As we all know, political polarisation is a big problem. There have been plenty of articles (e.g. this one I picked off Google, from the New Yorker) written about why it’s getting worse and why that’s a bad thing — it even has a Wikipedia article. But all of it leaves me confused. It just doesn’t make sense.
You see, there’s plenty of evidence that political parties are moving apart — for example, Wikipedia has this graph showing DW-NOMINATE scores for the major American political parties:
Also, if you’d like a long footnote on the geographical scope of this post, I’ve got a treat for you: 1
The lines in the graph above are clearly moving apart, and have been since some time in the late ’70s. The weird thing is that this isn’t a graph of the opinions of regular voters, who are susceptible to the kinds of complex cultural phenomena all the articles talk about; it’s based on the votes of actual elected members of the House of Representatives, who have other priorities.
I’m sure they do believe things, want to do the right thing, make a difference to the country, address the issues that matter to their constituents, et cetera… But also, I’m cynical enough to know that what they really want is to:
Keep their jobs, and
Obtain or retain power.
These are completely rational things for them to want — politics is all about getting your party (and ideally yourself) into power, because that offers the best opportunities to make change happen and get things done.
But that graph above really isn’t the way to do that!
At the beach
Let’s digress.
Imagine you’re a young entrepreneur in the hot tourist destination of your choice, lounging on the beach one day when you spot some nearby tourists looking a bit overheated.2
“Gee,” one of them says, “I wish that ice cream stall weren’t so far away.”
Being the enterprising young thing you are, you take notice of this gap in the market. You head into town to get the required permits and buy an ice cream maker, then wander back down to the beach to set up shop.
But where?
Your only competitor — let’s call it Nigel’s Ices — is about a third of the way along the beach, where this bluish dot is:
So you decide to set up somewhere on the other side of the beach, to serve the people who currently have a long way to walk to get ice cream. Anyone who’s closer to your stall than Nigel’s will, naturally, choose you — so he gets the blue areas of the beach, and you get the reddish ones:
Of course, Nigel isn’t too happy about you stealing his customers, so he moves a bit to the right. He remains the best option for everyone to his left, but manages to steal some more of your customers from the centre ground, so things look like this:
You, of course, retaliate by moving a bit to the left, stealing your customers back — and then some:
Nigel retaliates in kind, and so on. Before long, you’re both clustered together somewhere, each serving your own section of the beach. You can’t move any further left without colliding with Nigel, and you can’t move any further right without losing customers — and providing space for Nigel to steal more ground. Seemingly at an impasse, you settle down for the long haul:
But wait! You, being the Apprentice winner in the making you are, have an idea! If you leapfrog Nigel, you can steal most of his (larger) side of the beach, leaving him your smaller section (plus a bit extra). A cunning plan:
Maybe there’s space for Nigel to leapfrog you in the same way. Then you can leapfrog him again, and so on. Each time, the smaller share gets slightly larger and you both get slightly closer to the centre of the beach, until you’re both sat right in the middle, controlling equal shares of its length.
Now neither of you has anywhere to go. You’ve found a Nash equilibrium — if either of you were to move in any direction, you’d lose customers. So you just stay where you are, making snide remarks to each other through your serving hatches, forever.
(Until, of course, some enterprising individual notices how far people at the ends of the beach are having to walk for ice cream, the sun beating down on their skin, and decides to set up a new stall… But we’ll get to her shortly.)
The big question
The big question is, why doesn’t this happen in politics?
Why don’t the major parties converge in the middle of the political spectrum? Someone on the far left would still prefer a tiny-bit-leftish candidate over a tiny-bit-rightish one. In terms of the world’s most famous two-party system, why aren’t the Democrats just enough to the left of the Republicans to make themselves the best option for the entire left-leaning population? Why don’t the two parties have DW-NOMINATE scores of ±0.001?
To be clear, this wouldn’t be a good situation — it would represent an enormous failure mode of democracy in which it decayed into something that wasn’t really democracy at all. But, intuitively, it should happen every time.
I have various hypotheses about why it might not, and I think the real answer is some combination of them all. Let’s walk through them.
Hypothesis Αʹ: Third parties
The first obvious explanation is that there are few true two-party systems — and when the largest parties converge on more-or-less indistinguishable, another gap appears in the market.
Just as Nigel once had a monopoly on ice cream, the two of you now form what is, from the point of view of an ice cream consumer, basically a single ice-cream-selling entity. There’s an opening for someone new; enter the Ice Queen.
By positioning a new ice cream stand at the edge of the beach, she is able to sweep up a lot of your customers, threatening to squeeze you out of existence.
Your best defence is to leapfrog Nigel, stealing his side of the beach. Meanwhile, the Ice Queen — despite her noble intentions of faithfully serving the far end of the beach — has an incentive to inch ever closer, stealing some more customers.
This cycle continues for a while, your new rival claiming more and more territory while you and Nigel fight over an ever-decreasing share…
Of course, you could always decide to move over to the other side of the beach; instead of fighting over a tiny sliver with Nigel, why not claim that massive space to the right?
Oh, how the tables have turned!
You can all keep going on like this, if you fancy it. But eventually, I think there’s only one (more-or-less) stable equilibrium here — in which you all take an equal share of the market.
I don’t think this is a proper Nash equilibrium — whoever takes the middle ground has nothing to gain by moving anywhere, though they are vulnerable to encroachments from the flanks — but any changes to this situation are limited to relatively minor tweaks, as any dramatic overhaul will result in the return of chaos.
This is an elegant explanation for political polarisation. It conveniently pushes the parties apart, and also explains why centrist parties tend to see less electoral success, being squeezed on the flanks.
But it has a host of problems:
This theory predicts that third parties should arise on the fringes, in response to major parties that have converged too closely. This doesn’t seem to match reality, in which (major, successful) third parties tend to pop up in the centre in response to high levels of polarisation. (For example, the Liberal Democrats arose from a merger of the centrist-ish Liberal party with the SDP, which split from the Labour party because they felt it had become too left-wing.)
Politics isn’t one-dimensional. In 2D (and higher dimensions, like the very messy real world), the obvious equilibrium has the parties clustered tightly in the centre of the political compass.
What about America? I mean, I know the US has third parties, but they don’t seem to pose any real electoral threat to the real ones, and yet political polarisation is very much on the rise.
Perhaps most fatally, if this is the whole story we’d expect to see changes in the level of polarisation driven exclusively by new parties popping up (or threatening to). It doesn’t explain the ongoing phenomenon.
Let’s change tack.
Hypothesis Βʹ: Activists
We’ve been focusing a lot on voters, but they’re not the people who make the decisions. And we’ve been assuming parties are led by rational beings with a long-term plan that focuses primarily on political success in the most cynical sense.
Enter the activists.
These are the party members, the canvassers, the people who truly believe in the mission — and they are a tiny bunch. The UK’s largest political party by membership, the Labour Party, has about 430,000 members (more than twice as many as the next largest), against a population of 67 million. Normal, boring, moderate, middle-of-the-road people don’t join political parties.
But those who do join get a lot of influence for their money. It’s well known, for example, that the Labour membership liked Jeremy Corbyn (a firmly left-wing figure, even within a left-wing party) more than either the public or the party’s members of parliament (MPs). But the members choose the party leader, and they chose him.3
This is an obvious recipe for polarisation — especially if members choose candidates as well as the leadership. There’s a tenuous balance. Appeal too much to the membership, and the (more moderate) general public won’t vote for you; appeal too much to the public, and the (more extreme) members will have you deselected.
In this model, we should see increasing polarisation as a result of decreasing political party membership and higher levels of intra-party democracy such as leadership elections and closed primaries to select candidates. Some lovely graphs from the Guardian show declining (UK) party memberships across the board up to 2013, as we might expect. But shortly after that, this happened:
That massive jump in Labour Party membership is hundreds of thousands of people joining, mostly to vote for Jeremy Corbyn — a polarising candidate.4
This is political party membership rising right before an increase in polarisation. If the party members and activists are the most politically extreme people around, where did these new, more extreme members come from?
Overall, though, I think the divide between the views of members and the public is a big part of the story. And that’s not a bad thing — remember, this polarising force is what stops democracy from degenerating into an entirely false choice. But activism gone wild is the road to extremism and tribalism, so tread with care.
A functioning party machine will carefully tune the party constitution to allow just the right amount of activist involvement — enough democracy to distinguish themselves from their rivals, but not enough to catapult themselves to the fringes.
Meanwhile, a few other forces are at play too.
Hypothesis Γʹ: Turnout
Imagine yourself on the beach from the start of this post, sunning yourself while shaking your head at the lunacy of putting two identical ice cream stalls right next to each other. The local government wants to hand out a Best Ice Cream Stall award and asks you to vote for your favourite. Would you vote?
People don’t vote if they don’t really have an opinion. Political clustering suppresses turnout — voters have better things to do than vote on whether the basic rate of income tax should be 20% or 20.1%. They need to be excited.
This is why political landslides are won by getting out the vote.
It also explains the Jeremy Corbyn thing — people on the extreme left of the political spectrum weren’t excited enough by the Labour party to bother actually joining the party and paying their dues, until Jeremy Corbyn came along and got them enthused.
But what about Australia? They have compulsory voting, but they’re becoming more polarised anyway. This can’t be the whole story, but I think it’s part of it.
Hypothesis Δʹ: Constituencies
This one is more subtle. None of the countries we’ve talked about have proper proportional representation — votes are aggregated at a constituency level.5 And constituencies aren't all the same.
There are big geographical divides in political views — north vs south, urban vs rural, England vs Scotland. Appealing to voters in rural Yorkshire is an entirely different kettle of fish from inner London.6 “The public” looks different in every constituency. “The centre” is in a different place.
But these constituencies cluster together into parties anyway. Labour tends to win large urban centres, university campuses, and (until recently) northern working class towns, even though these are all very different places. So each winning candidate sits somewhere different on the spectrum, shifting the party’s centroid away from the overall centre of public opinion.
In reality, I think the effect of this is pretty small — not least because very few people actually care who their local candidate is — but I’d be curious about how things look in countries with full proportional representation.
Regardless, the theory is sound enough to be almost obvious (if I do say so myself), so I think the impact is likely non-zero.
Hypothesis Ϛ: Imperfect information
Even more subtlety incoming. Politics is messy. We’ve been treating it as a one-dimensional system, which everyone knows is false.7 It exists in a very high-dimensional space, with a separate (not necessarily orthogonal) axis (beach) for every issue.
And nobody plots themselves nicely on a graph for you. There’s no widely accepted methodology for converting the collection of policies in a party’s manifesto into a nice, precise set of numbers. Even if there were, nobody would know how to interpret it.
In reality, each candidate exists at a single point in this space. But voters see a low-dimensional projection of the space through heavy fog while wearing glasses with the wrong prescription. What they see is a blurry red blob and a blurry blue blob (or maybe one big, extra blurry purple blob).
And unless the blobs are sufficiently well separated, voters don’t know which one is closest to their own views — and they’re unlikely to be particularly certain of those either!
So the natural strategy to distinguish oneself is to move yourself away from your rivals — not too far, just enough to be clearly different. With low information — in times of low trust in the media, say — polarisation should rise to overcome this.
Of course, this whole process is less necessary with obvious, easily comparable questions like most economic issues — no amount of poor information can obscure that one tax rate is higher than another, for example. Where it really kicks in is on the big social questions — the wedge issues. The perfect strategy for low-information times is to stick close to the centre on legible issues like the economy, while driving an enormous wedge between yourself and your opponent on soft, illegible, Culture War issues.
Could anything more perfectly explain the times we live in?
Most of the easy-to-find sources on the Internet focus on US politics — which is understandable, as it’s the largest and most influential English-speaking country and the (de facto) two-party system makes analysis particularly clean — but this is a global (or perhaps just Western) problem.
For example, there were similar shifts in the political and social views of supporters of different British parties between 2015 and 2017 (though obviously the Brexit vote was an aggravating factor here — my inside view is that things have eased slightly as Brexit tensions have cooled, but I don’t think this bucks the overall trend), and I think most people would agree that the professional party politics of recent years has also been particularly polarised. This coincides with the rise of the far right in several European countries.
However, I’ll probably end up referring to American data a lot because there’s a lot of it and it’s one of relatively few countries with just two major parties so it’s conceptually simpler. Sorry about that.
There’s no real reason this has to happen at the beach, but it’s a nice basically one-dimensional scenario, which makes things a bit easier to grasp. Also, it’s been very cold recently, so naturally I have warmer climes on my mind.
Similarly, while nobody could have predicted the scale of the Liz Truss disaster, it was widely expected that she would be one — but she played well to the party membership, so she got to serve (very briefly) as prime minister.
This isn’t necessarily a criticism of Jeremy Corbyn. He was, at the very least, a principled man who seemed to be trying to do the right thing; like most politicians, he got a many things right and many other things wrong. The fact that his stance on most issues was somewhere towards the left of the Overton window doesn’t mean it was the wrong stance, just that it was an extreme one.
Or state, congressional district, or whatever else your country calls it.
Who even cooks fish in the kettle though?
The 2D political compass is a popular improvement, but politics isn’t really 2D either — though the first two principal components seem to explain a fair bit of the variability.