Trying to make some sense of it all,— Gerald Rafferty and Joe Egan (1972)
But I can see that it makes no sense at all,
Is it cool to go to sleep on the floor,
‘Cause I don’t think that I can take anymore
Clowns to the left of me, jokers to the right,
Here I am, stuck in the middle with you
Okay, bear with me on this for a minute. I promise there’s a payoff.
The marriage of the two dominant social sciences of the 20th and 21st centuries – economics and political science – happens in a field of study called social choice theory. Very roughly speaking, social choice theory (and its cousin, public choice theory) believes it is possible to aggregate and compare individual utility functions within a society, apply math to those aggregations and comparisons, and thus generate laws and policies that are “scientifically” grounded to maximize the collective welfare of that society. If you’re familiar with the philosophical bent of the University of Chicago, then you have a good sense of what social choice theory is all about.
The lions of social choice theory are guys (yes, they’re all guys) like Ken Arrow and Ronald Coase and James Buchanan, all of whom won the Nobel Prize because they were economists and were part of the Chicago mafia, and Duncan Black and Bill Riker, neither of whom won the Nobel Prize because they were political scientists and hung out in places like Glasgow and Rochester. There are lots of other famous social choice guys, so don’t @ me if I missed your personal fave.
I knew Duncan Black slightly and Bill Riker a bit more than slightly from graduate school days. They were great teachers and great guys! Bill Riker in particular is something of an academic hero of mine, and I think that his book “The Art of Political Manipulation” is still the best introduction to practical game theory ever written.
I also think that social choice theory is a crock.
But that’s a subject for another day (and if you’ve read more than a paragraph of Epsilon Theory, my views shouldn’t surprise you that much). What’s important for today’s note is to tell you about one of the core pillars of social choice theory: the median voter theorem.
The median voter theorem is an old, intuitive idea, going back at least as far as the mathematician Nicolas de Condorcet in the 1700s. But it was Duncan Black who made the assumptions and the math explicit in a 1948 paper called “On the Rationale of Group Decision-making”, where he “proved” (I know it’s snarky of me, but I put this in quotes because the proof – like all proofs – is a mathematical exercise resting on a ton of assumptions that don’t exist in the real world) that a majority rule voting system will select the outcome most preferred by the median voter.
Who is the median voter? It’s the hypothetical person in the middle of whatever dimension of preferences is being decided on here. In our typical left-right red-blue way of thinking about things, it’s the voter with just as many people to their left as to their right.
Now I have zero interest in going through the assumptions of the median voter theorem and challenging this on the terms set by social choice theorists. Sorry, but I’m still suffering PTSD from five years of this in grad school. I’m just saying two things:
- The median voter theorem is a central pillar of social choice theory and public choice theory, enormously powerful schools of thought that have generated multiple Nobel prizes and sustained literally thousands of academic careers.
- At its core, the median voter theorem can best be understood through the mathematics of distance, where the geometry of issue dimensionality and the voter preferences mapped to those dimensions generates its elegant results.
So here’s the payoff.
The structural analysis of narratives can also best be understood through the mathematics of distance.
There is a median narrative theorem that can serve as a central pillar of a NEW approach to social choice theory, an approach less … pedantic … in its assumptions about human nature and less … naive … in its assumptions about modes of social power.
The median narrative theorem generates powerful predictive hypotheses about elections, hypotheses that predicted Trump’s Republican primary victory in 2016 and – if current data holds – predicts Sanders’ Democratic primary victory in 2020.
Here’s a narrative map of news articles about the US election over the two month period of December 2015 – January 2016. This is a visualization of an enormous linguistic connection matrix created by natural language processing (NLP) technology we license from our friends at Quid. As a visualization, it’s squeezing multiple dimensions into two dimensions, so some multi-dimensional distance information that we capture in the math is lost in the visualization process (yes, all you budding social choice theorists, there’s lots of math in narrative analysis … LOTS of math … but also a different KIND of math), but what’s really great about the median narrative theorem is that the results are so strong (in the political context, at least) that they lend themselves immediately to a clear visual interpretation.
There are 2,215 unique articles captured in this query, each represented by a single tiny dot. The articles are clustered according to the similarity of their language, with articles sharing more of the same words and phrases generating a gravitational “pull” on each other, and articles lacking similarity in words and phrases exerting a gravitational “push”.
Up and down and left and right mean nothing in this map. What is meaningful is node-to-node distance (for cluster construction) and cluster-to-center distance (for map construction).
In a very real sense, the topics that are at the center of the narrative-world map are, in fact, at the center of the real-world attention that people pay to the overall query, in this case the US elections of 2016.
And as you can see, the center of this narrative-world map going into the Iowa caucuses of February 2016 was absolutely, completely, overwhelmingly dominated by messaging around Donald Trump.
And yes, these same narrative dynamics persisted all the way through the general election.
Trump got significantly more coverage than Clinton in major media outlets.
Trump got significantly more positive coverage than Clinton in major media outlets.
Trump suffered from no infectious meme like Clinton suffered from Emails! in major media outlets.
I’m not saying whether all this is good or bad. I’m just saying that it IS. And what it isn’t.
This isn’t a Russia thing.
This isn’t a Facebook thing.
This is a mainstream media thing. A mainstream media thing comprised of people who, for the most part, would rather rip out one of their own fingernails with red-hot pincers than help Trump, but who, driven by the systemic pressures of their business and its utter reliance on Fiat News, did just that. … Continue reading
So where are we today in the Democratic primary? Well, first you need to read this, Rusty’s companion note to mine, or at least the pull-quote I’ve taken from it:
Bernie Sanders is the On-Narrative Candidate
As we have recounted in previous installments, Bernie Sanders has consistently had his story told more clearly in the media than any other candidate. What we mean is that these stories used the most similar and most interconnected language. Yang and Bloomberg, on the other hand, have struggled mightily to produce a clear narrative in the minds of US political media. Other than describing Bloomberg as a billionaire and talking about Yang’s UBI plans, articles about them are all over the map.
We think it is almost self-evident that Sanders is the candidate most attached to the chief ‘meta-narratives’ of the election described above, but this is also quantifiable. The chart below is similar to the topical attention measure described in the first section. In short, this is how much more or less than the average attention articles about each candidate have within the broader universe of election coverage. Think of it as a measure of ‘linguistic correlation’ between the articles written about a candidate and all articles written about the election.
Note that Yang and Bloomberg are missing from this chart. They simply don’t have enough articles for the full period to make them an apples-to-apples comparison; if included, their values would both be well below Warren’s, and would not at all change the observation about Bernie’s unique connection to the framing of election coverage.
What Rusty is describing when he talks about Cohesion and Attention are our contributions to narrative analysis. We think that we’ve figured out how to measure the structural attributes of narrative, in a way that we can track directly to investor/voter/consumer behaviors. Honestly, I think we’ve stumbled onto a pretty fundamental technology for understanding unstructured data in a novel way, as this is a very different approach from how it seems everyone else is thinking about narrative these days, which is sentiment analysis and social media engagement metrics.
Here’s a note we wrote on this research program last March.
Now I want to be really clear about something …
Bernie Sanders is not dominating narrative-world going into the 2020 Iowa caucuses like Donald Trump dominated narrative-world going into the 2016 Iowa caucuses.
The central narrative cluster here is not even exclusively about the Democratic candidates, it’s about the Democratic candidates in relation to Donald Trump!
But within that central narrative cluster, and this is the point of Rusty’s Election Index update, Sanders dominates the other Democratic candidates. Ditto with that big purple cluster just to the left of center (haha, accidental geometry, I’m sure), which IS exclusive to the Dems and is sans Trump.
Point being … a median narrative theorem would have predicted a landslide Trump victory in the 2016 Republican primary campaign going into Iowa in February. This ain’t that for Sanders in the 2020 Democratic primary. But he IS the clear favorite from a narrative perspective to take the nomination.
Sure, there’s many a slip twixt cup and lip.
Again, read Rusty’s note for clear evidence that Bloomberg, HuffPo, the New York Times, and the Washington Post are two months into a no-holds-barred, all-out narrative assault on the Sanders candidacy.
This stuff makes a difference. Sanders is not dominating the other Democratic candidates in narrative-world centrality today as much as he was two months ago.
But for now, the Sanders narrative is the median narrative of the Democratic primary. And until that changes … I think he’s the outcome this majority rule voting system will select.