A Random Fix to Polarization

From the New York Times; 2022-11-13

Following the midterm elections in the U.S. this week, the punditry is ablaze with stories about how voters have rejected the lies, spoken out about abortion, indicated that this inflation spike is not a primary concern, and that perhaps we are finally shaking off this fever-dream and turning a corner.

Really? By what margin? The mere fact that the balance of power has taken so long to determine (still not settled in the House, as of this writing) indicates to me that we are not collectively on the same page about these issues. The main story, to me, is that the deep polarization pattern continues to stymie our political system.

I have an idea that could make a huge difference. It’s an idea that has essentially no hope of gaining traction—especially since: who the hell am I? But I will put it out there all the same. Maybe others will see the logic.

History of the Problem

The U.S. has essentially always been a two-party system. But before 1970 the Democratic party had a healthy mix of both liberals and conservatives, as did the Republican party. Bipartisan deals were common, as liberals or conservatives from both sides of the aisle would find common cause and push legislation. I think of it as a well-shaken beaker of oil and water.

The turmoil of the 60s, culminating in a bloody and shocking 1968, precipitated reforms to how nominations were made for political candidates—moving control from smoke-filled back rooms to the populace via primaries and caucuses for each party.

The result has been a near-perfect separation of oil and water. It has taken 50 years (as “old” party affiliations die with their members), but partisan primaries have worked like a charm. By now, essentially all liberals associate with the Democratic party and all conservatives with Republicans. This process has been assisted by media outlets that segregate consumers into their respective camps—although I don’t want to draw a false equivalency: I place much of the blame on the right-wing for capitalizing on this modality most aggressively. They saw and exploited the opportunity first and most cleverly.

In any case, the partisan primary system has resulted in giving the more ideologically extreme elements of each party a bigger voice. The math is pretty simple. Since the U.S.. is roughly 50/50 in the two parties, anyone gaining over half their party’s support in a primary gets the nomination, but this corresponds to just 25% of the electorate. It’s actually even worse if the contest is between more than two candidates: a candidate of little appeal to the center can walk away with the prize.

Now in the general election, we end up with candidates that are not attractive to the 50% in the middle, and the winner is only well-liked by a distinct minority. We therefore create a machine that disfavors centrist candidates who have cross-over appeal. Because primaries are partisan, any cross-over appeal is wasted in the nomination process, and therefore a liability rather than an asset. Of course we end up in this ultra-polarized mess. How could our partisan primary system accomplish anything else?

A Random Fix

Fixing polarization, therefore, must involve fixing the nomination process. To do this, we need to allow centrist voices to speak. That only happens if we bring down the artificial partition dividing the primary process, cleaving centrists into two non-interacting camps.

Step 1: Imagine that you, as a voter, could weigh in not only on who you’d like to represent your own party (if you identify with one), but also who would represent the other party. For independents, it’s the same: an expression of who you like best from either party (can also weigh in on third parties: no real limit here).

Step 2: Ah, you realize: this will never work because one side would try to select the weakest opposition candidate. There’s even a vulgar term for that behavior that I will decline to propagate. So the remedy: randomness.

This is the heavy lift here. Don’t run away in horror until you hear me out. Basically I suggest a coin flip (not literally; nuances to follow) to decide which nominee advances to hold office. Because this final selection is out of the hands of voters, they’ll think twice about selecting the worst candidate from the other team: a good chance they’ll be stuck with the choice and suffer the consequences.

What emerges is good-faith efforts to pick candidates who you could tolerate, even if not your overall preference. Lots of consequences come out of this, which I’ll return to after addressing some nuances.


Nuance 1: Not a literal coin. The “coin” should not produce 50/50 odds, but should be weighted to represent the district’s composition. For instance, in Wyoming (a deep “red” or Republican state), maybe the electorate is 70% Republican, 25% Democratic, and 5% Libertarian. So fashion a “coin” that transparently produces outcomes following those odds (maybe with a lower cutoff: “you must be this tall to go on this ride”). Ideas for implementing this appear farther down. So all the Republicans in Wyoming will have a say in the candidates from each party, knowing that the odds are in their favor that their selection will advance to office. All the same, they’ll have a say in a Democratic candidate who they believe could represent their state well enough, if it came to that.

Nuance 2: Need not be equal voices. Why would a Republican want a Democrat to have equal say about their candidate? One could imagine a weighting system, where you have a weight of 1.0 in your own party, and 0.75 (or whatever) in others. You’d want it to be significant enough to hold sway, but would want to prevent outsiders from dominating decisions within the party. In all likelihood, the weights would be set in accordance with the electorate composition to maintain dominant weight within the party. It’s easy to come up with prescriptions for this, even in a multi-party system.


Most significantly, imagine how the elected official behaves in office. They know that in order to get re-elected, they have to satisfy all their voters—not just their polarized base. Isn’t this what we mean by democracy? Aren’t we aiming for our officials to represent all of us? Shouldn’t we be seeking the collective good?

This structure moves the situation to one that is more true to the “of the people, by the people, for the people” creed, rather than the current version that replaces “the” with “some.”

What we’d get is more centrist office holders that reflect the bulk of the electorate. I think of the ideological spectrum as being distributed like a Gaussian (bell curve). Yes, we have (loud) voices on the wings, but most people sit in the middle. These are exactly the people who have been marginalized over the last few decades as the oil and water have fully separated. In fact, the oil/water metaphor misses exactly the point that most people are not hard-over into being one or the other.

Isn’t it time we stop listening to the shouting and grandstanding from the political extremes, and pay attention to the competent and pragmatic middle? Such a transformation might make us think we’re all in this together, and lead to decisions that are not about team victories but about the collective good. At the very least, we become far less dysfunctional and usher in a new age of bipartisanship, while maintaining vibrant political parties that can see things differently. Discussions become more civil and constructive.

Another key advantage: think about what this does for voter participation. The poor Democrat in Wyoming knows now that their candidate doesn’t have a snowball’s chance to make it to office. So why bother? Why vote? But if there’s a 25% chance of getting your wish? Anyone who has rolled dice knows that even a one-in-six chance is very motivating. So voter participation could go way up.

Increased interaction also becomes more common: right now the teams stay in their locker rooms until the big game. Under this proposed system Democratic voters will talk to Republican voters and vice versa about the players on the others’ teams. More common ground; more shared interests; more relevance.

I encourage you to think about other upsides (and sure, downsides) to the scheme.


A brief detour into how to make the random selection truly random, transparent, and unhackable. I have not gone to great lengths to think this through—not worth the effort at this conceptual stage. But imagine this as a template.

Have you seen Pachinko, or Plinko? It’s a set of pegs on a grid and a ball or disk that falls through the pegs getting bounced left and right before landing somewhere along the bottom. Dropping 1,000 balls and binning the bottom into vertical “stalls” results in the emergence of a gorgeous Gaussian. From this, one could pick any weighting one wanted (akin to rolling a 1,000-sided die), and of course 1,000 is not a limitation: could be more or less.

In a 50/50 district, the simple interpretation of “left side goes to the Democrat” and “right to the Republican” would apply. But you can imagine the dividing line migrating left or right depending on district composition. Likewise, the partitioning need not be confined to two parties.

Now, a second stage of randomness selects which ball of the 1,000 is the “winner” ball. This is done in such a way that no one knows as the balls fall which is the winner, to discourage nefarious attempts at the difficult prospect of interfering with the physics of the fall (magnetic fields, etc.). Also, any such influence applied to all drops to bias the result can be self-calibrated out by not drawing the partition lines until all the balls have fallen, self-proclaiming their actual distribution rather than adhering to some theoretical expected distribution. This second-stage would, for instance, announce after the drops are complete that ball 428 was the winner, and wherever that one landed indicates the result of the “coin flip.”

The exact mechanism is not critical, and this is just one serving suggestion. It should not be hard to devise a fair and transparent system immune to foul play. Plus, it would be fun to watch as a national event!

General Election?

In the scheme outlined here, the “coin flip” becomes the general election. Voters express their preferences at what we now call the primary stage, then sit back and watch the random result. It does not have to be this way. A second round of voting could be injected after the “primary” winners have had a chance to state their cases, debate, etc. Either a straight vote or ranked choice could establish voter preference and set the lines for the random partitions. It is important to keep the randomness at the end, though. Otherwise we slip right back into the tyranny of the majority wing and marginalize the minority (and centrist majority) voices.

The electoral college could co-exist with this scheme, if wished, or it could be abandoned.  Compelling arguments exist for either. Without it, a single random process could be used for national office (i.e., President).  Otherwise each state conducts its own, as they also would for Senators.  And so it goes as districts get smaller: each contest ultimately comes down to a weighted “coin flip.”

Final Reflection

It is amusing to ponder how many ills a random element could fix. We tend to think of randomness as a lack of control. The fallacy is in thinking that we’re good at control. How is it working out? Nature is full of randomness, and in fact thrives and evolves as a result. Our political system could perhaps use a bit more of it.

Coming back to reality, I suffer no illusions that this scheme is anything more than an intellectual exercise of “fantasy football.” Vested interests have little or no incentive to fix our current system. Those enjoying majority rule have no incentive to open the door to the minority. Why would Wyoming Republicans entertain even a 25% chance of Democratic representation? It sounds to them like playing Russian Roulette.

So there we are. It’s not the only case where I feel like Dorothy in The Wizard of Oz: all along we have the power to make transformative changes, and it’s really not that hard. But it’s the last thing we’re going to try. As my wife’s mother often told her as a child: Suffer, then.

Hits: 3528