-mail systems, they could mandate that the elections be held on Saturdays. By letting people vote during a nonstandard workday, voters may spread out more evenly throughout the day, and not cluster at the end of the workday, which leads to long lines and, often, voters being turned away as polling locations close. Other reasonable ballot access reforms could be explored as well.

Finally, though I think this is of significantly lesser importance, the party should also require states to adopt a semi-closed party primary, in which independents are allowed to vote in a given party’s primary, but not members of the other party. This would discourage vote spoiling while allowing the party to test how well candidates do in terms of attracting new potential members to the party coalition. Standardization across the states would decrease complexity and lessen confusion over voter eligibility.

Restrictive Campaign Finance Rules
The next major change should be to the campaign finance system. Simply put, a candidate’s personal wealth should not be able to improve their odds of winning the nomination. As previously discussed, in a general election, preventing this from happening is extremely difficult, as campaign finance restrictions will run into first amendment challenges, and subtler methods of controlling the use of money would depend on the passage of state or federal laws. In primaries, on the other hand, because parties ultimately get to determine the nomination process, there is considerably more leeway. One cannot make it illegal to spend hundreds of millions on political ads, but the party certainly could make doing so a violation of its rules.

I propose the following: no candidate may use their personal finances in any way, whatsoever, to influence the election. This means that candidates would not be allowed to make personal loans to their campaign PAC; they would not be allowed to give any of their money to an associated Super PAC; they would not be allowed to make ad buys directly out of their personal fortune. Any candidate who is found to have violated this requirement would immediately be removed from all ballots (if possible), or at least would be automatically eliminated in the first round of the ranked-choice vote count.

Candidates would be welcome to draw financial support from any other source. Some candidates, like Sanders and Warren, might attempt to raise their finances entirely from small donations from supporters; other candidates, like Biden or Buttigieg, might choose to rely more upon established interest group donor networks, or some combination of the two. Candidates such as Steyer and Bloomberg, who survived far into the primary largely due to their personal wealth, would have to find other ways to raise money if they want to be competitive in the race.

A New Primary Schedule
The current primary schedule has the virtue of revealing which candidates are adept at campaigning, and giving them a boost, due to the way the primary is staggered across states. However, this system also violates our primary goals by advantaging a small and highly unrepresentative set of states, and creates mistrust and dampens enthusiasm among voters who are in states that go later in the process.

In the case of the former, Iowa and New Hampshire are given far more power and attention than they should. Both states are small and overwhelmingly white. The fact that they always go first also likely distorts American policymaking; for instance, knowing that Iowa will be highly influential in any future presidential primary, any member of Congress who has aspirations to higher office is likely to think twice before voting in a way that might harm Iowans, such as against corn subsidies. No single state should have such an unfair advantage.

In the case of the latter, as the campaign season goes on, voters are less and less likely to feel that their opinion matters, and increasingly that they are simply left with the ability to choose between the top two or three candidates preferred by people in other states who voted earlier. For those who vote after a candidate has already achieved a majority of delegates, their election simply becomes meaningless. For those who are in states that vote late in the process but before a majority winner is determined, this dampened enthusiasm actually could change election outcomes.

In my view, then, an ideal primary system should maintain some kind of staggered schedule while eliminating the unrepresentativeness and discouragement that results from the present system. To solve this, I propose that primary elections should be held across two rounds. Round 1 would involve the vote of a small subset of states, and Round 2 the remainder.

Round 1: In this first stage, there would be four states, and the vote would be held in early February of the election year, just as now, but the similarities end there. Each year, the four states themselves would be chosen at random from a subset of US states. To be included in the random drawing, states would have to meet several requirements. First, states with extremely large populations should be discarded — we don’t want the first round to include a very large portion of the overall electorate. Personally, I would just permanently exclude CA, TX, NY, and FL, which are substantially larger in population than the other states in the country. Second, a state that qualifies must be easily travelable by candidates. This does not necessarily mean choosing smaller states like IA or NH — NV, for instance, is one of the largest states by area, but the state’s population is almost entirely clustered around Reno and Las Vegas. Parties could develop a formula which requires a certain percentage of state residents to live within a preset amount of square miles. Third, no two states should be selected two primaries in a row; to encourage variety, any state who went in Round 1 last time would be excluded from the drawing. Finally, to encourage geographic diversity, at least one of the four states must be west of Texas, and one of them east of Texas and south of the Mason-Dixon line.

Once the qualifying states are determined by the above rules, an algorithm would randomly select four states. This algorithm would compare the combined demographic makeup (at least race, but possible age and income) of these four states to some benchmark — either the demographic makeup of the country, or of the party. Any combination that deviated from representativeness by some predetermined amount would be rejected, and the algorithm would re-run, repeating until four states are chosen. By turning this process entirely over to the algorithm, this would insulate party officials from attempts by state officials or candidates to game the system in favor of one or more states.

The selection process would occur no earlier than July of the year prior to the general election. Waiting until this time prevents candidates from attempting to develop networks in early-voting states months or years in advance of the primary, and gives all candidates an equal opportunity to campaign in the chosen states. This also gives candidates about six months to campaign in these states, which should be plenty of time to reveal who is talented and who is not.

The chosen states would be required by the parties to hold their primaries on the same day — during the first week in February –and any state that refused would have its votes ignored by the party. On “Ballot 1”, the raw ballots will be counted and processed by each state separately, and then handed over to the central party committee. The party would then report two results: the “raw” vote count (every ballot’s first choice candidate), and the RCV winner (from the four states aggregated together). Both of these figures would be used to inform the public as to which candidates appeared to campaign impressively, which did not, and which were able to build a coalition that cut across multiple factions. The RCV ballots for all those who voted in the early states would then be held off to the side until the remainder of the country votes, and counted in aggregate with all other ballots.

In addition to determining a victor, the RCV count would also determine which candidates would be forced to drop out, as the candidate pool should remain relatively small going into the next round of voting. One reason for this is to simply clarify the decision voters have and reduce choice fatigue. Another important reason is that one of the only problems with RCV is that, with a very large number of candidates, the order in which lesser candidates are eliminated can have a big impact on the eventual majority winner. RCV therefore performs optimally with a number of candidates that is greater than two but smaller than some clearly large number of candidates (say, ten). I would argue that the RCV process should occur until five candidates remain. These five candidates would qualify to participate in the next round, and all other candidates would be eliminated from contention and not appear in future debates or on ballots in future states.

Round 2: The remaining 46 states (plus territories) would then vote on the same day three months later, during the first week of May. Again, any state who tried to hold its primary on a different day would have its ballot results ignored by the party. Having demonstrated their ability to campaign in a small number of intense local campaigns, the remaining candidates would then be free to play with whatever set of strategies they feel would best make them viable in the rest of the country, and would have three months to make their appeal to the public.

On “Ballot 2”, all states would vote by RCV, each state would count its ballots and send the results to the party’s central committee. At this point, the RCV ballots from the four early states would be added back into the aggregate. Any candidates on these ballots who were eliminated in Round 1, or drop out between Rounds 1 and 2 would be eliminated first in the RCV model, and their votes redistributed accordingly. The RCV process would repeat until one candidate has a majority of the votes. At this point, the candidate would ultimately be declared the party’s nominee, and the primary process would come to close.

This new schedule has numerous advantages. By randomizing the early states by algorithm, it minimizes corruption and ensures representativeness. By automatically narrowing the field to five candidates for most voters, it reduces confusion and choice fatigue. By having the remaining states all vote simultaneously, this reduces voter dissatisfaction and increases voter turnout. By eliminating delegates altogether, it simplifies the system while increasing voter trust by denying party elites any tools for possible machination. In short, it preserves the advantages of the current system while minimizing most of its disadvantages.

A New Debate Process
Finally, I believe that any reform of the primary process should occur alongside reform of the primary debate process. While political scientists generally believe that debates have very little impact on the outcomes of general elections, it is clear that primaries are very different. A quick comparison of aggregated polling trends to the dates of primary debates reveals that these debates do in fact impact voter attitudes. Unlike general elections, primary voters lack easy cues for voting (i.e. the candidate’s party), and are therefore much more susceptible to media messaging. The media has a good deal of leeway in how it chooses to cover candidates, but debates tend to produce consensus across the media landscape in terms of winners and losers in a fashion highly visible to voters. The effect of debates is probably not among those who actually watch them, but rather due to the headlines, which are much more widely read. If debates affect even short-term attitudes towards candidates, then the signals produced by the debates matter, and therefore, in turn, the structure of the debates matter as well.

First, some general rules that would improve the debates. To begin with, candidates who are aggressive enough to talk over the moderators in order to gain more time speaking should not be rewarded. The standard should be that all microphones are held off by default, and turned on only when a candidate is given their time to speak. When the time is up, the candidate is given a warning light and has ten seconds to wrap up. At the end of ten seconds, the microphone is automatically cut off and the moderator moves on.

Second, the time candidates should have to respond to questions should be increased from the standard 60 seconds to, say, 120 seconds. The longer candidates have to speak, the more likely they are to reveal inadequacies either of charisma or knowledge. Trump and Bloomberg are examples of candidates who had multiple awful performances in primary debates, but actually benefit from higher numbers of candidates on-stage minimizing their time talking. Bad candidates, given sufficient rope to hang themselves, will more easily be eliminated from contention.

Third, visual aids should be allowed and encouraged. It is ridiculous to me that most people are used to consuming news via charts, pictures and chryons, and educational material via powerpoint, but we still expect debate consumers to understand complicated claims without simple graphics to help back them up. The extra time in responses would give candidates the ability to walk viewers through very simple charts (of course, those who use complicated graphics and jargon will be punished by viewers accordingly). This should improve informativeness while mirroring the modern learning environments of the average American.

Moving onto the timeline of the debates, a key goal of the debates should be to facilitate a large number of candidates in the beginning, and then winnow the field as time goes on. The DNC this season, for instance, has attempted to do this by using polling and fundraising numbers as requirements, and incrementally raising the requirement as time goes on. This ruleset should largely be maintained. However, this has nevertheless resulted in a process in which, on any given night, the number of candidates might be as high as ten, and only two of eleven debates have featured less than seven candidates. Generally, the number of candidates at any given debate should be held at six and only occasionally more. This will allow any given candidate to talk for a sufficient amount of time, and will help a confused voting public to actually remember the differences between candidates.

In terms of debate format, the debates currently do a poor job of highlighting the differences between candidates in a way that is simple, clear and memorable, especially in early debates. Many candidates end up trying to simultaneously find ways to distinguish themselves from other candidates while taking essentially the same positions, which ends up confusing voters into thinking differences between candidates are larger than they are. Other candidates deliberately deflect or hedge when asked to offer policy specifics, leading voters to presume they support policies that they may not. I propose that some debates, particularly early debates, should make use of “lightning rounds” in which candidates are asked to make a couple dozen up-or-down votes in response to a series of policy questions asked by the moderators. Candidates in these rounds would have mics turned off and would not be able to provide nuance, at least initially. The moderators could provide these questions in advance of the debate, so candidates would not be able to claim they misunderstood the question or were unprepared to take a clear position. These lightning rounds could be used later in the same debate, or in future debates, as fertile grounds from which to pluck questions that will highlight clear differences between selected candidates. They would also help news organizations report differences clearly to their audiences.

Finally, towards the end of the primary process, candidates should be forced to spend a significant length of time debating particular topics. Most policy areas are simply far too complicated to cover in 15–20 minutes segments. Entire debates should be devoted to major policy areas of public concern such as health care, immigration, jobs, and so on.

Combining all of the above ideas, I propose the following as a formalized debate schedule:

Debate 1: Held in September. All candidates who meet very low threshold (e.g. at least 1% in 3 polls plus some low fundraising amount) are allowed in. No limit on the amount of candidates on-stage. In the first half, each candidate simply gets 3 minutes to introduce themselves and say whatever they’d like. The remainder of the debate will be spent on a lightning round with many quick questions, as described above, and then the debate would end.

Debate 2: Held in October. Requirements ratchet up slightly. Up to three nights of debate, with six candidates each, for a maximum of eighteen candidates. Debate questions should be drawn from a combination of traditional questioning and multiple follow-ups on distinctions generated from the lightning round in the previous debate.

Debate 3: Held in November. Requirements ratchet up slightly. Up to two nights of debate, with six candidates each, for a maximum of twelve candidates. Again, the debate format can mirror the traditional format and continue to delve into distinctions revealed in the first debate.

Debate 4: Held in December, on one night. Top eight candidates in some combination of polling and fundraising qualify. At the beginning, the lightning round is held for a second time, with a combination of old questions (to see if candidates have moved on issues) and new (to pick up on distinctions that may have been ignored). For the remainder of the debate, moderators will focus on the distinctions revealed from this lightning round.

Debates 5 and 6: Held at the beginning and end of January. Again, one night only for each, top eight candidates get in. Debate format resembles that of Debates 2 and 3.

Debates 7–10: These debates take place after voting on Ballot 1 has occurred, staggered every 2–3 weeks. By now, the candidate field has been narrowed to no more than five, and these five are pre-qualified for all remaining debates. Prior to these debates, the party will take a straw poll from a random list of a couple thousand registered party members in the mass public on what issues they care about most. From the list, the top four issues will be taken, and each of these debates will focus on that issue for the entirety of the debate. Candidates will be forced to go into extreme detail on these policy areas, and will be unable to hide behind generalities.

Debate 11: This debate takes place at the end of April just before the final vote. Candidates will have one last opportunity to address the voting public in a debate with a traditional, multiple-issue format.


The purpose of this article was to animate discussion of what would produce a better primary process. This is just one set of solutions, and they represent what I think would work best. That said, I have advanced a large number of ideas, and of course many of them are open for debate and tweaking. For instance, most of my ideas about the debate process are completely separable from my ideas about the primary voting process, and do not have to go together. I consider this a jumping off point, not an infallible blueprint.

Still, I think it is fairly clear that this process would produce results that are superior to our existing process. The primary that I have outlined would leave very little decision-making power to elites, helping to reduce cynicism and distrust in the process. It would involve a voting format that would incentivize candidates who can build a big tent coalition, but also facilitate voters who wish to support, at least initially, less popular candidates of high personal or ideological congruence. It would prevent the very wealthy or very well known from automatically dominating the process. It would restore fairness to people of color and poorer voters, the former of whom are poorly represented in present early primary states, and the latter of whom are unlikely to participate in caucuses. It would prevent states from gaming the process and gaining unfair institutional advantages. The debate format would improve informativeness on candidate positions, and do a much better job of revealing weak and insufficiently prepared candidates.

Some might argue that elites might have very little incentive to change this process, an important point given that it would be entirely up to party elites to implement these changes. Indeed, they would effectively be giving up their power to control the process. What would they benefit from this? In recent years, party elites have been viewed with great suspicion by their bases, and the insurgent support for candidates who openly rail against the establishment, such as Trump and Sanders, is proof that party elites are deeply unpopular. If party elites truly care more about the health of their party than their personal power, and want to win elections, they should be glad to usher in a process that would allow them to fairly claim minimal influence over the process, and which would produce automatically what they seek to achieve manually — consensus among the base, to whatever extent that is possible.

Assistant Professor of Political Science at University of North Florida. US politics and political psychology. Lover of music, science, sports and comedy.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store