I agree that countries with more than two parties have more opportunities to create majority winning coalitions that cut across painful cleavages. In a three party system (for simplicity I'll do equal sized parties) there are four majority coalitions: all three parties, or each of the three two-party coalitions. This is very good for healing wounds and building bridges and very bad for stability (because there is a majority coalition that can always improve on the status quo, no matter what the status quo is, creating cycles between each of the coalitions). I definitely agree that it makes it easier for regular citizens to disconnect the party they support from sort of more mundane social interactions. I definitely agree it's better for negative out-group affect. It's much harder to hate the other if the boundaries of who is in the other changes regularly and people cycle in and out.

But the US is a presidential system. While the presidency isn't everything, it is a big enough prize. No system can get around the winner-take-all nature of the presidency. I favour ranked choice and/or MMP and would support any reform proposed to achieve better institutions of electoral representation, but nothing is going to get around the fundamentally two-party nature of the contestation of the presidency. If we look at the whole history of the US, the party system has changed a number of times, but each time it has been around assembling roughly two groups contesting the presidency, regardless of how those groups map to formal parties (not at all in the JQA/Jackson era, very strongly today).

Still, regardless of the institutional setup, it's the case that inter-party polarization can emerge from either increasing extremeness, or better correlation between party and ideology, or better correlation between multiple dimensions of ideology, and the same processes I mention above are visible in other multi-party countries as well, to a lesser degree.

What, why is the presidency two party in nature? Any transferable vote system would change this surely?

This is a long digression, sorry. Let's say you have a presidency and that among first choice votes, 40%+1 prefer L, 40%-1 prefer R, and 20% prefer C.

In a plurality system, would-be C voters have to choose between voting sincerely (knowing their candidate has no chance of winning, or to exert a plausible threat in the next election for candidates to take them seriously) and voting strategically (picking between R and L, which is closest to them). This creates a system where the closer of the two has every incentive to co-opt the votes as early as possible and everyone who donates money or time to try to affect the major candidates early. That's what we see now. In the US, most of the negotiation happens within parties. The Tea Party and Sanders movements were both things that acted as caucuses within an existing party instead of forming new parties, because of this incentive.

In a ranked-choice system, the C voters vote for C... and then either L or R. C gets eliminated, and L or R wins. So while the strategic versus sincere tradeoff for C voters is dismantled by the voting system, there's still a center of gravity pushing people into two camps (the L/C->L and the R/C->R camp).

Besides voting, this impacts organization, campaigning, issue messaging, donations, etc. Given that the election is winner take all and there are two camps, what infrastructure should resources pass through? And we've decided that for branding purposes, the answer is mostly stable, long-term professional parties. If you're extremely conservative, you'd rather your choice win the Republican primary than go it alone in an election for senate as an independent conservative. Going it alone does sometimes happen -- but in these cases there still tend to be two main camps for the winner take all election, like how Alaskan senate seats are recently I versus R instead of D versus R, but this normally only happens when there is dysfunction in the infrastructure that should have been one of the two camps.

So, now, take France, which has runoff voting for President. The first round of voting essentially functions as a primary to determine who represents the right and who the left. The institution makes the two surviving camps explicit for the second round of voting, but even in an instant runoff framework this camp structure always would have happened because there's only one president and you can't divide the presidency.

Canada is an interesting case in that it doesn't have a presidency. It has more than two parties. But within each district, there are typically just about two parties -- the country typically has around 2.7-3.5 political parties measure by ENP but the mean district typically has about 2.2 parties. I haven't run the numbers on the 2019 election, so I'm goin by memory. The two party pressure still exists, because electoral districts are winner take all, like the presidency, but it exists at a slightly different level of organization. Canadian parties often try to nationalize their elections (make a vote for your local Liberal a vote for Trudeau, or what have you) but conceptually the competitive dynamics are multiparty only at a national level and two party in most districts. Where they aren't, the third party almost always acts against its own interests as a spoiler, and so they should be. The pressure is powerful.

Thus Duverger's law is best formalized as: in any winner take all election, there is strong strategic pressure to reduce the number of parties competing. Where more parties emerge, it is often a transition state that will in equilibrium reduce down to 2 over time.

The best way to reduce the two camp effect is to reduce the number of things that are winner take all. An MMP or PR electoral system is best suited to ensure this at the voting stage. Even in these systems, the two camp effect often emerges during coalition formation (see: Netherlands; GL and PVV and VVD and Labor can all exist in parliament, but the ways in which they form coalitions are mostly left versus right depending on relative party sizes).

To fully get rid of it, you'd also have to figure out a way to divide the executive in parliament in a stickier way than coalitions do. Something like the Lebanese system (which is designed to ensure every ethnicity controls one major political role) could work. If you imagine a parliamentary country with PR and with a restriction that winning coalitions must rotate the prime minister by party every six months, that would be an interesting start. I am not aware of any countries that go this far.

My post is way too long. I suggest you read the short book "An Economic Theory of Democracy" by Anthony Downs. It is maybe 100 pages and has a wonderful quality of starting with very simple models and then relaxing assumptions until it approximates reality. It's the canonical political science reference to this and one of the main things that kicked off the sort of "median voter" approach to social choice, which as other commentors here mentioned is one of the main approaches to thinking about this problem.

It would change the dynamic, but the issue is the fact that the presidency is "winner takes all." In a multiparty legislature, they parties compete at the polls, and THEN they form a coalition. The stakes for the presidency are such that it encourages coalition building BEFORE the polls, to produce a candidate with an expected majority.

You can see the dynamic during some of the less polarized political cycles of the past. The parties and candidate had to build coalitions of "interests" groups that weren't necessarily pre-committed to one party or another. I could imagine a transferable vote system looking getting us back to that "coalition" state where the parties platforms changed based on the dynamic coalitions.

It adds alot of complexity with the questionable benefit of driving more extremism.

That’s not a party problem, it’s an artifact of our mass experiment with various flavors of social media and unfiltered mass communication. The situation with Twitter giving maniacs a megaphone, Facebook radicalizing anyone and YouTube offering a twisty tunnel of crazy to drive engagement isn’t sustainable.

The counter response will be a lockdown and that will address most of the issue. Unfortunately there is plenty of suffering and discord to be had before that happens.

> But considering the fact that political positions appear to correlate with brain structures

The do now, but didn't historically. So (presuming the correlation is true) are any of the historical factors relevant to changing things today?

It seems like historically there were more dimensions to consider such as class, regional interests, religion and then within those combination of factors you could have a more-liberal or conservative view.

> There's nothing in evolutionary theory that says the impediments can't be social and behavioural instead of anatomical.

There is some data to support this. Charles Murray's Coming Apart cited a big trend of separation amongst the urban/liberal/educated and rural/conservative/religious. It really isn't enough to drive evolution, it certainly has an impact on culture and politics. There is no "reversion to the mean" if the distribution is binomial.

The U.S. recently briefly had a three party system with the emergence of the Tea Party. All it did was radicalize the GOP. Just ask John Boehner.

Arguably this radicalization was only possibly because of the Hastert Rule. Under the Hastert Rule the smart strategy was to only caucus with the GOP, and obstruct the GOP when they didn't get what they wanted. That provided the Tea Party with maximum effectiveness. Without the Hastert Rule and some other technical developments, the Tea Party could have and would have occasionally shifted their support.

Another way to look at it is that without the Hastert Rule and similar mechanisms that bolster the power of parties, a two party system is largely a two party system in name only. What matters isn't the number of nominal parties, but the effective autonomy of individual representatives. You don't need 3 or 5 or 12 parties, which in almost every country invariably come to represent niche interests (similar to all the caucuses in Congress) but otherwise usually align with one of the main parties. You can have the same outcomes so long as politicians and informal caucuses can more easily defect. Not only will they defect, but it softens the official stance of the remaining parties.

We don't need to copy the structure of multi-party European systems. Rather, if we want the same outcomes[1], we should understand the dynamics and replicate or the types of dynamics that matter. Without the latter, simply copying the structure is unlikely to be effective. With the latter, copying is unnecessary.

[1] That assumes the outcomes are different and desirable. But the notion that multi-party systems tend to more centrist and inclusive politics seems stale to me. The rise of populism and even nationalism has hardly been confined to traditionally two-party systems, like in the Anglosphere. Heck, when it come to issues of abortion and immigration the U.S. has always been and remains more liberal than almost every European country.

The Tea Party was not a separate political party. It was a faction in the Republican party that was focused and non-compromise with the Democratic party.It's hard to argue that this would be the same dynamic with a true third party.

"The U.S. recently briefly had a three party system with the emergence of the Tea Party. All it did was radicalize the GOP. Just ask John Boehner. "

In a multiparty system they probably would have been a separate minority party.

I think what the US is really suffering from is that the country is maturing. For a long time people could just claim new areas, start living there and create new lives. The country is now fully owned and populated so moving around to start anew from scratch is not that much of an option anymore. Basically the country is filling up with people. Europe had this going on for much longer and the US is going that way too.